Can't see many ways to achieve that without either having to split BIOS versions and have user's hopefully flash the right one, or sunsetting support for earlier generation Zen CPUs on those boards for everyone.
The download will require you to key in the Zen 3 serial number or some other way of evidencing that you have the Zen 3 to hand, so you can boot the board back up after you install it - i.e., you'll have to have a Zen 3 CPU ready to pop in to run the BIOS.
It'll forever be a BETA BIOS, and clearly labelled with Zen 3 support. If they're sensible, it will ask the user to confirm they have a Zen 3 CPU ready to install before flashing as well.
There will be some crossover I expect. They'll likely remove Athlon-A and Zen 1000 (non-AF) support, maybe Zen 2000 APU. Depends on how much space they need. So it might still support Zen+, Zen 2 even with the added Zen 3 support.
Keep in mind that it may not be possible to compress large parts of the BIOS images, since the code in them may be needed even before RAM to decompress them into has been initialized.
What pisses me off the most, was I helped build some systems with Zen1, and convinced many buyers to skip the B350 and upgrade to the X370 motherboard instead. My argument was that in the future, their mainstream X370 board was going to help them upgrade from the 1500X/1600/1700 CPUs to the 4600/4700/4900 CPUs. And that was back in late-2016.
So now, essentially those owners have to replace their motherboards, if they want to upgrade their CPUs... which is exactly the opposite of what AMD promised with AM4. I feel sorry that they spent extra money which they didn't want to, based on my recommendation. I've definitely let them down, and probably lost their trust.
AMD promised 3 years, and they got 3 years. They can also get a NEW chip design with higher IPC, and double the cores. I think they can't complain. The issue with B450 support is different, people are buying those now, and MAX versions, specifically with the expectation of future support.
AMD has always provided CPU support based on the socket rather than the chipset. Having built many AMD computers over the last 20+ years, this has always been the case.
If you recall when Zen+ shipped, AMD's official statement was that the first-generation chipsets didn't officially support Zen+, although it was possible for motherboard makers to add support via BIOS updates. This was a huge surprise to me, as I assumed that every AM4 board would support every AM4 chip including Zen 3. I would not have invested so heavily in a brand new Ryzen 1700X build if I knew that not just Zen 2 and 3, but Zen+ support would have been questionable.
I also wasn't anticipating the release of another chipset alongside the release of Zen+. What's even more annoying about this is that X470 offers very little improvement over X370 anyway, and while it officially supports Zen 2, X370 does not. No motherboard maker in their right mind would forego Zen 2 support for higher-end boards that easily support those CPUs.
Now, it seems AMD is officially trying to put an end to X370 CPU support, by excluding Zen 3. It will be interesting to see if boards like the Crosshair VI Hero get unofficial support because I don't see a damn reason why it's not technically possible. If X470 can support it, X370 can, too. I do hope that X370 boards get support somehow, but I'm extremely disappointed that AMD has gone this "unofficial support" route from the moment they started shipping Zen+. Zen freaking plus.
Ryzen 3000 is a very decent upgrade option from Ryzen 1000, but it's not what the majority of people who have used AMD for years were expecting. We were expecting the usual socket-based CPU compatibility, which means Zen 3. AMD users are used to AMD changing the socket when changing CPU support.
I'm getting the distinct impression that some people are mixing up memories of their assumptions when Zen came out with information that wasn't available until later. I definitely don't think it's fair to say that people were "expecting" Zen 3 compatibility - unless I'm seriously misremembering things, I didn't think we knew back then if AM4 would live past Zen 2, let alone that Zen 2 would be such a robust upgrade.
That said, I'm generally in agreement that X470 compatibility with Zen 3 really ought to imply X370 compatibility. There must be a reason why not, and it would definitely be better if they explained that. The Zen+ X370 mess was absurd, too.
Zen, Zen+, Zen 2, and Zen 3 were all shown prior to the launch of Ryzen to be on the same AM4 roadmaps. Zen 4 was always intended to be the clean break.
There is no way to please everyone. X370/B350 owners may feel left out, but its able to get support for 3 generations of CPU, which I don't think is a bad thing. The jump from first gen Ryzen to the 3rd gen is anyway a decent upgrade. Switch to Intel side, and you probably only get 1 generation of CPU support and MAYBE another generation if Intel feels generous enough.
im waiting to build 1700 based system. in 3 years when the 3900x will be available for 100-150 i will upgrade to it after updating the bios on the x370 motherboard. not yet feeling left out. :D
That's actually incorrect. B350 and X370 didn't officially support Zen+ when it launched. AMD said support was possible, but left it up to the motherboard makers to do it. The same goes for Zen 2.
APU support benefits me in no way whatsoever. In the past, AMD gave them their own socket, which made sense. Why would anyone buy an expensive X370 board just to put an APU in it? Never mind the fact that the only APUs available at the launch of AM4 were old Bristol Ridge chips.
My point about Zen+ is that it's not really a generational step, but a warmed-over refresh much like Intel has been doing since Skylake. Going from Zen to Zen+ is a lateral move. Zen 2 is the second evolution of the Zen architecture.
lost their huh-what now? on x370 with ur manufacturer bios supplied you can upgrade zen 1 1700 style processor to zen 2 3700x or even 3900 for 16+ cores if ur bios is upgraded from the manufacturer. this is quite a bit jump over 3-4 years. they shouldn't be let down.
They can still upgrade all the way to a 3rd gen which is still equivalent to Intels current gen.
I mean hey, I’m in club intel and we generally get support for one or two generations of CPU per chipset, not three or four generations.
What’s really annoying about Intel is they KNOW they could support everything from Skylake to Comet Lake, because they are all the same power envelope, manufacturing process node (admittedly somewhat irrelevant) and PCIe/memory bus. But it’s presumable they have the same concern AMD does...how to fit support for that many CPU’s in a BIOS. And Intel makes a LOT more individual models of CPU’s per generation.
"I mean hey, I’m in club intel and we generally get support for one or two generations of CPU per chipset, not three or four generations."
No shit, that's why I typically avoid buying Intel.
Had AMD not set expectations by doing the same thing for 20+ years, I wouldn't have minded so much. Had they given their customers a heads up before launch that, hey, socket support no longer means CPU support, I wouldn't have minded so much. The fact that they just decided to completely reverse course after decades of doing the same thing without bothering to tell anyone is a bit upsetting to anyone that bought in early based on years of expectations.
Really? Because I don't recall Zen 3 even being roadmapped back then. In fact, I just did a quick search and I can't see it showing up in their slides until 2018. If you gave that specific advice that they'd be able to upgrade to Zen 3 then it was bad advice.
Thing is, most (not all, but most) X370 owners can jump up from an original maximum of the 1800X to, say, a 3800X for a ~25% performance boost with the same number of cores - or a 3950X to get twice the number of cores. So advising that they'd get a solid upgrade path was totally reasonable, and remains that way.
Side point - you couldn't get that kind of performance uplift out of a 2-generation Intel upgrade any time between Sandy Bridge and the present, and you had precisely zero chance of doing it on the same board. I'll take "some chance" over "none" any time.
It's really annoying to keep hearing the Intel argument. Yes, we know that Intel doesn't often carry over CPU support, that's one of the reasons why we buy AMD. Lots of people buy AMD because their CPU support was based around the socket, and you didn't have to worry about new CPUs dropping support for different chipsets. The fact that they did this right out of the gate, by releasing X470 only a year later and not officially supporting Zen+ on X370 was infuriating even if motherboard manufacturers did add support. If AMD wants to shift CPU support to the chipset, then fine, but they should have told their customers first. I wouldn't have bought an X370 board at launch if I had known AMD was going to pull this stuff.
There's no technical reason why X370 can't support Zen 4, and a large reason why this problem even exists is because AMD decided to put all of their APUs on the same socket as the CPUs for some reason. Had they just kept them on different sockets, this probably never would have been an issue.
Not to keep this going indefinitely, @rarson, but those slides are from 2 months after Ryzen launched.
The "Intel argument" keeps being made because it's relevant. None of this tech exists in a vacuum. I think they're doing a decent job all things considered. Could the situation be better? Yes. Could it be a lot worse? Oh my, yes.
Pointing to older architectures and the decisions made then as a reason to expect them to stay the same isn't entirely helpful, either - AMD were in a woeful competitive position from AM2 onwards.
My mistake, I was thinking Ryzen launched in May, not March.
That being said I'm fairly sure that I saw that slide before May, and before Ryzen launch. I could be mistaken, but the existence of Zen 3 and its use of AM4 was at least rumored before Zen launched and AMD would have been planning all of this well before they released the slide anyway.
The "Intel argument" is not relevant here. Not for AMD customers. I've already explained ad nauseum why this is the case. After decades of chips being tied to sockets, the least AMD could have done was to let people know they were changing the rules. One of the reasons why I've built so many AMD systems in the past and so few Intel builds is because of the fact that upgrade paths were much clearer.
In this case, pointing to older architectures is the entire point: decades of setting customer expectations led to people assuming nothing had changed.
It not your fault. You couldn't predict how the 4000 series are going to affect the BIOS. I think AMD need to upgrade the BIOS 16mb to maybe 32mb. Since the CPU are becoming more powerful and has the ability to do more, they need to build BIOS adaptability for future upgrade.
While I agree a GUI BIOS is a waste of space (personally I hate them, they are slow to navigate and less intuitive) it will still be a huge challenge to fit microcode for 100 individual CPU’s (taking into account individual designs, stepping, generation, etc) into 16MB. Most microcode files are around 200kb give or take. You still need UEFI drivers, USB legacy initialization for HID devices, a network stack for live BIOS updating, and so on.
I think AMD is in a tough spot here, and the consensus is they should just drop 1000 series support since they have the most errata (and hence the largest microcode updates, though they don’t “have” to use the latest microcode I guess...) and I can’t imagine people running first gen parts anymore on brand new boards unless their old board failed.
It’s really interesting to see AMD commit to supporting a socket interface this long.
AMD can't eliminate support for the older CPUs as these are popular options in foreign markets.
I'm all for getting rid of the stupid graphics in BIOS. I've always hated fancy interfaces that make it more difficult to find and change the settings that you need, especially since once I get the BIOS set up as I want, I don't intend to continue going back into it anyway. I prefer the old school low-res, text-based, keyboard only BIOSes. I don't need a freaking "EZ mode" or any of that nonsense, I just need a clean, simple menu that I can easily navigate through.
I think it was GN who said that if they went back to old school tabulated interfaces in the BIOS then reviewers would complain and make them look bad. I say fine. It lets me know who to take seriously. The A E S T H E T I C S of the BIOS shouldn't matter. Its there for setting system configuration and maybe overclocking. Not to look cool. Most people will only ever be there once or twice.
Any reviewer who values how a flashy interface over practical support and features isn't credible in the enthusiast space. They can go review Apple products for all I care.
Yeah, I generally agree with Steve and I understand his sentiment, but I don't think the issue would be as big a deal as he thinks if it means Zen 3 support is added to existing products. Surely the people who are buying and using them wouldn't mind the trade-off.
I WANT a BIOS that looks old, because those old BIOSes were stupidly simple to navigate and change settings. I don't want to have to use a mouse to change settings. These stupid graphical UEFI environments often leave me wondering where exactly I need to click on something or what is a dropdown/toggle, etc. A simple text interface is infinitely cleaner and easier to use. A graphics-less BIOS is a feature that I actually want, not a detriment.
For clarification, Steve agrees that the flashy interfaces are dumb and would be fine with simple ones. He just argues that there are many in the industry who would complain and give AMD and motherboard makers bad press if they did that.
The difference is AMD promised to support AM4 for 4 years, which implies they will support every COU released on every chipset for that socket.
This still isn't what they promised.
They made promises they didn't realize they couldn't keep. Probably because they didn't think about it, as is obvious by the board partners only using 16 MB bioses and the CPUs only being able to address 16MB.
They NEVER promised they would support every single CPU released for the socket on the same BIOS version... You're pulling crap out of your butt. Literally all that was said was that they'd be supporting AM4 through 2020. Anything else is simply you reading way too damn far into things.
Some of you guys are missing the point: AMD has been doing this for decades. People assumed that AM4 support means that all AM4 CPUs would be supported because that's the expectations AMD has set for their customers, FOR DECADES.
Every single previous AMD socket has supported every single AMD CPU for that socket, and then some (the plus sockets even supported newer CPUs in older boards). Every single one.
I built my first AMD system in 1997 and since then, have rarely used Intel not only for pricing reasons, but because of chipset compatibility. That is to say, with Intel it's always been a crapshoot as to what chipsets will support what processors, but with AMD it has always been determined by the socket. When AMD changes CPU support, they've always changed the socket. Without fail. Until now.
You guys who are thinking this is some kind of bizarre assumption clearly haven't built many AMD systems over the last 20 years.
I was gonna post the same but refreshed before I started typing. :)
I'm on AMD since Super Socket 7 (good old K6-2 300MHz). And it has always been the case that if the CPU physically fits the socket the chipset would run it. At least I don't remember a CPU where that wasn't the case. If you know of one, let me know.
AM2+/AM3 CPUs are a special case. A Phenom X4 9950 won't work in an AM3 board because it doesn't have a DDR3 IMC, whereas a Phenom II X4 940 will work in both sockets thanks to having support for both DDR2 and DDR3.
This is true due to the memory controllers, but all of the "plus" chips were still backwards-compatible with the older sockets. I mean, it's kind of a special case if you need to put an older chip into a newer board, not many people would attempt that.
this was never true though, even back in the socket A days, the mixes of different FSBs and chipset support meant you couldn't expect true forwards compatibility. Try putting a 2002 XP 2600+ (333) (which came in both 333 AND 266 FSB versions for extra confusion!) in that 2001 KT133A board you bought for your XP 1800+, it wouldn't work, and they came out 10 months apart, not 3.5 years apart.
This pretty much summarises why I feel the way I do about this. The situation is still way better than it was at many points in the past, so while I get some level of disappointment and am pleased by AMD's change of heart, I also don't understand why this appears to have stirred up such opprobrium.
"I also don't understand why this appears to have stirred up such opprobrium."
Then you're simply not reading what has been written. This was one example in over 2 decades of chips.
