Final Words
If AMD's Socket-AM2 only offers a minimal performance increase, then why on Earth is AMD moving to it?
AMD has done a tremendous job of making DDR-400 last with their architecture. When Intel first talked about moving to DDR2 there was concern that AMD's delayed move to the new memory technology would result in it being behind the curve, but the absolute opposite held true; Intel showed no benefit from DDR2 initially and AMD did just fine with only DDR-400.
However times are changing, and after a very long hiatus Intel will soon resume increases in FSB frequency, not to mention that their new Core architecture is considerably more data hungry than anything we've seen to date. So on the Intel side of the fence, the greater bandwidth offered by DDR2 will finally have a real use. With Intel DDR2 demand increasing and more manufacturing shifting away from DDR, it now makes sense for AMD to jump on the DDR2 bandwagon as well. If AMD does it early enough, the transition to DDR2 will be complete before any of its products desperately need it, which is always a better route.
It's not the most convincing reason to switch to DDR2 today, but AMD has stayed on DDR1 far longer than anyone expected and it's better to be early than never. The fact of the matter is that CPUs will get more cores, reach higher clock speeds and feature more data-hungry architectural changes, all of which require more memory bandwidth. AMD's options are to either add more memory bus pins to the already staggering 939-pin package, or to embrace a higher bandwidth (and lower voltage) memory standard; the option it chose makes a lot of sense.
There's also this issue of efficiency; based on our ScienceMark results, AMD was able to build an extremely efficient DDR-400 memory controller into their processors. The Rev E processors are able to deliver over 5GB/s of memory bandwidth, which is extremely close to the 6.4GB/s theoretical maximum offered by a 128-bit DDR-400 memory interface. The Rev F AM2 processors we've tested aren't able to break 7GB/s yet, which albeit an increase of 35% over the best Socket-939 numbers we've seen, still ends up being only 53% of the peak bandwidth offered by a 128-bit DDR2-800 memory controller compared to the almost 80% we saw on the Rev E.
If we use history as our predictor of the future, it may take a few more revisions of AM2 before we see that sort of efficiency, if we ever do. AMD has come a very long way since the performance we saw back in January, and if that's any indication we may just end up seeing better performance out of Rev G and H processors in the future. The verdict is also not out on Rev F; although the launch is only two months away, we keep on hearing that availability won't be until July. While that's not enough time for AMD to be making major changes to the silicon, it is quite possible that the changes have already been made and they're just waiting to get new chips back from the fab.
Based on what we saw with the Rev E cores and DDR-500, coupled with our results here with DDR2-800, it looks like Socket-AM2 will offer minor performance gains across the board if paired with very low latency DDR2-800, but otherwise it looks like it'll offer performance as good as Socket-939. If you're looking for numbers, with DDR2-800 at 3-3-3 we'd expect to see 2 - 7% gains across the board, with the 7% figure being reserved for applications like Quake 4 or DivX and the 2% figure being far more common.
Why would you move to Socket-AM2? If you're well invested in an up-to-date Socket-939 system, and if these numbers we've seen here today hold true for shipping AM2 platforms, then there's no reason to upgrade immediately. However, if you're buying or building a brand new system, then by all means AM2 makes a lot more sense than Socket-939. Like it or not, DDR2 is the future, and AM2 will be the new socket for AMD's future 65nm parts as well. DDR2 is also competitively priced with DDR memory while generally offering higher bandwidths, and with most manufacturers transitioning to DDR2 now we expect to see further DDR2 price cuts.
With AM2 you are investing in memory that will have a longer lifespan and a motherboard that will have a better upgrade path than Socket-939 today. The only other advantage other than a more secure upgrade path that AM2 offers is AMD's upcoming Energy Efficient desktop CPUs. We're particularly intrigued by the 35W Athlon 64 X2 3800+; if you thought AMD's processors were cool and quiet, a 35W X2 should blow you away. (It might overclock really nicely as well!)
The disheartening news for AMD and its fans alike is that if AM2 can't offer significant performance increases over what we have now, then all Intel has to do is execute Conroe on schedule, delivering the performance we've been promised and 2006 will be painted blue. AMD has been telling us that 2007 is the year we'll see major architectural changes to their processors, so AM2 may very well be as good as it gets for now. That's still very good, of course - the fastest X2 chips still outperform the fastest Pentium D chips - but it looks like after three years K8 may finally get some competition for the performance crown.
Facebook
Twitter
RSS
107 Comments
View All Comments
Shintai - Tuesday, April 11, 2006 - link
I love how that site only compared Conroes worst results, also the Conroe used crappy memory timings, DDR2-533, preproduction boards, ES sample CPU. Oh, and did we forget its a 300$ CPU vs a 1200$ AMD CPU? Ye we might have forgotten than...didn´t we.mino - Tuesday, April 11, 2006 - link
Most important it's NON-EXISTENT vs. $1200 CPU.Seems many people forget about it.
TrogdorJW - Tuesday, April 11, 2006 - link
"Facts"? That site is merely an ultra-pro-AMD/ultra-anti-Intel blog factory. It might as well be titled, "The Ravings of a Lunatic Fanboy!" I think just about every article ends up ripping on Intel in some fashion. He's actually still proclaiming the near-future death of Intel, I think. "Intel is dead, and the only reason they haven't realized it is because they don't have a brain." I think that was what he stated on one of the "articles".Isn't this the same Sharky that has been talking about the demise of Intel for years, like since the Athlon days? What will he say when Intel ships 2.66 GHz Conroe chips that easily outperform FX-60 chips in meaningful benchmarks at $300 or so, and then follow it up by having 3.0+ GHz parts? I can already guess what he'll say: Intel is just being weak by including more cache, making more chips, selling more processors, yada yada yada.