There are always edge cases of certain boards not supporting certain CPUs. Usually this comes down to BIOS updates. My IP35 doesn't support all Core 2 chips because Abit went out of business before finishing the BIOS updates. There are even certain CPU instructions that aren't supported with older BIOS revisions (I ran into this problem when trying to install Windows 8 64-bit on that board and had to download an unreleased beta that someone had modified to make it work). That's totally different than the CPU manufacturer just outright deciding not to support chips that technically would work fine with a given chipset (like Zen 3 on X370). Yes, I understand the reasons why AMD would make such a decision, but to go back and offer support for X470, which is practically the same damn thing, is a bit absurd. Nobody that bought X370 expected AMD to support ONLY Zen 1 (not even Zen+), which is what they did right from the start.
Yet we should feel "lucky" that we got unofficial support for Zen+ and Zen 2. This is not how this normally works with AMD platforms.
That's kind of a unique example though. Even in the Pentium days boards had configurable FSBs, and it was just a matter of setting jumpers (or if you had a jumper-less board like I did, changing settings in the BIOS). And even if your board didn't support a higher FSB, you could still run faster chips using different multiplier settings.
Regardless, once AMD moved the memory controller to the CPU, extended support became the norm, especially with stuff like the "plus" sockets.
This is not quite true. I think Ryzen 3000 / X570 socket could have been AM4+ to resemble what happened in AM3/AM3+ (granted, they were different architectures). But if you take a look to Socket A (462)...
It is not FSB frecuency that defines socket A compatibility, but most likely it is the package. Most KT133A boards will run ceramic Palominos just fine, but will refuse to boot any T-bred. Not all nForce2 boards will run Thunderbirds (I think this is based on multiplier rather than package).
Same goes for AM3 and AM3+: I have a M4A79T wich will not take a Bulldozer, even if it's power delivery and RAM cupport should allow it.
I think we can just conclude that naming has been pretty much FUBAR for Ryzen (older gen APUs with current gen names, B550 release date closer to 4000 than 3000 series processors) and call it a day.
Technically if AMD only allows x670 boards to work with the new Zen3 chips they still have met the standard of using AM4. They never said your x370 board would work with CPU's 3 years later.
Yes, they never explicitly stated that X370 would support Zen 3, however they've spent decades setting customer expectations by ensuring CPU support based on socket compatibility.
True socket support in the past meant CPU support but I simply don't see any reason to get outraged. It’s not like they are arbitrarily limiting support or Like Intel creating a new socket for no technical reason. They have real reasons of fitting all that AGESA code into the BIOS.
I agree with you completely on the outrage part. People are reacting a bit too extremely about this. But to me, this whole situation is a little disappointing for a few reasons. Primarily, I'm disappointed that AMD was caught so off guard by this problem, as BIOS limitations should have been pretty obvious before they started shoving everything onto AM4. Previous generation APUs had their own socket and I'm not sure why AMD felt the need to consolidate everything to AM4, especially the pre-Zen stuff like Bristol Ridge. They already had Zen, Zen+, Zen 2, and Zen 3 on the public roadmaps LONG before Ryzen was launched and should have known well before this that the CPU support list was going to become massive with all these different chips to support. Basically, it's a little embarrassing for a company who has manufactured CPUs for this long to run into this problem, let alone be blind-sided by it. I'll chalk it up to ambition, but they should have seen this coming.
Also, CPU support based on the socket always made planning builds with future upgrades in mind a lot easier with AMD. Given that they have usually been the budget option, this is plus for their target audience, it makes going with AMD an even easier decision. If AMD wants to shift CPU support to the chipset, that's fine. I may not like it, but if that's what they determine they should do, then I accept that. However, it's not right to suddenly introduce this change when you're 75% of the way through your socket road map and many of your loyal customers bought in at the beginning under the assumption that the company would continue maintaining support the same way they had for decades.
And honestly if they do shift to a chipset-centric support model (which, let's face it, they have because X370 support is done), that will make me think twice the next time I want to build an AMD computer. I'm not saying that I'm boycotting them or anything, just that I'll be a little more weary about CPU support.
Lolwut? Why'd you wait 4 years to complain about this? Couldn't you have made similar complaints right at the start claiming that amd failed to keep their promises because cpus like 1800x and cheap a320 mobos existed at the same time?
Part of the problem is that in the past, AMD has typically changed the socket when changing CPU support, and even then, as was the case with the "plus" sockets, still retained some compatibility between new and old boards and CPUs. Usually, AMD would announce the socket and state how long they intend to support the socket, and that was it. Intel on the other hand was always tied to specific chipsets, and the socket had no real bearing on what processors would be supported by a given board. This has always somewhat frustrated me about Intel, because you can never assume that a given chipset will support future processors. Case in point, I feel like anyone that invested in Z270 got really ripped off. That aside, I will give Intel a small bit of credit for avoiding this kind of mess completely by setting customer expectations.
I built my first PC in 1997, and have always switched between AMD and Intel, based on my budget, the parts currently available, and the best bang for the buck with upgrade path in mind. Granted, parts are a lot cheaper than they used to be (graphics cards aside), but generally I start with a good motherboard and a cheap CPU so that I can upgrade down the road. It's ALWAYS been easier to plan very far ahead with AMD, because you knew that when they said they would support a socket for a number of years, the newer CPUs would work with all boards that used the same socket. In fact, when they first announced AM4, I assumed that meant that every AM4 board would support all CPUs up to and including Zen 3 (based on the timelines it seemed pretty clear that the third Zen iteration would arrive within that support window). It's a little frustrating to have AMD suddenly change CPU support without changing the socket after decades of doing the opposite. I would have rather seen them change the socket for Zen 3 than continue with the same socket while breaking support. But it's pretty clear they realized far too late that this BIOS thing would be a problem.
I don't necessarily fault them for being overly ambitious, however it kind of makes you wonder why they bothered with stuff like Bristol Ridge on AM4 or why a company that has been making CPUs and chipsets for decades wasn't more aware of the BIOS problems arising from supporting too many CPUs on a single socket. If it comes down to removing BIOS graphics altogether and going back to a normal, plain menu interface then I'm all for it, as I really loathe useless BIOS graphics.
Actually, and I can't believe I have to keep pointing this out, it's closer to Intel: "We tried to make multiple generations work on the same socket before and we got absolutely *slated* by the enthusiasts and press when changing power budgets meant not every chip worked - so now we don't make promises about how long sockets will last"
Intel has never made promises about how long sockets will last, CPU support has always been tied to the chipset and not the socket.
What Intel got blasted for is creating warmed over refreshes (Kaby Lake) and useless chipsets (Z270) fully aware of the impending release of higher core chips with Coffee Lake.
I recently upgraded from a 1700 + X470 to a 3700X + X570. But I almost decided to just put the 3700X in the X470 board. I don't currently have anything that takes advantage of PCIe 4 so there is really no performance advantage to be had from the motherboard upgrade. I mainly did it for future upgrades and because I could use the old motherboard and CPU in another system.
The random consumer likely doesn't even know how to upgrade their CPU, and I'd wager a lot of people run their computers to the ground. Upgrading is a moot point by then.
Most people don't upgrade every year so I agree. However those with a Ryzen 1000 could upgrade to 4000 if their motherboard supports it. Always good to have the option.
Yup, that‘s it. The good thing is that older gen Ryzen get very cheap because they are competing with the new models for the same customers.
Also, at least for me installing a motherboard is the part about building you own that I like the least. So not having to do that while being able to get a noticeably better CPU is a great plus for me, even if it means losing out on some of the new features like PCIe 4.
Got my Ryzen 2700x for €150 new on special offer a few months ago and being able to upgrade to either a cheap Ryzen 2 or 3 down the road is a nice outlook. Heck, with a 12 core Ryzen 4000 I may even be able to get much better performance in the same or lower power budget.
At this point in time, given that every X570 board has a fan on its chipset to cool the thing, I'd consider the lack of PCIe 4.0 to be a positive, not a negative. Those tiny little fans clog with dust and seize up so easily, I don't want any part of my PC relying on one of those things for cooling.
There are no PCIe 4.0 graphics cards yet, and even if there were, I doubt running them at 3.0 speeds would be any kind of bottleneck. Similarly, for NVMe drives, while the speed difference likely exists, I probably would never know the difference comparing the two side-by-side for my normal computer usage.
That little fan you speak of I've never heard once in the 6 months I've been on a X570 board. I would also never let the inside of my PC get to the point where dust will clog a fan. This isn't the 90s anymore the fans are of decent quality and quiet. Anyone using an Air Cooler will hear more noise from that than the chipset fan.
No, I work in PC repair, where I see these tiny fans die all the time. It's cute that you think your 6 month-old X570 board is a valid example case, but having seen years of motherboards and graphics cards with these fans die from the slightest amount of dust, I'm just going to go with my experience over yours.
If you've ever worked with a large amount of these things, you'll know just how little the amount of dust is required to seize these things up.
Not all fan designs are the same. I too have been burned by years of dodgy fans, but there is a greater prevalence of FDB fans over the older sleeve bearings that wore out so easily with heat and dust.
I don't know for sure whether most X570 fans are FDB or not, but based on people's comments on the noise they make and the cost of the boards involved, it seems quite likely many are.
The 6 month comment on X570 is funny because my X79 board is 8 or more years old and still works great. Yes it's another point of failure, but it's far from a guaranteed point of failure.
I think nVidia‘s and AMD‘s next gen GPU may change that. Having PCIe 4 for the top of the line models will probably be a plus. Same for nVme storage once the newer controllers come out.
But yes, right now it‘s not a problem and if you are not going for the high end GPU wise it should also be OK.
Regarding fans - I have no big issue with them but not needing to have one on the mainboard is a plus, so B550 would be the preferable option for me.
With this regard I am kind curious about the upcoming Z490 boards as they seem to have several smaller fans to cool the VRM.
Presumably the same reason someone would "upgrade" from a 3000 series to a 4000 series. My guess is they have money, too much of it, they hate money. So they spend hundreds of dollars on a part that isn't anywhere near being a bottleneck, literally every year.
My grandpa ran A dealership and he has customers who would buy a new car, drive it to the first oil change, or 3 months, or 6 months, or one year, then buy a whole new car. Often the exact same thing. They simply had so much money this was more convenient to them than dealing with any maintenance at all. It also put some very clean used cars into the used car market and my grandpa got some very nice margins on both ends of those deals.
Cause yeah, motherboard/cpu/ram should last you 10 years easy. 5 minimum. Upgrading just the CPU hasn't made any sense at all since LGA 775.
Upgrading just the CPU makes perfect sense if you buy a bang-for-the-buck CPU at the start of the build, and then upgrade to the best CPU you can a few years later after prices have dropped significantly. That's what I've always done, and even after buying 2 CPUs I've still paid significantly less than the initial cost of the high-end CPU.
This. I have a Gigabyte GA-AB350-Gaming 3 which was originally paired with a Ryzen 5 1600 but now sports a Ryzen 5 3600. The cost of both CPUs is a little higher than the launch price of the Ryzen 7 1700X, and the 3600 can outpace it in most areas despite a two core deficit. The motherboard is still fit for purpose, even if the LED strip on the left side of the board has turned a shade of green.
Lol this is the dumbest ever comment I have read in the tech space. Upgrading the cpu doesn't make sense? I went from a ryzen 5 1600 at launch for 225 to a 1700x for $215 (sold the 1600 for $155) a half a year later and used that until the 3900x came out. I started with a b350 tomahawk arctic for 85$ on sale. Sold that board for $105 7mo later when it was discontinued. got an x370 carbon (for $77 on sale) and have been using that since then. upgrading the motherboard in a windows machine requires a new license or a complicated activation process (which isn't always successful as it depends on your windows key type). people who upgrade motherboards and spend their time for no benefit are the ones who love to waste money. If I sold my 3900x now (which is a great bin btw) I could get 380$ for it. Considering I paid 450 last August with my best buy birthday discount... That would be $70 for almost a year of use; and I could go to a faster 4000 chip again. this is by far the most economical way to upgrade and always stay on the newest platforms. being forced to purchase all new hardware to upgrade the cpu just basically means that you will lose alot of money to upgrade and are likely to keep your pc until its worthless and then spend far more doing upgrades.
AMD never lied. You're a misremembering salty fanboy. (And lol we all know that "people you personally know" is you). You got 3 full generations of CPU's on the same board, including the MASSIVE leap that was Zen 2. AMD continued supporting your board through 2020 as they promised (and even into it, with the 3300X/3100). I simply cannot fathom how a 300 series owner can be that mad at having a board that supports nearly 100 CPU's, and thinking they got shafted on compatibility.
As an X370 owner Im not mad. AMD never promised I could use my x370 board for 3 years of CPU generations. If Gigabyte somehow ports support over for the 4000 series, even in a beta bios, I would be thrilled but I am not counting on it.
Zen+ is not a new generation of CPU, it's a re-spin. It's what Zen would have been had AMD not had to rush it to market. And while Zen 2 is nice and all that, Zen 3 being supported on AM4 was the expectation.
Why should anyone care about lots of CPUs being supported when a good chunk of them are completely irrelevant? Is anyone still using Bristol Ridge on AM4? Why did AMD even bother putting Bristol Ridge on AM4? Does anyone who is building a high-end PC with discrete graphics give two shits about APU support? Why didn't AMD just give the APUs their own socket like they did with the FM sockets?
AMD should have known and planned better for this. As it is, it's pretty clear they didn't realize until way too late that this was a problem. There's no excuse for a company who has made CPUs for as long as they have to run into this problem.
You're missing the point. AMD's CPU and APU sockets were always different in the past. They should have just left Bristol Ridge on FM2+ or created a new socket for it.
But that was a PITA for everyone (buyers, manufacturers, etc.) - I can totally forgive them for wanting to resolve that, despite it having created a new problem in the process.
APU and CPU sockets used to be different because the used different interconnect. AM2/3 used Hyper Transport FM1/2 used Unified Media Interface. AMD CPU-s are SoC-s, so there is no need for dedicated interconnect to the PCH.
*AMD CPU-s are SoC-s, so there is no need for dedicated interconnect to the PCH. AM4 CPU-s are SoC-s, so there is no need for dedicated interconnect to the PCH.
AMD did not lie about the socket support, they are still technically supporting the socket. However, the problem is that for the last 2+ decades "socket support" meant that CPU support was tied to the socket, not the chipset (unlike Intel). For them to suddenly change this is a real problem for customers whose expectations AMD has set for decades. I, too, assumed that AM4 support meant that Zen 3 would be supported on all AM4 boards, because this is how AMD has done it for the 2 decades that I've been building my own computers.
While I dislike the idea of AMD now determining CPU support based on chipsets rather than sockets, like Intel does, I wouldn't have a major problem with them doing so had they made it clear from the outset that "socket support" no longer guarantees CPU support. You can't just throw out a huge change like this years after selling people the product on the assumption that CPU support would continue.
I went from a 2700X to a 3900X on my X470. Nice jump in video rendering performance and increased frame rates in older games that are lightly threaded. Sold the 2700X to a friend, so it didn't cost a lot to do. Flash BIOS, remove heatsink, clean things up, install new CPU and you're good to go. No real issues or hassles. If the next gen has a sizable increase in the things I use my PC for, I will do it again. I already know where the 3900X will go. Then probably hold onto that for a few years until DDR5 and CPU advancements entice me to upgrade.
Chopping off a head is a terrible analogy. An old head is useless to you (it would be a downgrade) and useless to almost everyone else.
A better analogy would be: do you replace your clothes dryer when buying a new high-end clothes washer? You could upgrade your dryer at the same time, or you could use your old dryer. You could get a far better dryer while you are at it and sell off your old dryer. Or, you can keep chugging along with your older, possibly slower, used dryer with fewer features.
Upgrading a CPU while keeping an old motherboard is a small subset of the already small enthusiast subset. There just really isn't that many people that do it. If you are extremely budget constrained and short-term budget focused, it might be your only option. But, in the end it is a better long-term financial deal to wait until you can afford replace both.
I personally just can't get over gimping a good new CPU with an old motherboard. It just feels wrong. In this case there isn't too much different from B450 and B550. But, if you need the differences such as PCI 4, the B550 is well worth the upgrade. Sell off your old motherboard with your old CPU and the motherboard upgrade to you costs next to nothing.
"Upgrading a CPU while keeping an old motherboard is a small subset of the already small enthusiast subset."
Sums up everything nicely. While we readers of Anandtech generally have the technical background, knowledge, and tools to perform a CPU upgrade with no difficulty or drama, we represent a VERY small percentage of the general population.
Maybe so but but keeping this small part of the community happy and supported it only brings greater goodwill and standing to the AMD brand and will encourage others to stay with AMD or swap from Intel. Good marketing and PR goes a long way and AMD still needs to keep hungry for customers and revenue. It's also karma for screwing over the Threadripper 2000 customers...
Why do you care about AMD or Intel? I just want competition. I purchase whatever product gives me the best feature set for the best price. You act like you're going to divorce AMD over this.
I don't think the enthusiasts are such a small market for AMD. For Intel that market may be small change, and that's why they do not care much about compatibility, but AMD is not that strong in the OEM market, but very strong in the DIY market. And those who DIY also want to upgrade, but often avoid the cost and hassle of changing the board. And no, for most users an old AM4 board does not gimp a new AM4 CPU; but if it does for you, keeping compatibility does not prevent you from changing the board along with the CPU.
AntonErtl is correct, and to add to it, AMD is often the choice for budget-oriented enthusiasts, who may opt for a cheaper CPU at the outset of a build so that they can replace it later on with something better. I've always based my builds this way, buy a good board and a great bang-for-the-buck CPU and upgrade it a few years later. Additionally, anyone that has kept any kind of finger on the pulse of the used market knows that Intel chips are obscenely overpriced while AMD's prices tend to drop as you would expect them to. So buying used is a huge benefit for AMD customers who want to upgrade later on.
Enthusiasts are exactly the people who upgrade CPUs. Non-enthusiasts don't generally do that. Not everyone can afford spending thousands of dollars every time a new generation of CPU comes out.
The appliance analogy doesn't even come close. Most people upgrade appliances when their old one dies, and the washer and dryer operate completely independently from each other. Appliances are also major purchases that require delivery and installation. I have absolutely no idea how you think appliances are comparable to CPUs and motherboard.
"I personally just can't get over gimping a good new CPU with an old motherboard."
But you're not. The chipsets have nothing to do with the performance of the CPU. The only thing gained with X570 of any significance is PCIe 4.0, which isn't even supported by graphics cards yet and is inconsequential to the average user from an NVMe performance perspective. If the CPU had a DDR5 controller in it, then sure, you'd probably be somewhat gimping the CPU, but the rest is basically irrelevant. I don't spend $200+ on a motherboard just to replace it in a couple years, that's just dumb.
"I have absolutely no idea how you think appliances are comparable to CPUs and motherboard." Because you use them in pairs, and you can upgrade one without upgrading the other. Same with CPUs and motherboards: you use them in pairs and can upgrade one without upgrading the other. That is a decent analogy. Comparing upgrading a motherboard to cutting off your head is not a good analogy.
"The chipsets have nothing to do with the performance of the CPU." They do if you need a new feature of the new CPU, if you need faster memory to gain a lot of the benefits of the new CPU, if you need a faster interconnect, if you need more ports with the new CPU, etc. Even if you didn't need any of those a slight chipset upgrade can also come with a speed boost. Up to 5% speed difference here: https://www.techspot.com/review/1872-ryzen-9-on-ol...
"Because you use them in pairs, and you can upgrade one without upgrading the other."
I can also use one without using the other, and I don't ever buy "upgraded" appliances just to wash or dry clothes faster.
"They do if you need a new feature of the new CPU, if you need faster memory to gain a lot of the benefits of the new CPU, if you need a faster interconnect, if you need more ports with the new CPU, etc."
All of those things preclude using a newer chip in an older motherboard anyway, with the exception of a feature like PCIe 4.0. That is, if a new CPU has a DDR5 controller on it, physical DDR4 slots obviously aren't going to work. So you're not "gimping" the CPU because the CPU doesn't work at all.
In this case, the ONLY "gimping" you're getting is PCIe 4 which is essentially useless at this point, especially since PCIe 3 isn't any sort of bottleneck, especially for the CPU. Regarding the article you linked to, did you even read the title, or any of the rest of it for that matter? The differences are clearly chalked up to VRMs, power consumption, and firmware. Or did you not realize that a CPU can be slightly faster when it is allowed to consume more power?
My main problem with X570 is the heat of the chipset. The damn thing runs so hot that it needs a fan on it. I run far, far away from those tiny little fans because they suck in dust and seize up with the slightest amount of build-up. I don't want my motherboard dying because the tiny-ass chipset fan stopped working. For that reason alone, I'm more than happy to forego PCIe 4 for a board that doesn't require a chipset fan.
Changing board requires a Windows reactivation or reinstallation, while changing CPU doesn't. It's also a lot less work to remove a CPU cooler and swap a chip than it is to remove nearly everything from your case to get the motherboard out, and you stand a lot less chance of accidentally killing a major component with ESD or an errant screwdriver.
This is before you get anywhere near the e-waste implications.
On a desktop (or even a laptop - especially a laptop), I would imagine plenty... especially with people who are looking to save on upgrade costs and minimize the upgrade to just 1 or two components.
Some might not even be aware of the possibility they could save a lot of money on just updating the motherboard BIOS which would give them the option to upgrade to a far better CPU which would improve system performance.
Swapping out the motherboard on a desktop adds extra time and cost for the users. In a laptop, its impossible to swap out the motherboard because laptops use custom-made mobo's for specific chassis (but then again, most OEM don't even update mobo BIOS-es on laptops which makes matters worse... although in the past they used to - which made it possible for a person to swap out a Core2Duo for say Core2Quad - if the cooling had the capacity to handle the newer CPU).
"Most laptop CPUs are BGA so you cannot change the CPU anyways."
You can, but it's a LOT more difficult.
I don't think I've seen a socketed laptop motherboard since Sandy Bridge. I've looked a lot, but that's about when most manufacturers started transitioning all of their laptops to thinner designs and BGA chips.
"most OEM don't even update mobo BIOS-es on laptops which makes matters worse"
This isn't true, as HP laptops will often update the BIOS on their own, without warning. Windows can also update the UEFI through Windows Updates. But they don't do BIOS updates for support, they do them for bug fixes, as the support is already "baked in" by the hardware choices of the OEMs and doesn't change.
I personally went from an R7 1700 to an R7 3700X on my MSI B350 Tomahawk board. This means that I had to use one of these "permanently beta" bios versions that drops older CPU support (I think A-series chips?) and it removed some of the fancy GUI bios config features like graphical fan curves. Worth it to me personally, since the perf jump from zen 1 to 2 was decent, along with the better memory support. resulting system is a bit more stable.
Quite a few do. There are people running Ryzen 3000 CPUs on X370, for example. The upgrade path from Ryzen 1000 or 2000 to 3000 is pretty compelling - much better IPC and more cores without having to buy a new MB, that's quite convenient.
(If Intel had allowed newer CPUs to be installed on the B250 -- which was technically possible -- I'd have made such an upgrade.)
I have in the past upgraded from K6-2 300 to K6-2 500 and from Athlon 800 to Athlon 1200; someone interested in more cores may easily want to upgrade from a 2700X to a 4950X or so.
Another use case is replacement after failure: when my LGA 775 board died, I needed another LGA 775 board (and despite Intel, one could still get such things). When my Core i7-6700K (tray) died, I bought a Core i5-6600K to fit in the board; if that happened today, I would curse Intel for not supporting the 9600K on the same board.
Socket A lived a long time, but certainly evolved - in three years, users went from AGP 2x to 4x to 8x, FSB went from 100MHz up to 200MHz in 33MHz steps, DDR replaced SDR, dual channel became a thing (that wasn't worth bothering with at that time on AMD), and there were so many chipsets, many of which were problematic (here's looking at you, VIA). I went through four in that time - AMD 750, SiS735 (on the infamous ECS K7S5A which slowly eroded its ability to run at 133MHz FSB, until one day it wouldn't even run at 100MHz), KT266A (ridiculous how much extra performance it offered over even the 735) and nForce2 400. All this time, it was still one generation of socket - even if people did need to buy new boards for Zen 3, this is still an order of magnitude less hassle than the upgrade path not even 20 years ago, though to be fair, CPU speeds more than doubled in a three year period.
It seems like every socket I've ever used that VIA made chipsets for had at least one very problematic VIA variant and one really good one. I kind of miss those days because there was more variety in features and performance. You really had to do your research before buying, though.
Exactly, VIA would release an A variant to fix the issues (or add performance) in some cases. The KT400 didn't even officially support a 400MHz FSB (the clue certainly wasn't in the name) until the A variant, and the KT266A was a huge jump over the KT266. Luckily for VIA, pin compatibility meant it was easier for motherboard manufacturers. I didn't mind the research when you had sites like this providing you with the ins and outs, certainly helped back then where it seems there were double the number of manufacturers (as well as numerous chipsets).
I certainly would if I could and it makes sense on a $ / perf basis. I have a PC now with a i5-6600K that could be upgraded to a i7-7700K, but spending <=$379 on the CPU alone doesn't make a bit of sense versus spending the same or even less money to buy a NIB Ryzen CPU and mobo. Add in the option to upgrade the CPU on that board in a year or two and it's all gravy.
This is exactly why I tend to go with AMD. I pieced together an HTPC using parts that were laying around the shop. It's got a Z270 m-ITX board with a 1080 that is significantly limited by an i5-6400T. It was the best CPU that I had laying around, and not only can I not overclock it but it's so SLOW with a max of 2.8 GHz. The 7700K is the best I can put into this thing, but even used prices for those are absolutely ridiculous. I have a motherboard that is completely useless because there are zero upgrade options that make any sense.
I'm probably just going to replace it with a B550 board and 3300X. That'll give me about the same performance in games as a 7700K for less money, while allowing me to upgrade to Zen 3 in the future.
That's literally how I determine my builds. I buy a good motherboard, one that has all of the features I want and can support plenty of upgrade options, then I buy the best bang-for-the-buck CPU for it, with the intentions of upgrading to the best processor a couple years later when their prices have plummeted.
I stuck with Core 2 for 10 years. My initial CPU purchase was the $65 Pentium Dual-Core which overclocked from 1.8 to 3.0 GHz stupid easy on the stock cooler. Then a few years later, upgraded to a Q9550 for less than half its original MSRP. Kinda the same story with the graphics card. There was no need for me to replace the board because it was a great board. With a SATA SSD, an overclocked C2Q, and a good graphics card, it ran better than a lot of newer machines. It even ran Doom 2016 pretty well.
If my x370 board gets Ryzen 4000 support I would upgrade. I have a higher end board with 32GB of Samsung B.die so Im all set. Would love to get a 30%+ performance boost over my 1800x.
It used to be quite commonplace. Hell, they used to make special versions of new processors to keep old boards running(486-class parts that fit 386 sockets, Pentium-class parts that fit 486 sockets, an absolute metric buttload of P2-class parts that fit socket 7) Intel worked hard to stop it from being an option.
That's my understanding too. Maybe chips that can address 32MB ROMs can use the "back" half? It's conceivable that MSI's MAX series boards could support every single AM4 CPU, but I doubt it is worth their time and effort
It's like an entirely different company, aside from the continued marketing blunders. I think the plan was something like "fix the CPUs, fix the GPUs, then fix the marketing team." Hopefully they'll eventually get there.
Question: What is limited/left on the table by having this one socket (A) from Excavator to Ryzen 4000 and now these BIOS limitations? To be clear, I support and commend AMD's decision to not pull an Intel move here (new chips, new sockets, sucks to be you, go pay for new MBs), but doesn't sticking with an old socket and limited BIOS also come with significant downsides?
Yes. Go watch GamersNexus' most recent video on this whole situation. AMD was screwed & gonna piss somebody off either way they went. This was a "lose-lose" situation all around.
The 16MB BIOS ROM size limitation to me hearkens to what I feel is something that's holding modern x86 processors back: it has to start up in Real Mode, for no reason other than to maintain backwards compatibility.
We can ridicule Intel all we want for its lack of delivering something other than 14nm, but can we focus on getting rid of the baggage that is BIOS and Real Mode? Or at the very least, if you really want it, you can have it but don't make it the default anymore.
Maybe I am misreading the chart, but it appears to show that Ryzen 3000 CPUs are not supported on A320. I’ve got a secondary machine at home running a 3700X on an ASRock A320M-HDV R4.0. Works perfectly, no issues whatsoever. Very happy with AMD.
It isn't officially supported lol. OEMs just said "we don't care what you say on this one" and updated most A320 boards anyways as they already had the necessary AGESA code from the other 300 series boards.
That was the whole point of UEFI, to get around the limitations of BIOS. They're technically two different things, most people generally still refer to both as BIOS.
I don't think this has anything to do with Real Mode since Real Mode only has 1MB of addressable space. This is more to do with fancy graphics, large CPU support lists, and early Zen processor limitations.