Look, Intel isn't great, and AMD isn't the promised Mesiah. Both are mega-corporations looking to make money. End of story. What has AMD really done new in the past three years? They've integrated the memory controller into K7, and they've pushed 64-bits onto us. I still haven't done any 64-bit computing, because for 99% of what I do, it's meaningless. Any benchmark that shows 64-bit computing as substantially faster (like Sharky illustrated on the Conroe article with ScienceMark) has just unveiled itself as a synthetic test that most people will never use. Tasks that can really benefit from 64-bit floats/integers can already be done more efficiently with MMX/SSE/SSE2/SSE3. The only thing 64-bits really adds is the ability to address more than 4GB of RAM, so until we're all running 4GB of RAM it isn't necessary.
AnandThenMan - Tuesday, April 11, 2006 - link
You have no idea what you are talking about. Zero. x64 is very, very important and valuable. Just because YOU don't see any benefit is meaningless. So I suppose Vista 64 bit edition is meaningless too. And I suppose x64 in the server space is meaningless, even though smart people are demanding it to the point where AMD can not make the Opteron fast enough. 64 bit applications are going to be EVERYTHING, especially in 4,8,16 and more processor/core settings. Unless for starters, you think 4 gigs of RAM is more than enough moving forward. :nuts: Meaningless indeed.
AMD has the interconnect infrastructure that will take them into multi-core, mult processor territory that Intel can only DREAM about. Couple that with x64 and Intel seems lost in the wilderness by comparison. What's next from Intel, a 64 meg shared cache?
Conroe is a short term solution, period. It may put Intel temporarily back in the thick of things. But it does not solve Intel's hopelessly outdated FSB mentality. It does not solve the fact that Intel has no idea how to add x64 to their mobile parts. It does not change the fact that AMD will be further extending x64, forcing Intel to play catch up again. It does not change the fact that Intel's x64 bit hack is actually slower than running in 32 bit mode.
Sorry, but we have all seen this movie before. Go back and read the Intel hype and press releases from the last 3 years. They are nearly a carbon copy of what is happening now.
jjunos - Tuesday, April 11, 2006 - link
lol fanboi alert.If you read his post, he was stating that general users will probably never see a tangible performance improvement from 64bit computing.
Tho that doesn't account for architectual changes to implement 64bits.
NullSubroutine - Monday, April 10, 2006 - link
reason for changing:1) lower power requirements for primarily laptops and servers
2) changing to ddr2 helps the industry not make two different main memory types
3) gives more bandwidth for quad cores across entire product line
4) released before back to school business (time related - gives them enough time to saturate the market with the technology)
Calin - Thursday, April 13, 2006 - link
Unified socket interface - this way, anyone could use the same mainboard with processors from the lowly Semprons (at some $50) to the ultrahigh FX line ($1000). This helps the small shops in the retail channel, as they won't need to stock so many mainboard types (socket related).fikimiki - Monday, April 10, 2006 - link
You forgetting one fact regarding AM2 platform. It is going to support four-core CPU.And using DDR2 AMD will be able to support four memory hungry cores (ups forbidden word).
The product launch and availability of X4/Opterons will be September I think to be faster than Intel for sure.
AMD is playing with us, because need best performance processors to stay on top of Intel offer. Why? The capacity is the answer. It is better to sell 1 CPU for 2000$ than 10 for 170$. Q2/Q3 AM2/939 is enough for first wave of Conroes. Then we will see X4 and other fun stuff like K8L to keep Intel and to extend the FAB3x capacity.
Shintai - Monday, April 10, 2006 - link
There is a few errors tho.Making 4cores on 90nm is suicide. It would be a massive sized chip, and we would talk of speeds of like 2.0-2.4ghz peak and a massive TDP. I guess a chip like that fastly would cost 2000$+
So don´t expect a 4 core chip from AMD before 65nm, so sometimes in 2007. The chance of having a 4core chip failing under production right now is massive. Also a 4core chip takes waffer space as said, so again, AMD wont ship many CPUs. So unless you want AMD to be a server vendor only..then no. 4 core AMD chips will start in the highend servers i presume, we talk Socket F here.
Intel comes with a 4 core (2 DC bolted) in Q1 2007 as a desktop. Later on they will make a "real" 4core chip. But this is the most wise and profitable way to start with, take 2 working dualcores and get a quadcore. Cheap and easy.
Sure one 2000$ CPu is better than 10 or 20 170$ chips. But they are just killing themselves that way and you wont see them in any desktop anymore then. But maybe thats Hectors plan, all new AMD developments look very serverside.
Also no 2006 roadmaps for AMD shows quadcores...
Calin - Thursday, April 13, 2006 - link
Even with the FSB interconnect (core to core) in Intel's current Dual Cores, their performance (in real applications) isn't degraded so much. That is, considering their performance in core-to-core data transfers is lower than half of AMD's current dual cores.If you can make single cores at 80% rate of success (and 20% failures), then dual cores on a single chip would take some 20% + (1-20%)x20%, or 36% bad, and quad cores would take almost 60% failures. Compare this to two bolted dual cores at 36% failures, and as a result, Intel can produce almost twice as many two chips quad cores than single chip quad cores. Is the performance hit significant? We don't know very well (for quad cores). For dual cores, the performance hit is existant but (right now) not so significant (based on scalability from two to four cores servers between Opteron and Xeons).