To be fair I celebrate that AMD keeps it promise, but as a consent consumer I can no be buying a pc every other year. I try to make my PC and laptop last as log as they can (10+ years), so a strong processors and a strong board is critical for me (with slots and technology for today and tomorrow devices). This year with all new technology in the pass years justify the upgrade for my PC (USB 3.2, DDR4, PCIx 4.0, M.2 drives) and went from a Core i5 750 and some mombo I dont recall to a Ryzen 9 3900x and a x570 mombo. The next upgrade will be GPU a couple years on when the current GTX1060 begins to really lack (in one or two year).
CPU upgrade path is nice to have, but there is no way to support a platform for a long time, i'll consider to upgrade my CPU if eventually AMD came with a CPU using 32 cores over the same AM4 x570 board (which is unlikely in the short term).
AMD would have to massively shrink their CPU dies in order to fix additional or higher core count chiplets on the existing AM4 package. I doubt even 5nm would be enough to accomplish such a shrink.
I start with a good board and buy the best bang-for-the-buck with the intent of upgrading to the best CPU possible in a couple years from the used market. I can invest in a platform without the huge upfront cost of a high-dollar CPU, while still reaping the benefits of maximizing the hardware platform later on. For a minor upgrade cost, I get a nice bump in CPU performance that, along with storage and GPU upgrades, keeps the computer useful for several more years. I'm basically getting about 10 years out of each computer at this point.
If the BIOS was only 1MB then there would be a good reason to limit support - with 16MB there is plenty of room. It might be necessary to make the BIOS a non-graphical text mode one (like most 486 motherboards) but there would be plenty of room for the essentials. (A mouse driven graphical interface and fancy logos are not requirements.) Remember a complete Windows 3.1 system could run on a system with 8MB of storage - this is only a BIOS.
I wish they would do this, every computer I've built up to and including my Core 2 had the simple text interface, and I loved it. Using a mouse in BIOS drives me nuts, as do the stupid fancy graphics which obfuscate menu options and makes things more difficult to navigate. It adds nothing but frustration to the experience.
Just do the smart thing and buy x570--or B550. Skip all of the hassle this will entail. Support for B450/X470 will come a good amount of time after the x570, B550 Zen 3 CPUS are launched and shipped, and bios support for Z3 AGESAs will always lag behind such support, just as they have with Zen 2. But if AMD goes the forked bios route they are talking about--how does this not entail buying a new x470/B450 motherboard? What about PCIx4 mode support? If it does, then I cannot see a reason to want to do this at all. Just buy an x570/B550 mboard, instead.
Is there a technical reason they can only read 16MB BIOSes? If not I'd like to see them support a much larger BIOS for the AM5 package so they can really deliver on the idea (I know they didn't *promise* support for all future Zen generations, but one could reasonably assume all AM4 platforms would work with any AM4 chip) of being able to buy a motherboard for Zen 4 and have it work with Zen 5, 6, and 7. I assume they still have time as Zen 4/AM5 won't be out before late 2021 or early 2022
In the table it says that the A320 chipset doesn't support Ryzen 3000 series CPUs. This is at the very list not generally true. I have an ASUS PRIME A320I-K motherboard, running an AMD Ryzen 5 3600. The CPU is officially supported: https://www.asus.com/Motherboards/PRIME-A320I-K/He...
i meant the A320M-HDV board, of course, and the max. TDP allowance for CPU's on this motherboard. Sadly, the Ryzen 9 3900 is quite more expensive atm than the 3900X-model
Zen+ was not originally officially supported by first-generation Ryzen chipsets (A320/B350/X370). Instead, AMD said that support should be possible, but left it up to the motherboard manufacturers themselves. I believe AMD was being overly cautious due to increased TDPs and the concern that the VRMs on existing boards might not be good enough to handle the power draw. The 1700X was originally a 95W part but the 2700X bumped that up to 105W.
As far as I knew, Zen+ support was still unofficially supported, but the same thing happened again when Zen 2 started shipping, with support on first-gen chipsets being left up to the motherboard manufacturers. There's very little difference between X370 and X470, so other than concerns about VRMs, which seem largely unnecessary, I have no idea why this was assumed to be such a problem. Zen 2 support on X470 was no issue.
That said, this is the reason why Ryzen 3000 works on A320 boards, because the motherboard manufacturers enabled support for them.
Maybe they already knew there were ROM problems on lower-end boards and didn't want to store up more problems for themselves when the next set of CPUs came out.
I'm glad AMD is allowing the new chips to be used, but really now, this verified chip nonsense? Sure, downgrading to earlier BIOS versions may be a potential issue. But two side-by-side BIOS versions, with the same date, one for the new chips, one for all the old chips, with free switching between them would avoid the issue of any motherboard not being able to support any chip. Why gratuitously make problems for people?
If I good remember the original AMD statement was that AM4 platform will support first tree generation of processors. So depending if Zen+ can be considered a generation the support might end with Ryzen 3000 processors. If not I expect support for Ryzen 4000 for all but not the most cheap and BIOS constrained motherboards. I have a Asus B450 Plus mobo with a Ryzen 5 2600X, I will have anyway the option to upgrade up to a Ryzen 9 3900X that's more than enough.
I must give AMD credit for backtracking and saying we will try our best to make it happen. You going to lose the fancy graphic or customized option but you have to start somewhere. s most motherboards have 16 MB, and the CPUs can only address the first 16 MB of a BIOS chip. I think it time to upgrade the 16 MB to maybe 32 with room to spare for down the road.
I don't understand why having another forked bios is an issue. I have a B450M Pro 4 and it already has two different sets of BIOSs, one for 1000 and 2000 series and one for 3000 series.
I would like to point out that the chart is incorrect. Ryzen 3000 CPUs are listed as beta for B350/X370. I own the ASRock X370 Fatal1ty Professional Gamer motherboard and Ryzen 3000 was officially supported with UEFI edition 6.20, nothing beta about it.
"AMD reiterated to AnandTech that after the launch of B550 into the market, they do recommend the B550 motherboards as the best option for Ryzen 4000 support in upgrades."
I'm relieved, excited, confused and a little apprehensive all at the same time regarding these two official announcements made by AMD in only two weeks. The well deserved community backlash and now this travesty of requiring purchase / ownership confirmation of both 400 series motherboard and Ryzen 4000 CPU seems too controlling in my opinion. It goes against the simple concept of buying an unlocked hardware and being forced to lock it up to have support for a future CPU. I don't really understand the reasoning behind AMD stipulations here. Just give the Zen 3 microcode to the board partners and let them post updated BIOSes online like they've been doing for decades. Clueless users will brick their boards and increase RMAs? Well, tough sh*t. Read and follow instructions and make sure you ask all questions about the flashing process BEFORE flashing anything in your system. When in doubt, call your boards' tech support number and have them walk you through the flash process. It's not rocket science.
It's easy to say "tough sh*t" when you're not the one on the hook for the support bills.
Remember, AMD only just got done losing the dumbest lawsuit ever against people who decided they could pull a fast one by arguing over how AMD defined a "core" in Bulldozer. They know the true depths of ignorance some customers are capable of plumbing.
In Bulldozer it was actually not reasonable to define their core as a core. It was reasonable to define it as a core imo in Piledriver onward. Certainly wasn't a dumb lawsuit.
They won't have to worry about processors past Zen 3 in AM4 in any case. Zen 4 will almost certainly require DDR5 memory, and should come with a redesigned socket so you can't accidentally put the new processors into the old sockets or vice versa. But please, AMD, keep the cooler mounting arrangement the same!
Nice and well informative blog. QuickBooks is provided with the QuickBooks Connection Diagnostic Tool to resolve the issues of network and various other issues occurring in it.To know more visit https://accountingerrorsolution.com/learn-how-to-u...
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
164 Comments
Back to Article
edzieba - Tuesday, May 19, 2020 - link
Can't see many ways to achieve that without either having to split BIOS versions and have user's hopefully flash the right one, or sunsetting support for earlier generation Zen CPUs on those boards for everyone.psychobriggsy - Tuesday, May 19, 2020 - link
The download will require you to key in the Zen 3 serial number or some other way of evidencing that you have the Zen 3 to hand, so you can boot the board back up after you install it - i.e., you'll have to have a Zen 3 CPU ready to pop in to run the BIOS.It'll forever be a BETA BIOS, and clearly labelled with Zen 3 support. If they're sensible, it will ask the user to confirm they have a Zen 3 CPU ready to install before flashing as well.
There will be some crossover I expect. They'll likely remove Athlon-A and Zen 1000 (non-AF) support, maybe Zen 2000 APU. Depends on how much space they need. So it might still support Zen+, Zen 2 even with the added Zen 3 support.
peevee - Tuesday, May 19, 2020 - link
These CPU compatibility packs must be very similar to each other and as such easily compressible.16MB is plenty for any BIOS if they don't just waste space on uncompressed logo images etc.
shing3232 - Tuesday, May 19, 2020 - link
nah, X370 has to cut support for AM4 old APU to support Zen2 .Dolda2000 - Tuesday, May 19, 2020 - link
Keep in mind that it may not be possible to compress large parts of the BIOS images, since the code in them may be needed even before RAM to decompress them into has been initialized.Kangal - Tuesday, May 19, 2020 - link
I was thinking the same thing.What pisses me off the most, was I helped build some systems with Zen1, and convinced many buyers to skip the B350 and upgrade to the X370 motherboard instead. My argument was that in the future, their mainstream X370 board was going to help them upgrade from the 1500X/1600/1700 CPUs to the 4600/4700/4900 CPUs. And that was back in late-2016.
So now, essentially those owners have to replace their motherboards, if they want to upgrade their CPUs... which is exactly the opposite of what AMD promised with AM4. I feel sorry that they spent extra money which they didn't want to, based on my recommendation. I've definitely let them down, and probably lost their trust.
Alistair - Tuesday, May 19, 2020 - link
AMD promised 3 years, and they got 3 years. They can also get a NEW chip design with higher IPC, and double the cores. I think they can't complain. The issue with B450 support is different, people are buying those now, and MAX versions, specifically with the expectation of future support.rarson - Tuesday, May 19, 2020 - link
AMD has always provided CPU support based on the socket rather than the chipset. Having built many AMD computers over the last 20+ years, this has always been the case.If you recall when Zen+ shipped, AMD's official statement was that the first-generation chipsets didn't officially support Zen+, although it was possible for motherboard makers to add support via BIOS updates. This was a huge surprise to me, as I assumed that every AM4 board would support every AM4 chip including Zen 3. I would not have invested so heavily in a brand new Ryzen 1700X build if I knew that not just Zen 2 and 3, but Zen+ support would have been questionable.
I also wasn't anticipating the release of another chipset alongside the release of Zen+. What's even more annoying about this is that X470 offers very little improvement over X370 anyway, and while it officially supports Zen 2, X370 does not. No motherboard maker in their right mind would forego Zen 2 support for higher-end boards that easily support those CPUs.
Now, it seems AMD is officially trying to put an end to X370 CPU support, by excluding Zen 3. It will be interesting to see if boards like the Crosshair VI Hero get unofficial support because I don't see a damn reason why it's not technically possible. If X470 can support it, X370 can, too. I do hope that X370 boards get support somehow, but I'm extremely disappointed that AMD has gone this "unofficial support" route from the moment they started shipping Zen+. Zen freaking plus.
Ryzen 3000 is a very decent upgrade option from Ryzen 1000, but it's not what the majority of people who have used AMD for years were expecting. We were expecting the usual socket-based CPU compatibility, which means Zen 3. AMD users are used to AMD changing the socket when changing CPU support.
Spunjji - Wednesday, May 20, 2020 - link
I'm getting the distinct impression that some people are mixing up memories of their assumptions when Zen came out with information that wasn't available until later. I definitely don't think it's fair to say that people were "expecting" Zen 3 compatibility - unless I'm seriously misremembering things, I didn't think we knew back then if AM4 would live past Zen 2, let alone that Zen 2 would be such a robust upgrade.That said, I'm generally in agreement that X470 compatibility with Zen 3 really ought to imply X370 compatibility. There must be a reason why not, and it would definitely be better if they explained that. The Zen+ X370 mess was absurd, too.
rarson - Wednesday, May 20, 2020 - link
Zen, Zen+, Zen 2, and Zen 3 were all shown prior to the launch of Ryzen to be on the same AM4 roadmaps. Zen 4 was always intended to be the clean break.watzupken - Tuesday, May 19, 2020 - link
There is no way to please everyone. X370/B350 owners may feel left out, but its able to get support for 3 generations of CPU, which I don't think is a bad thing. The jump from first gen Ryzen to the 3rd gen is anyway a decent upgrade. Switch to Intel side, and you probably only get 1 generation of CPU support and MAYBE another generation if Intel feels generous enough.rabidpeach - Tuesday, May 19, 2020 - link
im waiting to build 1700 based system. in 3 years when the 3900x will be available for 100-150 i will upgrade to it after updating the bios on the x370 motherboard. not yet feeling left out. :Drarson - Tuesday, May 19, 2020 - link
Zen+ is little more than a re-spin, not an actual generation of CPU. They called it 2000 for obvious marketing reasons.dotjaz - Wednesday, May 20, 2020 - link
So? B350/X370 actually supported A-series APU, then Zen/Zen+, and if you want Zen2, you lose A-serious and Zen.I upgraded from A6-9500 to 2200G, flashed the beta BIOS so I can upgrade to 3600 if I want to.
rarson - Wednesday, May 20, 2020 - link
That's actually incorrect. B350 and X370 didn't officially support Zen+ when it launched. AMD said support was possible, but left it up to the motherboard makers to do it. The same goes for Zen 2.APU support benefits me in no way whatsoever. In the past, AMD gave them their own socket, which made sense. Why would anyone buy an expensive X370 board just to put an APU in it? Never mind the fact that the only APUs available at the launch of AM4 were old Bristol Ridge chips.
My point about Zen+ is that it's not really a generational step, but a warmed-over refresh much like Intel has been doing since Skylake. Going from Zen to Zen+ is a lateral move. Zen 2 is the second evolution of the Zen architecture.
Spunjji - Wednesday, May 20, 2020 - link
APU support may not benefit you specifically, but it's definitely helpful to AMD and OEMs who don't have to support two parallel sockets.rabidpeach - Tuesday, May 19, 2020 - link
lost their huh-what now? on x370 with ur manufacturer bios supplied you can upgrade zen 1 1700 style processor to zen 2 3700x or even 3900 for 16+ cores if ur bios is upgraded from the manufacturer. this is quite a bit jump over 3-4 years. they shouldn't be let down.Samus - Wednesday, May 20, 2020 - link
They can still upgrade all the way to a 3rd gen which is still equivalent to Intels current gen.I mean hey, I’m in club intel and we generally get support for one or two generations of CPU per chipset, not three or four generations.
What’s really annoying about Intel is they KNOW they could support everything from Skylake to Comet Lake, because they are all the same power envelope, manufacturing process node (admittedly somewhat irrelevant) and PCIe/memory bus. But it’s presumable they have the same concern AMD does...how to fit support for that many CPU’s in a BIOS. And Intel makes a LOT more individual models of CPU’s per generation.
rarson - Wednesday, May 20, 2020 - link
"I mean hey, I’m in club intel and we generally get support for one or two generations of CPU per chipset, not three or four generations."No shit, that's why I typically avoid buying Intel.
Had AMD not set expectations by doing the same thing for 20+ years, I wouldn't have minded so much. Had they given their customers a heads up before launch that, hey, socket support no longer means CPU support, I wouldn't have minded so much. The fact that they just decided to completely reverse course after decades of doing the same thing without bothering to tell anyone is a bit upsetting to anyone that bought in early based on years of expectations.
Spunjji - Wednesday, May 20, 2020 - link
Really? Because I don't recall Zen 3 even being roadmapped back then. In fact, I just did a quick search and I can't see it showing up in their slides until 2018. If you gave that specific advice that they'd be able to upgrade to Zen 3 then it was bad advice.Thing is, most (not all, but most) X370 owners can jump up from an original maximum of the 1800X to, say, a 3800X for a ~25% performance boost with the same number of cores - or a 3950X to get twice the number of cores. So advising that they'd get a solid upgrade path was totally reasonable, and remains that way.
Side point - you couldn't get that kind of performance uplift out of a 2-generation Intel upgrade any time between Sandy Bridge and the present, and you had precisely zero chance of doing it on the same board. I'll take "some chance" over "none" any time.
rarson - Wednesday, May 20, 2020 - link
Zen 3 was on slides before Zen even launched.https://videocardz.com/69571/amd-reveals-vega-and-...
It's really annoying to keep hearing the Intel argument. Yes, we know that Intel doesn't often carry over CPU support, that's one of the reasons why we buy AMD. Lots of people buy AMD because their CPU support was based around the socket, and you didn't have to worry about new CPUs dropping support for different chipsets. The fact that they did this right out of the gate, by releasing X470 only a year later and not officially supporting Zen+ on X370 was infuriating even if motherboard manufacturers did add support. If AMD wants to shift CPU support to the chipset, then fine, but they should have told their customers first. I wouldn't have bought an X370 board at launch if I had known AMD was going to pull this stuff.
There's no technical reason why X370 can't support Zen 4, and a large reason why this problem even exists is because AMD decided to put all of their APUs on the same socket as the CPUs for some reason. Had they just kept them on different sockets, this probably never would have been an issue.
Spunjji - Friday, May 22, 2020 - link
Not to keep this going indefinitely, @rarson, but those slides are from 2 months after Ryzen launched.The "Intel argument" keeps being made because it's relevant. None of this tech exists in a vacuum. I think they're doing a decent job all things considered. Could the situation be better? Yes. Could it be a lot worse? Oh my, yes.
Pointing to older architectures and the decisions made then as a reason to expect them to stay the same isn't entirely helpful, either - AMD were in a woeful competitive position from AM2 onwards.
rarson - Sunday, May 24, 2020 - link
My mistake, I was thinking Ryzen launched in May, not March.That being said I'm fairly sure that I saw that slide before May, and before Ryzen launch. I could be mistaken, but the existence of Zen 3 and its use of AM4 was at least rumored before Zen launched and AMD would have been planning all of this well before they released the slide anyway.
The "Intel argument" is not relevant here. Not for AMD customers. I've already explained ad nauseum why this is the case. After decades of chips being tied to sockets, the least AMD could have done was to let people know they were changing the rules. One of the reasons why I've built so many AMD systems in the past and so few Intel builds is because of the fact that upgrade paths were much clearer.
In this case, pointing to older architectures is the entire point: decades of setting customer expectations led to people assuming nothing had changed.
headloser - Wednesday, May 20, 2020 - link
It not your fault. You couldn't predict how the 4000 series are going to affect the BIOS. I think AMD need to upgrade the BIOS 16mb to maybe 32mb. Since the CPU are becoming more powerful and has the ability to do more, they need to build BIOS adaptability for future upgrade.Arnulf - Wednesday, May 20, 2020 - link
Perhaps they should have gone with Intel which was selling 1xx/2xx series chipsets for Skylake/Kaby lake in 2016 ...Samus - Wednesday, May 20, 2020 - link
While I agree a GUI BIOS is a waste of space (personally I hate them, they are slow to navigate and less intuitive) it will still be a huge challenge to fit microcode for 100 individual CPU’s (taking into account individual designs, stepping, generation, etc) into 16MB. Most microcode files are around 200kb give or take. You still need UEFI drivers, USB legacy initialization for HID devices, a network stack for live BIOS updating, and so on.I think AMD is in a tough spot here, and the consensus is they should just drop 1000 series support since they have the most errata (and hence the largest microcode updates, though they don’t “have” to use the latest microcode I guess...) and I can’t imagine people running first gen parts anymore on brand new boards unless their old board failed.
It’s really interesting to see AMD commit to supporting a socket interface this long.
rarson - Wednesday, May 20, 2020 - link
They should have just left the APUs on their own socket.rarson - Tuesday, May 19, 2020 - link
AMD can't eliminate support for the older CPUs as these are popular options in foreign markets.I'm all for getting rid of the stupid graphics in BIOS. I've always hated fancy interfaces that make it more difficult to find and change the settings that you need, especially since once I get the BIOS set up as I want, I don't intend to continue going back into it anyway. I prefer the old school low-res, text-based, keyboard only BIOSes. I don't need a freaking "EZ mode" or any of that nonsense, I just need a clean, simple menu that I can easily navigate through.
zmatt - Tuesday, May 19, 2020 - link
I think it was GN who said that if they went back to old school tabulated interfaces in the BIOS then reviewers would complain and make them look bad. I say fine. It lets me know who to take seriously. The A E S T H E T I C S of the BIOS shouldn't matter. Its there for setting system configuration and maybe overclocking. Not to look cool. Most people will only ever be there once or twice.Any reviewer who values how a flashy interface over practical support and features isn't credible in the enthusiast space. They can go review Apple products for all I care.
rarson - Tuesday, May 19, 2020 - link
Yeah, I generally agree with Steve and I understand his sentiment, but I don't think the issue would be as big a deal as he thinks if it means Zen 3 support is added to existing products. Surely the people who are buying and using them wouldn't mind the trade-off.I WANT a BIOS that looks old, because those old BIOSes were stupidly simple to navigate and change settings. I don't want to have to use a mouse to change settings. These stupid graphical UEFI environments often leave me wondering where exactly I need to click on something or what is a dropdown/toggle, etc. A simple text interface is infinitely cleaner and easier to use. A graphics-less BIOS is a feature that I actually want, not a detriment.
zmatt - Tuesday, May 19, 2020 - link
For clarification, Steve agrees that the flashy interfaces are dumb and would be fine with simple ones. He just argues that there are many in the industry who would complain and give AMD and motherboard makers bad press if they did that.rarson - Tuesday, May 19, 2020 - link
Yeah, I know. I was trying to make that clear in my own comment as well, but it's too easy to read both ways!docwily - Wednesday, June 17, 2020 - link
I dont know why they dont just have swappable bios chips you can buywilsonkf - Tuesday, May 19, 2020 - link
AMD: We couldn't make old motherboard support new CPU because the CPU support list is too long ... OK we will try.Intel: We couldn't make old motherboard support new CPU, because ... maybe we actually could, yea, but we simply don't want.
XabanakFanatik - Tuesday, May 19, 2020 - link
The difference is AMD promised to support AM4 for 4 years, which implies they will support every COU released on every chipset for that socket.This still isn't what they promised.
They made promises they didn't realize they couldn't keep. Probably because they didn't think about it, as is obvious by the board partners only using 16 MB bioses and the CPUs only being able to address 16MB.
willis936 - Tuesday, May 19, 2020 - link
And? Curbing a 4 year promise by six months beats promising that you will never support a platform for more than two years.ksec - Tuesday, May 19, 2020 - link
Is it me or having been with Intel for a long time you know the same socket doesn't mean the same support.Where did this idea they think AM4 = support of all AM4 CPU came from.
Cooe - Tuesday, May 19, 2020 - link
They NEVER promised they would support every single CPU released for the socket on the same BIOS version... You're pulling crap out of your butt. Literally all that was said was that they'd be supporting AM4 through 2020. Anything else is simply you reading way too damn far into things.rarson - Tuesday, May 19, 2020 - link
Some of you guys are missing the point: AMD has been doing this for decades. People assumed that AM4 support means that all AM4 CPUs would be supported because that's the expectations AMD has set for their customers, FOR DECADES.Every single previous AMD socket has supported every single AMD CPU for that socket, and then some (the plus sockets even supported newer CPUs in older boards). Every single one.
I built my first AMD system in 1997 and since then, have rarely used Intel not only for pricing reasons, but because of chipset compatibility. That is to say, with Intel it's always been a crapshoot as to what chipsets will support what processors, but with AMD it has always been determined by the socket. When AMD changes CPU support, they've always changed the socket. Without fail. Until now.
You guys who are thinking this is some kind of bizarre assumption clearly haven't built many AMD systems over the last 20 years.
misel228 - Tuesday, May 19, 2020 - link
I was gonna post the same but refreshed before I started typing. :)I'm on AMD since Super Socket 7 (good old K6-2 300MHz). And it has always been the case that if the CPU physically fits the socket the chipset would run it. At least I don't remember a CPU where that wasn't the case. If you know of one, let me know.
silverblue - Tuesday, May 19, 2020 - link
AM2+/AM3 CPUs are a special case. A Phenom X4 9950 won't work in an AM3 board because it doesn't have a DDR3 IMC, whereas a Phenom II X4 940 will work in both sockets thanks to having support for both DDR2 and DDR3.rarson - Tuesday, May 19, 2020 - link
This is true due to the memory controllers, but all of the "plus" chips were still backwards-compatible with the older sockets. I mean, it's kind of a special case if you need to put an older chip into a newer board, not many people would attempt that.drexnx - Wednesday, May 20, 2020 - link
this was never true though, even back in the socket A days, the mixes of different FSBs and chipset support meant you couldn't expect true forwards compatibility. Try putting a 2002 XP 2600+ (333) (which came in both 333 AND 266 FSB versions for extra confusion!) in that 2001 KT133A board you bought for your XP 1800+, it wouldn't work, and they came out 10 months apart, not 3.5 years apart.Spunjji - Wednesday, May 20, 2020 - link
This pretty much summarises why I feel the way I do about this. The situation is still way better than it was at many points in the past, so while I get some level of disappointment and am pleased by AMD's change of heart, I also don't understand why this appears to have stirred up such opprobrium.rarson - Sunday, May 24, 2020 - link
"I also don't understand why this appears to have stirred up such opprobrium."Then you're simply not reading what has been written. This was one example in over 2 decades of chips.
There are always edge cases of certain boards not supporting certain CPUs. Usually this comes down to BIOS updates. My IP35 doesn't support all Core 2 chips because Abit went out of business before finishing the BIOS updates. There are even certain CPU instructions that aren't supported with older BIOS revisions (I ran into this problem when trying to install Windows 8 64-bit on that board and had to download an unreleased beta that someone had modified to make it work). That's totally different than the CPU manufacturer just outright deciding not to support chips that technically would work fine with a given chipset (like Zen 3 on X370). Yes, I understand the reasons why AMD would make such a decision, but to go back and offer support for X470, which is practically the same damn thing, is a bit absurd. Nobody that bought X370 expected AMD to support ONLY Zen 1 (not even Zen+), which is what they did right from the start.
Yet we should feel "lucky" that we got unofficial support for Zen+ and Zen 2. This is not how this normally works with AMD platforms.
rarson - Wednesday, May 20, 2020 - link
That's kind of a unique example though. Even in the Pentium days boards had configurable FSBs, and it was just a matter of setting jumpers (or if you had a jumper-less board like I did, changing settings in the BIOS). And even if your board didn't support a higher FSB, you could still run faster chips using different multiplier settings.Regardless, once AMD moved the memory controller to the CPU, extended support became the norm, especially with stuff like the "plus" sockets.
Andyburgos - Thursday, May 21, 2020 - link
Was about to post the same as drexnx.This is not quite true. I think Ryzen 3000 / X570 socket could have been AM4+ to resemble what happened in AM3/AM3+ (granted, they were different architectures). But if you take a look to Socket A (462)...
It is not FSB frecuency that defines socket A compatibility, but most likely it is the package. Most KT133A boards will run ceramic Palominos just fine, but will refuse to boot any T-bred. Not all nForce2 boards will run Thunderbirds (I think this is based on multiplier rather than package).
Same goes for AM3 and AM3+: I have a M4A79T wich will not take a Bulldozer, even if it's power delivery and RAM cupport should allow it.
I think we can just conclude that naming has been pretty much FUBAR for Ryzen (older gen APUs with current gen names, B550 release date closer to 4000 than 3000 series processors) and call it a day.
Spunjji - Friday, May 22, 2020 - link
Two thumbs up to this comment.FreckledTrout - Tuesday, May 19, 2020 - link
Technically if AMD only allows x670 boards to work with the new Zen3 chips they still have met the standard of using AM4. They never said your x370 board would work with CPU's 3 years later.rarson - Tuesday, May 19, 2020 - link
Yes, they never explicitly stated that X370 would support Zen 3, however they've spent decades setting customer expectations by ensuring CPU support based on socket compatibility.FreckledTrout - Tuesday, May 19, 2020 - link
True socket support in the past meant CPU support but I simply don't see any reason to get outraged. It’s not like they are arbitrarily limiting support or Like Intel creating a new socket for no technical reason. They have real reasons of fitting all that AGESA code into the BIOS.rarson - Tuesday, May 19, 2020 - link
I agree with you completely on the outrage part. People are reacting a bit too extremely about this. But to me, this whole situation is a little disappointing for a few reasons. Primarily, I'm disappointed that AMD was caught so off guard by this problem, as BIOS limitations should have been pretty obvious before they started shoving everything onto AM4. Previous generation APUs had their own socket and I'm not sure why AMD felt the need to consolidate everything to AM4, especially the pre-Zen stuff like Bristol Ridge. They already had Zen, Zen+, Zen 2, and Zen 3 on the public roadmaps LONG before Ryzen was launched and should have known well before this that the CPU support list was going to become massive with all these different chips to support. Basically, it's a little embarrassing for a company who has manufactured CPUs for this long to run into this problem, let alone be blind-sided by it. I'll chalk it up to ambition, but they should have seen this coming.Also, CPU support based on the socket always made planning builds with future upgrades in mind a lot easier with AMD. Given that they have usually been the budget option, this is plus for their target audience, it makes going with AMD an even easier decision. If AMD wants to shift CPU support to the chipset, that's fine. I may not like it, but if that's what they determine they should do, then I accept that. However, it's not right to suddenly introduce this change when you're 75% of the way through your socket road map and many of your loyal customers bought in at the beginning under the assumption that the company would continue maintaining support the same way they had for decades.
And honestly if they do shift to a chipset-centric support model (which, let's face it, they have because X370 support is done), that will make me think twice the next time I want to build an AMD computer. I'm not saying that I'm boycotting them or anything, just that I'll be a little more weary about CPU support.
JoeWright - Tuesday, May 19, 2020 - link
Lolwut?Why'd you wait 4 years to complain about this?
Couldn't you have made similar complaints right at the start claiming that amd failed to keep their promises because cpus like 1800x and cheap a320 mobos existed at the same time?
rarson - Tuesday, May 19, 2020 - link
Part of the problem is that in the past, AMD has typically changed the socket when changing CPU support, and even then, as was the case with the "plus" sockets, still retained some compatibility between new and old boards and CPUs. Usually, AMD would announce the socket and state how long they intend to support the socket, and that was it. Intel on the other hand was always tied to specific chipsets, and the socket had no real bearing on what processors would be supported by a given board. This has always somewhat frustrated me about Intel, because you can never assume that a given chipset will support future processors. Case in point, I feel like anyone that invested in Z270 got really ripped off. That aside, I will give Intel a small bit of credit for avoiding this kind of mess completely by setting customer expectations.I built my first PC in 1997, and have always switched between AMD and Intel, based on my budget, the parts currently available, and the best bang for the buck with upgrade path in mind. Granted, parts are a lot cheaper than they used to be (graphics cards aside), but generally I start with a good motherboard and a cheap CPU so that I can upgrade down the road. It's ALWAYS been easier to plan very far ahead with AMD, because you knew that when they said they would support a socket for a number of years, the newer CPUs would work with all boards that used the same socket. In fact, when they first announced AM4, I assumed that meant that every AM4 board would support all CPUs up to and including Zen 3 (based on the timelines it seemed pretty clear that the third Zen iteration would arrive within that support window). It's a little frustrating to have AMD suddenly change CPU support without changing the socket after decades of doing the opposite. I would have rather seen them change the socket for Zen 3 than continue with the same socket while breaking support. But it's pretty clear they realized far too late that this BIOS thing would be a problem.
I don't necessarily fault them for being overly ambitious, however it kind of makes you wonder why they bothered with stuff like Bristol Ridge on AM4 or why a company that has been making CPUs and chipsets for decades wasn't more aware of the BIOS problems arising from supporting too many CPUs on a single socket. If it comes down to removing BIOS graphics altogether and going back to a normal, plain menu interface then I'm all for it, as I really loathe useless BIOS graphics.
mkaibear - Tuesday, May 19, 2020 - link
Actually, and I can't believe I have to keep pointing this out, it's closer to Intel: "We tried to make multiple generations work on the same socket before and we got absolutely *slated* by the enthusiasts and press when changing power budgets meant not every chip worked - so now we don't make promises about how long sockets will last"rarson - Tuesday, May 19, 2020 - link
Intel has never made promises about how long sockets will last, CPU support has always been tied to the chipset and not the socket.What Intel got blasted for is creating warmed over refreshes (Kaby Lake) and useless chipsets (Z270) fully aware of the impending release of higher core chips with Coffee Lake.
DigitalFreak - Tuesday, May 19, 2020 - link
How many people really upgrade their CPU and keep the same motherboard? Seems like a solution looking for a problem.ArcadeEngineer - Tuesday, May 19, 2020 - link
Surely everyone if they could? What's the reason for a random consumer to throw away a perfectly good B450 board and buy B550?PhysicsNurd - Tuesday, May 19, 2020 - link
PCIe 4 NVME SSD support?Ratman6161 - Tuesday, May 19, 2020 - link
I recently upgraded from a 1700 + X470 to a 3700X + X570. But I almost decided to just put the 3700X in the X470 board. I don't currently have anything that takes advantage of PCIe 4 so there is really no performance advantage to be had from the motherboard upgrade. I mainly did it for future upgrades and because I could use the old motherboard and CPU in another system.Guspaz - Tuesday, May 19, 2020 - link
That provides no real-world performance benefit over installing that same SSD in a PCIe 3 slot.Spunjji - Wednesday, May 20, 2020 - link
That's a reason, but not a good one!xenol - Tuesday, May 19, 2020 - link
The random consumer likely doesn't even know how to upgrade their CPU, and I'd wager a lot of people run their computers to the ground. Upgrading is a moot point by then.danbob999 - Tuesday, May 19, 2020 - link
Most people don't upgrade every year so I agree. However those with a Ryzen 1000 could upgrade to 4000 if their motherboard supports it. Always good to have the option.Irata - Tuesday, May 19, 2020 - link
Yup, that‘s it. The good thing is that older gen Ryzen get very cheap because they are competing with the new models for the same customers.Also, at least for me installing a motherboard is the part about building you own that I like the least. So not having to do that while being able to get a noticeably better CPU is a great plus for me, even if it means losing out on some of the new features like PCIe 4.
Got my Ryzen 2700x for €150 new on special offer a few months ago and being able to upgrade to either a cheap Ryzen 2 or 3 down the road is a nice outlook. Heck, with a 12 core Ryzen 4000 I may even be able to get much better performance in the same or lower power budget.
rarson - Tuesday, May 19, 2020 - link
At this point in time, given that every X570 board has a fan on its chipset to cool the thing, I'd consider the lack of PCIe 4.0 to be a positive, not a negative. Those tiny little fans clog with dust and seize up so easily, I don't want any part of my PC relying on one of those things for cooling.There are no PCIe 4.0 graphics cards yet, and even if there were, I doubt running them at 3.0 speeds would be any kind of bottleneck. Similarly, for NVMe drives, while the speed difference likely exists, I probably would never know the difference comparing the two side-by-side for my normal computer usage.
Makaveli - Tuesday, May 19, 2020 - link
you live in some kinda dusty workshop?That little fan you speak of I've never heard once in the 6 months I've been on a X570 board. I would also never let the inside of my PC get to the point where dust will clog a fan. This isn't the 90s anymore the fans are of decent quality and quiet. Anyone using an Air Cooler will hear more noise from that than the chipset fan.
rarson - Tuesday, May 19, 2020 - link
No, I work in PC repair, where I see these tiny fans die all the time. It's cute that you think your 6 month-old X570 board is a valid example case, but having seen years of motherboards and graphics cards with these fans die from the slightest amount of dust, I'm just going to go with my experience over yours.If you've ever worked with a large amount of these things, you'll know just how little the amount of dust is required to seize these things up.
Spunjji - Wednesday, May 20, 2020 - link
Not all fan designs are the same. I too have been burned by years of dodgy fans, but there is a greater prevalence of FDB fans over the older sleeve bearings that wore out so easily with heat and dust.I don't know for sure whether most X570 fans are FDB or not, but based on people's comments on the noise they make and the cost of the boards involved, it seems quite likely many are.
lmcd - Friday, December 4, 2020 - link
The 6 month comment on X570 is funny because my X79 board is 8 or more years old and still works great. Yes it's another point of failure, but it's far from a guaranteed point of failure.supdawgwtfd - Tuesday, May 19, 2020 - link
All RX series Radeon GPUs are PCIE 4.0...They have existed for quite some time now...
rarson - Tuesday, May 19, 2020 - link
Months. Yeah. Quite some time. And of no benefit whatsoever.Irata - Wednesday, May 20, 2020 - link
I think nVidia‘s and AMD‘s next gen GPU may change that. Having PCIe 4 for the top of the line models will probably be a plus.Same for nVme storage once the newer controllers come out.
But yes, right now it‘s not a problem and if you are not going for the high end GPU wise it should also be OK.
Regarding fans - I have no big issue with them but not needing to have one on the mainboard is a plus, so B550 would be the preferable option for me.
With this regard I am kind curious about the upcoming Z490 boards as they seem to have several smaller fans to cool the VRM.
Hrel - Tuesday, May 19, 2020 - link
Presumably the same reason someone would "upgrade" from a 3000 series to a 4000 series. My guess is they have money, too much of it, they hate money. So they spend hundreds of dollars on a part that isn't anywhere near being a bottleneck, literally every year.My grandpa ran A dealership and he has customers who would buy a new car, drive it to the first oil change, or 3 months, or 6 months, or one year, then buy a whole new car. Often the exact same thing. They simply had so much money this was more convenient to them than dealing with any maintenance at all. It also put some very clean used cars into the used car market and my grandpa got some very nice margins on both ends of those deals.
Cause yeah, motherboard/cpu/ram should last you 10 years easy. 5 minimum. Upgrading just the CPU hasn't made any sense at all since LGA 775.
rarson - Tuesday, May 19, 2020 - link
Upgrading just the CPU makes perfect sense if you buy a bang-for-the-buck CPU at the start of the build, and then upgrade to the best CPU you can a few years later after prices have dropped significantly. That's what I've always done, and even after buying 2 CPUs I've still paid significantly less than the initial cost of the high-end CPU.silverblue - Tuesday, May 19, 2020 - link
This. I have a Gigabyte GA-AB350-Gaming 3 which was originally paired with a Ryzen 5 1600 but now sports a Ryzen 5 3600. The cost of both CPUs is a little higher than the launch price of the Ryzen 7 1700X, and the 3600 can outpace it in most areas despite a two core deficit. The motherboard is still fit for purpose, even if the LED strip on the left side of the board has turned a shade of green.Jimster480 - Tuesday, May 19, 2020 - link
Lol this is the dumbest ever comment I have read in the tech space.Upgrading the cpu doesn't make sense? I went from a ryzen 5 1600 at launch for 225 to a 1700x for $215 (sold the 1600 for $155) a half a year later and used that until the 3900x came out. I started with a b350 tomahawk arctic for 85$ on sale. Sold that board for $105 7mo later when it was discontinued.
got an x370 carbon (for $77 on sale) and have been using that since then.
upgrading the motherboard in a windows machine requires a new license or a complicated activation process (which isn't always successful as it depends on your windows key type).
people who upgrade motherboards and spend their time for no benefit are the ones who love to waste money.
If I sold my 3900x now (which is a great bin btw) I could get 380$ for it. Considering I paid 450 last August with my best buy birthday discount... That would be $70 for almost a year of use; and I could go to a faster 4000 chip again.
this is by far the most economical way to upgrade and always stay on the newest platforms.
being forced to purchase all new hardware to upgrade the cpu just basically means that you will lose alot of money to upgrade and are likely to keep your pc until its worthless and then spend far more doing upgrades.
bigboxes - Tuesday, May 19, 2020 - link
Yeah, changing your car's motor oil is a lot easier than buying a new car. Even if you do it yourself. Nice story though.XabanakFanatik - Tuesday, May 19, 2020 - link
I personally know people that bought into AM4 at x370 because AMD lied about the socket support.Cooe - Tuesday, May 19, 2020 - link
AMD never lied. You're a misremembering salty fanboy. (And lol we all know that "people you personally know" is you). You got 3 full generations of CPU's on the same board, including the MASSIVE leap that was Zen 2. AMD continued supporting your board through 2020 as they promised (and even into it, with the 3300X/3100). I simply cannot fathom how a 300 series owner can be that mad at having a board that supports nearly 100 CPU's, and thinking they got shafted on compatibility.FreckledTrout - Tuesday, May 19, 2020 - link
As an X370 owner Im not mad. AMD never promised I could use my x370 board for 3 years of CPU generations. If Gigabyte somehow ports support over for the 4000 series, even in a beta bios, I would be thrilled but I am not counting on it.rarson - Tuesday, May 19, 2020 - link
Zen+ is not a new generation of CPU, it's a re-spin. It's what Zen would have been had AMD not had to rush it to market. And while Zen 2 is nice and all that, Zen 3 being supported on AM4 was the expectation.Why should anyone care about lots of CPUs being supported when a good chunk of them are completely irrelevant? Is anyone still using Bristol Ridge on AM4? Why did AMD even bother putting Bristol Ridge on AM4? Does anyone who is building a high-end PC with discrete graphics give two shits about APU support? Why didn't AMD just give the APUs their own socket like they did with the FM sockets?
AMD should have known and planned better for this. As it is, it's pretty clear they didn't realize until way too late that this was a problem. There's no excuse for a company who has made CPUs for as long as they have to run into this problem.
lmcd - Tuesday, May 19, 2020 - link
Ryzen 1 didn't have an APU and Ryzen 2 APUs barely arrived before Ryzen 3 arrived with Zen 2. Bristol Ridge was the placeholder CPU.rarson - Tuesday, May 19, 2020 - link
You're missing the point. AMD's CPU and APU sockets were always different in the past. They should have just left Bristol Ridge on FM2+ or created a new socket for it.Spunjji - Wednesday, May 20, 2020 - link
But that was a PITA for everyone (buyers, manufacturers, etc.) - I can totally forgive them for wanting to resolve that, despite it having created a new problem in the process.tyaty1 - Wednesday, May 20, 2020 - link
APU and CPU sockets used to be different because the used different interconnect.AM2/3 used Hyper Transport
FM1/2 used Unified Media Interface.
AMD CPU-s are SoC-s, so there is no need for dedicated interconnect to the PCH.
tyaty1 - Wednesday, May 20, 2020 - link
*AMD CPU-s are SoC-s, so there is no need for dedicated interconnect to the PCH.AM4 CPU-s are SoC-s, so there is no need for dedicated interconnect to the PCH.
rarson - Tuesday, May 19, 2020 - link
AMD did not lie about the socket support, they are still technically supporting the socket. However, the problem is that for the last 2+ decades "socket support" meant that CPU support was tied to the socket, not the chipset (unlike Intel). For them to suddenly change this is a real problem for customers whose expectations AMD has set for decades. I, too, assumed that AM4 support meant that Zen 3 would be supported on all AM4 boards, because this is how AMD has done it for the 2 decades that I've been building my own computers.While I dislike the idea of AMD now determining CPU support based on chipsets rather than sockets, like Intel does, I wouldn't have a major problem with them doing so had they made it clear from the outset that "socket support" no longer guarantees CPU support. You can't just throw out a huge change like this years after selling people the product on the assumption that CPU support would continue.
Itveryhotinhere - Tuesday, May 19, 2020 - link
It still can hold 16 cores 3950x...better than nothing. As owner of x370+1600 at 2017 now with x370+3700x, I have nothing to complaintDrKlahn - Tuesday, May 19, 2020 - link
I went from a 2700X to a 3900X on my X470. Nice jump in video rendering performance and increased frame rates in older games that are lightly threaded. Sold the 2700X to a friend, so it didn't cost a lot to do. Flash BIOS, remove heatsink, clean things up, install new CPU and you're good to go. No real issues or hassles. If the next gen has a sizable increase in the things I use my PC for, I will do it again. I already know where the 3900X will go. Then probably hold onto that for a few years until DDR5 and CPU advancements entice me to upgrade.deathBOB - Tuesday, May 19, 2020 - link
If you can sell a chip you can sell a board.gerbilicious - Tuesday, May 19, 2020 - link
Do you chop off your entire head if you want to change hairstyles?dullard - Tuesday, May 19, 2020 - link
Chopping off a head is a terrible analogy. An old head is useless to you (it would be a downgrade) and useless to almost everyone else.A better analogy would be: do you replace your clothes dryer when buying a new high-end clothes washer? You could upgrade your dryer at the same time, or you could use your old dryer. You could get a far better dryer while you are at it and sell off your old dryer. Or, you can keep chugging along with your older, possibly slower, used dryer with fewer features.
Upgrading a CPU while keeping an old motherboard is a small subset of the already small enthusiast subset. There just really isn't that many people that do it. If you are extremely budget constrained and short-term budget focused, it might be your only option. But, in the end it is a better long-term financial deal to wait until you can afford replace both.
I personally just can't get over gimping a good new CPU with an old motherboard. It just feels wrong. In this case there isn't too much different from B450 and B550. But, if you need the differences such as PCI 4, the B550 is well worth the upgrade. Sell off your old motherboard with your old CPU and the motherboard upgrade to you costs next to nothing.
PeachNCream - Tuesday, May 19, 2020 - link
"Upgrading a CPU while keeping an old motherboard is a small subset of the already small enthusiast subset."Sums up everything nicely. While we readers of Anandtech generally have the technical background, knowledge, and tools to perform a CPU upgrade with no difficulty or drama, we represent a VERY small percentage of the general population.
guycoder - Tuesday, May 19, 2020 - link
Maybe so but but keeping this small part of the community happy and supported it only brings greater goodwill and standing to the AMD brand and will encourage others to stay with AMD or swap from Intel. Good marketing and PR goes a long way and AMD still needs to keep hungry for customers and revenue. It's also karma for screwing over the Threadripper 2000 customers...bigboxes - Tuesday, May 19, 2020 - link
Why do you care about AMD or Intel? I just want competition. I purchase whatever product gives me the best feature set for the best price. You act like you're going to divorce AMD over this.AntonErtl - Tuesday, May 19, 2020 - link
I don't think the enthusiasts are such a small market for AMD. For Intel that market may be small change, and that's why they do not care much about compatibility, but AMD is not that strong in the OEM market, but very strong in the DIY market. And those who DIY also want to upgrade, but often avoid the cost and hassle of changing the board. And no, for most users an old AM4 board does not gimp a new AM4 CPU; but if it does for you, keeping compatibility does not prevent you from changing the board along with the CPU.rarson - Tuesday, May 19, 2020 - link
AntonErtl is correct, and to add to it, AMD is often the choice for budget-oriented enthusiasts, who may opt for a cheaper CPU at the outset of a build so that they can replace it later on with something better. I've always based my builds this way, buy a good board and a great bang-for-the-buck CPU and upgrade it a few years later. Additionally, anyone that has kept any kind of finger on the pulse of the used market knows that Intel chips are obscenely overpriced while AMD's prices tend to drop as you would expect them to. So buying used is a huge benefit for AMD customers who want to upgrade later on.Enthusiasts are exactly the people who upgrade CPUs. Non-enthusiasts don't generally do that. Not everyone can afford spending thousands of dollars every time a new generation of CPU comes out.
rarson - Tuesday, May 19, 2020 - link
The appliance analogy doesn't even come close. Most people upgrade appliances when their old one dies, and the washer and dryer operate completely independently from each other. Appliances are also major purchases that require delivery and installation. I have absolutely no idea how you think appliances are comparable to CPUs and motherboard."I personally just can't get over gimping a good new CPU with an old motherboard."
But you're not. The chipsets have nothing to do with the performance of the CPU. The only thing gained with X570 of any significance is PCIe 4.0, which isn't even supported by graphics cards yet and is inconsequential to the average user from an NVMe performance perspective. If the CPU had a DDR5 controller in it, then sure, you'd probably be somewhat gimping the CPU, but the rest is basically irrelevant. I don't spend $200+ on a motherboard just to replace it in a couple years, that's just dumb.
dullard - Tuesday, May 19, 2020 - link
"I have absolutely no idea how you think appliances are comparable to CPUs and motherboard." Because you use them in pairs, and you can upgrade one without upgrading the other. Same with CPUs and motherboards: you use them in pairs and can upgrade one without upgrading the other. That is a decent analogy. Comparing upgrading a motherboard to cutting off your head is not a good analogy."The chipsets have nothing to do with the performance of the CPU." They do if you need a new feature of the new CPU, if you need faster memory to gain a lot of the benefits of the new CPU, if you need a faster interconnect, if you need more ports with the new CPU, etc. Even if you didn't need any of those a slight chipset upgrade can also come with a speed boost. Up to 5% speed difference here: https://www.techspot.com/review/1872-ryzen-9-on-ol...
rarson - Tuesday, May 19, 2020 - link
"Because you use them in pairs, and you can upgrade one without upgrading the other."I can also use one without using the other, and I don't ever buy "upgraded" appliances just to wash or dry clothes faster.
"They do if you need a new feature of the new CPU, if you need faster memory to gain a lot of the benefits of the new CPU, if you need a faster interconnect, if you need more ports with the new CPU, etc."
All of those things preclude using a newer chip in an older motherboard anyway, with the exception of a feature like PCIe 4.0. That is, if a new CPU has a DDR5 controller on it, physical DDR4 slots obviously aren't going to work. So you're not "gimping" the CPU because the CPU doesn't work at all.
In this case, the ONLY "gimping" you're getting is PCIe 4 which is essentially useless at this point, especially since PCIe 3 isn't any sort of bottleneck, especially for the CPU. Regarding the article you linked to, did you even read the title, or any of the rest of it for that matter? The differences are clearly chalked up to VRMs, power consumption, and firmware. Or did you not realize that a CPU can be slightly faster when it is allowed to consume more power?
My main problem with X570 is the heat of the chipset. The damn thing runs so hot that it needs a fan on it. I run far, far away from those tiny little fans because they suck in dust and seize up with the slightest amount of build-up. I don't want my motherboard dying because the tiny-ass chipset fan stopped working. For that reason alone, I'm more than happy to forego PCIe 4 for a board that doesn't require a chipset fan.
rarson - Tuesday, May 19, 2020 - link
Changing a motherboard is a lot more of a pain in the ass than changing a CPU. Same goes for shipping it.I would never go to the trouble of selling a board when I can just continue using the one I already have and just upgrade the CPU.
Spunjji - Wednesday, May 20, 2020 - link
Changing board requires a Windows reactivation or reinstallation, while changing CPU doesn't. It's also a lot less work to remove a CPU cooler and swap a chip than it is to remove nearly everything from your case to get the motherboard out, and you stand a lot less chance of accidentally killing a major component with ESD or an errant screwdriver.This is before you get anywhere near the e-waste implications.
deksman2 - Tuesday, May 19, 2020 - link
On a desktop (or even a laptop - especially a laptop), I would imagine plenty... especially with people who are looking to save on upgrade costs and minimize the upgrade to just 1 or two components.Some might not even be aware of the possibility they could save a lot of money on just updating the motherboard BIOS which would give them the option to upgrade to a far better CPU which would improve system performance.
Swapping out the motherboard on a desktop adds extra time and cost for the users.
In a laptop, its impossible to swap out the motherboard because laptops use custom-made mobo's for specific chassis (but then again, most OEM don't even update mobo BIOS-es on laptops which makes matters worse... although in the past they used to - which made it possible for a person to swap out a Core2Duo for say Core2Quad - if the cooling had the capacity to handle the newer CPU).
schujj07 - Tuesday, May 19, 2020 - link
Most laptop CPUs are BGA so you cannot change the CPU anyways.rarson - Tuesday, May 19, 2020 - link
"Most laptop CPUs are BGA so you cannot change the CPU anyways."You can, but it's a LOT more difficult.
I don't think I've seen a socketed laptop motherboard since Sandy Bridge. I've looked a lot, but that's about when most manufacturers started transitioning all of their laptops to thinner designs and BGA chips.
"most OEM don't even update mobo BIOS-es on laptops which makes matters worse"
This isn't true, as HP laptops will often update the BIOS on their own, without warning. Windows can also update the UEFI through Windows Updates. But they don't do BIOS updates for support, they do them for bug fixes, as the support is already "baked in" by the hardware choices of the OEMs and doesn't change.
kepstin - Tuesday, May 19, 2020 - link
I personally went from an R7 1700 to an R7 3700X on my MSI B350 Tomahawk board. This means that I had to use one of these "permanently beta" bios versions that drops older CPU support (I think A-series chips?) and it removed some of the fancy GUI bios config features like graphical fan curves.Worth it to me personally, since the perf jump from zen 1 to 2 was decent, along with the better memory support. resulting system is a bit more stable.
Orange_Swan - Tuesday, May 19, 2020 - link
When I upgrade my CPU I always get a new motherboard. To be fair I do upgrade every 4 years.ET - Tuesday, May 19, 2020 - link
Quite a few do. There are people running Ryzen 3000 CPUs on X370, for example. The upgrade path from Ryzen 1000 or 2000 to 3000 is pretty compelling - much better IPC and more cores without having to buy a new MB, that's quite convenient.(If Intel had allowed newer CPUs to be installed on the B250 -- which was technically possible -- I'd have made such an upgrade.)
AntonErtl - Tuesday, May 19, 2020 - link
I have in the past upgraded from K6-2 300 to K6-2 500 and from Athlon 800 to Athlon 1200; someone interested in more cores may easily want to upgrade from a 2700X to a 4950X or so.Another use case is replacement after failure: when my LGA 775 board died, I needed another LGA 775 board (and despite Intel, one could still get such things). When my Core i7-6700K (tray) died, I bought a Core i5-6600K to fit in the board; if that happened today, I would curse Intel for not supporting the 9600K on the same board.
silverblue - Tuesday, May 19, 2020 - link
Socket A lived a long time, but certainly evolved - in three years, users went from AGP 2x to 4x to 8x, FSB went from 100MHz up to 200MHz in 33MHz steps, DDR replaced SDR, dual channel became a thing (that wasn't worth bothering with at that time on AMD), and there were so many chipsets, many of which were problematic (here's looking at you, VIA). I went through four in that time - AMD 750, SiS735 (on the infamous ECS K7S5A which slowly eroded its ability to run at 133MHz FSB, until one day it wouldn't even run at 100MHz), KT266A (ridiculous how much extra performance it offered over even the 735) and nForce2 400. All this time, it was still one generation of socket - even if people did need to buy new boards for Zen 3, this is still an order of magnitude less hassle than the upgrade path not even 20 years ago, though to be fair, CPU speeds more than doubled in a three year period.rarson - Tuesday, May 19, 2020 - link
It seems like every socket I've ever used that VIA made chipsets for had at least one very problematic VIA variant and one really good one. I kind of miss those days because there was more variety in features and performance. You really had to do your research before buying, though.silverblue - Wednesday, May 20, 2020 - link
Exactly, VIA would release an A variant to fix the issues (or add performance) in some cases. The KT400 didn't even officially support a 400MHz FSB (the clue certainly wasn't in the name) until the A variant, and the KT266A was a huge jump over the KT266. Luckily for VIA, pin compatibility meant it was easier for motherboard manufacturers. I didn't mind the research when you had sites like this providing you with the ins and outs, certainly helped back then where it seems there were double the number of manufacturers (as well as numerous chipsets).mrvco - Tuesday, May 19, 2020 - link
I certainly would if I could and it makes sense on a $ / perf basis. I have a PC now with a i5-6600K that could be upgraded to a i7-7700K, but spending <=$379 on the CPU alone doesn't make a bit of sense versus spending the same or even less money to buy a NIB Ryzen CPU and mobo. Add in the option to upgrade the CPU on that board in a year or two and it's all gravy.rarson - Tuesday, May 19, 2020 - link
This is exactly why I tend to go with AMD. I pieced together an HTPC using parts that were laying around the shop. It's got a Z270 m-ITX board with a 1080 that is significantly limited by an i5-6400T. It was the best CPU that I had laying around, and not only can I not overclock it but it's so SLOW with a max of 2.8 GHz. The 7700K is the best I can put into this thing, but even used prices for those are absolutely ridiculous. I have a motherboard that is completely useless because there are zero upgrade options that make any sense.I'm probably just going to replace it with a B550 board and 3300X. That'll give me about the same performance in games as a 7700K for less money, while allowing me to upgrade to Zen 3 in the future.
rarson - Tuesday, May 19, 2020 - link
That's literally how I determine my builds. I buy a good motherboard, one that has all of the features I want and can support plenty of upgrade options, then I buy the best bang-for-the-buck CPU for it, with the intentions of upgrading to the best processor a couple years later when their prices have plummeted.I stuck with Core 2 for 10 years. My initial CPU purchase was the $65 Pentium Dual-Core which overclocked from 1.8 to 3.0 GHz stupid easy on the stock cooler. Then a few years later, upgraded to a Q9550 for less than half its original MSRP. Kinda the same story with the graphics card. There was no need for me to replace the board because it was a great board. With a SATA SSD, an overclocked C2Q, and a good graphics card, it ran better than a lot of newer machines. It even ran Doom 2016 pretty well.
FreckledTrout - Tuesday, May 19, 2020 - link
If my x370 board gets Ryzen 4000 support I would upgrade. I have a higher end board with 32GB of Samsung B.die so Im all set. Would love to get a 30%+ performance boost over my 1800x.Lord of the Bored - Tuesday, May 19, 2020 - link
It used to be quite commonplace. Hell, they used to make special versions of new processors to keep old boards running(486-class parts that fit 386 sockets, Pentium-class parts that fit 486 sockets, an absolute metric buttload of P2-class parts that fit socket 7)Intel worked hard to stop it from being an option.
Khenglish - Tuesday, May 19, 2020 - link
What if I install a larger bios chip for a B450 system? Can I then flash an image with the full range of CPU support? I have a programmer.I may be in the minority of users here. I'm not sure though.
XabanakFanatik - Tuesday, May 19, 2020 - link
The article says the CPU can only address the first 16MB of a chip larger than 16MB.SNESChalmers - Tuesday, May 19, 2020 - link
That's my understanding too. Maybe chips that can address 32MB ROMs can use the "back" half? It's conceivable that MSI's MAX series boards could support every single AM4 CPU, but I doubt it is worth their time and effortgouthamravee - Tuesday, May 19, 2020 - link
From the article it seems like only 'early AM4' processors had the 16MB limitation. Hard to tell if that's only the Ryzen 1000 series or others.Cooe - Tuesday, May 19, 2020 - link
This is only true for Zen/Zen+. Zen 2 can address the full 32MB.shing3232 - Tuesday, May 19, 2020 - link
MSI MORTAR B450 MAX does come with a 32MB bios, so you can by modified bios to get support to all cpu. it happen during AM3 AM3+ era.Achaios - Tuesday, May 19, 2020 - link
Chipzilla must rly be hurtin' today. AMD is dealin' them blow after blow.I admit I never expected Lisa Su's AMD to amount to anything much after #FAILDOZER.Looks like Lisa Su must have done a lot of housecleaning.
rarson - Tuesday, May 19, 2020 - link
It's like an entirely different company, aside from the continued marketing blunders. I think the plan was something like "fix the CPUs, fix the GPUs, then fix the marketing team." Hopefully they'll eventually get there.eastcoast_pete - Tuesday, May 19, 2020 - link
Question: What is limited/left on the table by having this one socket (A) from Excavator to Ryzen 4000 and now these BIOS limitations?To be clear, I support and commend AMD's decision to not pull an Intel move here (new chips, new sockets, sucks to be you, go pay for new MBs), but doesn't sticking with an old socket and limited BIOS also come with significant downsides?
Cooe - Tuesday, May 19, 2020 - link
Yes. Go watch GamersNexus' most recent video on this whole situation. AMD was screwed & gonna piss somebody off either way they went. This was a "lose-lose" situation all around.xenol - Tuesday, May 19, 2020 - link
The 16MB BIOS ROM size limitation to me hearkens to what I feel is something that's holding modern x86 processors back: it has to start up in Real Mode, for no reason other than to maintain backwards compatibility.We can ridicule Intel all we want for its lack of delivering something other than 14nm, but can we focus on getting rid of the baggage that is BIOS and Real Mode? Or at the very least, if you really want it, you can have it but don't make it the default anymore.
allenb - Tuesday, May 19, 2020 - link
Maybe I am misreading the chart, but it appears to show that Ryzen 3000 CPUs are not supported on A320. I’ve got a secondary machine at home running a 3700X on an ASRock A320M-HDV R4.0. Works perfectly, no issues whatsoever. Very happy with AMD.Cooe - Tuesday, May 19, 2020 - link
It isn't officially supported lol. OEMs just said "we don't care what you say on this one" and updated most A320 boards anyways as they already had the necessary AGESA code from the other 300 series boards.rarson - Tuesday, May 19, 2020 - link
That was the whole point of UEFI, to get around the limitations of BIOS. They're technically two different things, most people generally still refer to both as BIOS.I don't think this has anything to do with Real Mode since Real Mode only has 1MB of addressable space. This is more to do with fancy graphics, large CPU support lists, and early Zen processor limitations.
ahenriquedsj - Tuesday, May 19, 2020 - link
Nice.balamacab - Tuesday, May 19, 2020 - link
To be fair I celebrate that AMD keeps it promise, but as a consent consumer I can no be buying a pc every other year.I try to make my PC and laptop last as log as they can (10+ years), so a strong processors and a strong board is critical for me (with slots and technology for today and tomorrow devices).
This year with all new technology in the pass years justify the upgrade for my PC (USB 3.2, DDR4, PCIx 4.0, M.2 drives) and went from a Core i5 750 and some mombo I dont recall to a Ryzen 9 3900x and a x570 mombo.
The next upgrade will be GPU a couple years on when the current GTX1060 begins to really lack (in one or two year).
CPU upgrade path is nice to have, but there is no way to support a platform for a long time, i'll consider to upgrade my CPU if eventually AMD came with a CPU using 32 cores over the same AM4 x570 board (which is unlikely in the short term).
rarson - Tuesday, May 19, 2020 - link
AMD would have to massively shrink their CPU dies in order to fix additional or higher core count chiplets on the existing AM4 package. I doubt even 5nm would be enough to accomplish such a shrink.I start with a good board and buy the best bang-for-the-buck with the intent of upgrading to the best CPU possible in a couple years from the used market. I can invest in a platform without the huge upfront cost of a high-dollar CPU, while still reaping the benefits of maximizing the hardware platform later on. For a minor upgrade cost, I get a nice bump in CPU performance that, along with storage and GPU upgrades, keeps the computer useful for several more years. I'm basically getting about 10 years out of each computer at this point.
Duncan Macdonald - Tuesday, May 19, 2020 - link
If the BIOS was only 1MB then there would be a good reason to limit support - with 16MB there is plenty of room. It might be necessary to make the BIOS a non-graphical text mode one (like most 486 motherboards) but there would be plenty of room for the essentials. (A mouse driven graphical interface and fancy logos are not requirements.)Remember a complete Windows 3.1 system could run on a system with 8MB of storage - this is only a BIOS.
rarson - Tuesday, May 19, 2020 - link
I wish they would do this, every computer I've built up to and including my Core 2 had the simple text interface, and I loved it. Using a mouse in BIOS drives me nuts, as do the stupid fancy graphics which obfuscate menu options and makes things more difficult to navigate. It adds nothing but frustration to the experience.WaltC - Tuesday, May 19, 2020 - link
Just do the smart thing and buy x570--or B550. Skip all of the hassle this will entail. Support for B450/X470 will come a good amount of time after the x570, B550 Zen 3 CPUS are launched and shipped, and bios support for Z3 AGESAs will always lag behind such support, just as they have with Zen 2. But if AMD goes the forked bios route they are talking about--how does this not entail buying a new x470/B450 motherboard? What about PCIx4 mode support? If it does, then I cannot see a reason to want to do this at all. Just buy an x570/B550 mboard, instead.ishould - Tuesday, May 19, 2020 - link
Is there a technical reason they can only read 16MB BIOSes? If not I'd like to see them support a much larger BIOS for the AM5 package so they can really deliver on the idea (I know they didn't *promise* support for all future Zen generations, but one could reasonably assume all AM4 platforms would work with any AM4 chip) of being able to buy a motherboard for Zen 4 and have it work with Zen 5, 6, and 7. I assume they still have time as Zen 4/AM5 won't be out before late 2021 or early 2022The_Assimilator - Tuesday, May 19, 2020 - link
> Is there a technical reason they can only read 16MB BIOSes?Yes. Read the fucking article.
gruenebohne - Tuesday, May 19, 2020 - link
In the table it says that the A320 chipset doesn't support Ryzen 3000 series CPUs. This is at the very list not generally true. I have an ASUS PRIME A320I-K motherboard, running an AMD Ryzen 5 3600. The CPU is officially supported: https://www.asus.com/Motherboards/PRIME-A320I-K/He...rich42 - Tuesday, May 19, 2020 - link
same here, as my A320-HDV from AsRock apparently even supports the Ryzen 9 3900 (only 65W CPUs)rich42 - Tuesday, May 19, 2020 - link
i meant the A320M-HDV board, of course, and the max. TDP allowance for CPU's on this motherboard. Sadly, the Ryzen 9 3900 is quite more expensive atm than the 3900X-modelshing3232 - Wednesday, May 20, 2020 - link
3900X run at 65W mode perhaps?rich42 - Wednesday, May 20, 2020 - link
it don't know if this is possible ... the CPU support list does not mention any support for bigger CPUhttps://www.asrock.com/MB/AMD/A320M-HDV/index.asp#...
rarson - Tuesday, May 19, 2020 - link
Zen+ was not originally officially supported by first-generation Ryzen chipsets (A320/B350/X370). Instead, AMD said that support should be possible, but left it up to the motherboard manufacturers themselves. I believe AMD was being overly cautious due to increased TDPs and the concern that the VRMs on existing boards might not be good enough to handle the power draw. The 1700X was originally a 95W part but the 2700X bumped that up to 105W.As far as I knew, Zen+ support was still unofficially supported, but the same thing happened again when Zen 2 started shipping, with support on first-gen chipsets being left up to the motherboard manufacturers. There's very little difference between X370 and X470, so other than concerns about VRMs, which seem largely unnecessary, I have no idea why this was assumed to be such a problem. Zen 2 support on X470 was no issue.
That said, this is the reason why Ryzen 3000 works on A320 boards, because the motherboard manufacturers enabled support for them.
GreenReaper - Wednesday, May 20, 2020 - link
Maybe they already knew there were ROM problems on lower-end boards and didn't want to store up more problems for themselves when the next set of CPUs came out.quadibloc - Tuesday, May 19, 2020 - link
I'm glad AMD is allowing the new chips to be used, but really now, this verified chip nonsense? Sure, downgrading to earlier BIOS versions may be a potential issue. But two side-by-side BIOS versions, with the same date, one for the new chips, one for all the old chips, with free switching between them would avoid the issue of any motherboard not being able to support any chip. Why gratuitously make problems for people?shing3232 - Wednesday, May 20, 2020 - link
because it would make brick your motherboard easier than otherwise. I think this is up to the manufacturerMDD1963 - Tuesday, May 19, 2020 - link
Good news for current B450 owners, as I doubt many will lose much sleep over loss of support for 1000-series CPUs, etc...watzupken - Tuesday, May 19, 2020 - link
Kudos to AMD for listening and making an effort to make this backward compatibility possible. I think this is where Intel NEEDS to pay attention.davide445 - Wednesday, May 20, 2020 - link
If I good remember the original AMD statement was that AM4 platform will support first tree generation of processors.So depending if Zen+ can be considered a generation the support might end with Ryzen 3000 processors. If not I expect support for Ryzen 4000 for all but not the most cheap and BIOS constrained motherboards.
I have a Asus B450 Plus mobo with a Ryzen 5 2600X, I will have anyway the option to upgrade up to a Ryzen 9 3900X that's more than enough.
BikeDude - Wednesday, May 20, 2020 - link
If AMD had simply launched a AM5 socket along with Zen3, would there have been as many complaints then?Anyone here care to guess what AMD will do for Zen4..?
(that said, I too would welcome a return of textmode bios configuration menus)
headloser - Wednesday, May 20, 2020 - link
I must give AMD credit for backtracking and saying we will try our best to make it happen. You going to lose the fancy graphic or customized option but you have to start somewhere. s most motherboards have 16 MB, and the CPUs can only address the first 16 MB of a BIOS chip.I think it time to upgrade the 16 MB to maybe 32 with room to spare for down the road.
29a - Wednesday, May 20, 2020 - link
I don't understand why having another forked bios is an issue. I have a B450M Pro 4 and it already has two different sets of BIOSs, one for 1000 and 2000 series and one for 3000 series.ajlueke - Wednesday, May 20, 2020 - link
I would like to point out that the chart is incorrect. Ryzen 3000 CPUs are listed as beta for B350/X370. I own the ASRock X370 Fatal1ty Professional Gamer motherboard and Ryzen 3000 was officially supported with UEFI edition 6.20, nothing beta about it.ajlueke - Wednesday, May 20, 2020 - link
"AMD reiterated to AnandTech that after the launch of B550 into the market, they do recommend the B550 motherboards as the best option for Ryzen 4000 support in upgrades."Better than X570? Strange.
AndroidBR - Wednesday, May 20, 2020 - link
I'm relieved, excited, confused and a little apprehensive all at the same time regarding these two official announcements made by AMD in only two weeks. The well deserved community backlash and now this travesty of requiring purchase / ownership confirmation of both 400 series motherboard and Ryzen 4000 CPU seems too controlling in my opinion. It goes against the simple concept of buying an unlocked hardware and being forced to lock it up to have support for a future CPU. I don't really understand the reasoning behind AMD stipulations here. Just give the Zen 3 microcode to the board partners and let them post updated BIOSes online like they've been doing for decades. Clueless users will brick their boards and increase RMAs? Well, tough sh*t. Read and follow instructions and make sure you ask all questions about the flashing process BEFORE flashing anything in your system. When in doubt, call your boards' tech support number and have them walk you through the flash process. It's not rocket science.Spunjji - Friday, May 22, 2020 - link
It's easy to say "tough sh*t" when you're not the one on the hook for the support bills.Remember, AMD only just got done losing the dumbest lawsuit ever against people who decided they could pull a fast one by arguing over how AMD defined a "core" in Bulldozer. They know the true depths of ignorance some customers are capable of plumbing.
lmcd - Friday, December 4, 2020 - link
In Bulldozer it was actually not reasonable to define their core as a core. It was reasonable to define it as a core imo in Piledriver onward. Certainly wasn't a dumb lawsuit.prophet001 - Thursday, May 21, 2020 - link
Hamstringing your platform for a socket is dumb.shirleymarquez - Thursday, May 21, 2020 - link
They won't have to worry about processors past Zen 3 in AM4 in any case. Zen 4 will almost certainly require DDR5 memory, and should come with a redesigned socket so you can't accidentally put the new processors into the old sockets or vice versa. But please, AMD, keep the cooler mounting arrangement the same!alexpaul171292 - Friday, May 29, 2020 - link
Nice and well informative blog. QuickBooks is provided with the QuickBooks Connection Diagnostic Tool to resolve the issues of network and various other issues occurring in it.To know more visit https://accountingerrorsolution.com/learn-how-to-u...janolsen - Wednesday, June 17, 2020 - link
The 4200G or similar APU which is based on the previous year's Ryzen (Zen 2) might work with A320 ?