AMD Phenom Preview: Barcelona Desktop Benchmarks
by Anand Lal Shimpi on September 10, 2007 12:03 AM EST- Posted in
- CPUs
The Methodology
We've been asking AMD for months now to let us benchmark Barcelona and Phenom, and for months we've gotten the same answer: not yet. When asked why, AMD would always give us some terrible lie about how it was for competitive reasons, but when we actually put our hands on Barcelona at Computex we realized that these chips were simply not ready.
The Barcelona launch is finally upon us and we've got to wait another 45 - 60 days before we'll be able to bring you a review of Phenom, well, not exactly. Back when the Opteron launched, AMD was in a very similar situation to the one it's in today; AMD needed K8 to remain competitive, and it had been delayed so much that we were beginning to wonder if AMD would ever get the chip out on time. When the K8 finally launched, it was server-only but we took one of those server-only motherboards and ran a bunch of desktop tests on it to predict forthcoming performance.
We cracked open the Barcelona server and made some modifications; while the on-board ATI ES1000 graphics is sufficient for use as a server, it'd be too limiting for our desktop benchmarks. Luckily the Supermicro motherboard in the system had a plethora of PCIe slots, we just needed to gain access to them.
The PCIe Riser we removed from the system
We pulled out the PCIe riser card which plugs into the motherboard's sole x16 slot and divides it into a pair of x8s, then we modified a GeForce 8800 GTX by removing the backplate cover so we could just stick it into the open server.
The modded 8800
The 8800 GTX installed, the server is not really intended to be used like this
The end result was, as Johan put it, us using "such a beautiful, noble machine for such plebian activities". We couldn't help it, while AMD has already contacted us about Phenom briefings, we couldn't wait that long to get an idea of what we can expect from AMD on the desktop.
We needed an external PSU to power the graphics card, the server didn't have any PCIe power connectors
Barcelona is currently limited to DDR2-667, we were unsuccessful with attempts to run the memory any faster. Like all other MP Opterons, Barcelona requires the use of registered DDR2 memory, which is inherently slower than the unbuffered stuff we use on desktops. Because of these limitations we refrained from running any comparative benchmarks to desktop Athlon 64 X2s, instead we chose to run a single quad-core Opteron in our server platform against a pair of dual-core Opterons to simulate Phenom vs. K8 on the desktop.
The Opteron server, 2 CPUs, 8 DDR2-667 DIMMs
Keep this in mind as you're looking at these results, at best all we're offering is an idea of, at a minimum, how much faster Phenom will be over an identically clocked Athlon 64 X2. As Phenom is a more data hungry CPU than its predecessor, it will rely more on having a faster memory subsystem so the performance improvement could be even greater when we measure it on the desktop. That being said, at least we can set expectations within some amount of reason by performing this investigation.
Facebook
Twitter
RSS
70 Comments
View All Comments
kevinkga - Friday, November 23, 2007 - link
Interesting review but it looks like you've given the system to gamers to test. These systems seem to be built for server-class systems, where performance for stuff like encoding, gaming etc is not relevant. Instead, why not:1. Have a gentoo install done and bechmark against that? CFLAGS tweaking and stuff would truely reveal where this processor thrives or fails.
2. Why not use a process intensive application or thread intensive application such as apache under loads of stress or a highly threaded application server such as websphere? In these cases latency on communication between processes or threads running on different cores is much more important than GHz performance. AMD uses hypertransport between its processor, one of the things that its got an edge comapared to intel. IPC or interthread communication for server-class benchmarks make much more sense that normal multimedia benchmarks that are used for desktops.... c'mon guys.
spinportal - Wednesday, September 19, 2007 - link
So what is the going talk on Phenom vs. Peryn?Phenom launch is when? Late 2007? bad news
Peryn launch is Nov 12th. Wolfdale is dual/Yorkfield is quad (without interconnects).
Phenom is called PhenomX4 for quad-core? and PhenomX2 for dual-core?
Will Phenom support multi-cpu motherboards or is that just Opt/Barcelona?
Will Peryn support multi-cpu mobos or is that just Xeon/Harpertown?
What we need is a metric for the 2.33 or 2.5 GHz releases for Phenom & Peryn .
What is the comparison of a Athlon X2 6000+ to a Phenom X2 2 GHz in performance and price?
Wolfdale vs. Conroe (175$US) @ 2.33 GHz ~ +5%
PhenomX2 vs. K8 X2 3800+ (65$US) @ 2.0 GHz ~ +15%
Conroe @ 2.66 Ghz (320$US) vs. K8 X2 6000+ (165$US) @ 3.0 Ghz ~ +18%
(+24% vs. X2 4400+)
Now for the pulling numbers out of a hat:
Wolfdale @ 2.33 GHz vs. PhenomX2 @ 2.33 GHz ~ +14%
Street Worth: 200$US vs. 100$US
But you know AMD will need to price higher to protect its AM2 line, and Intel will put its price premium in as to not deflate Conroe.
Est Retail: 240$US vs. 180$US
cocoviper - Tuesday, September 11, 2007 - link
In light of Barcelona's "close but no cigar" performance I really think AMD should use a strategy from their own playbook from Spring/Summer of 99. K7 was looking to be a pretty good performer, but it wasn't quite there yet so AMD respun the K6-2 and K6-III's as the K6-2+ and K6-III+ (0.18 micron versions). These things were drop in replacements that were ultra cheap (both to produce and to buy), ran cool, overclocked really well and brought needed cash into AMD until they were able to stabilize the K7 and move to the Thunderbird.I say continue pushing hard to get Barcelona north of 3GHz, but in the mean time, have some people respin a either a cut down K8 derivitive to 65nm or a standard K8 to 45nm (only if 45nm will be up and running before Feb to March) and push it out as quick as possible. Then they can make some cash in the mainstream market from alot of inexpensive CPUs.
coolme - Friday, September 14, 2007 - link
Good idea, AMD can actually be the first cpu manufacturer to sell new x86 processors in the < $100 range. With new shrinked K8s and cheap cheap single-core Barcelona. (less cache, no power improvements, and possibly off-die memory controller to aviod low yields, but cheap cheap cheap)Visual - Wednesday, September 12, 2007 - link
even a simple return of the 939 x2s would give them quite a lot of sales, i think.i personally would not hesitate to buy even a 6200+ x2 if they made it available and as cheap as the am2 versions now.
also that will in no way replace my future purchase of a quadcore k10 (or two, in a 4x4 board) so they shouldn't worry about their products competing among themselves.
i know plenty other people who refused to buy new mobos just for the sake of ddr2, and are still sticking with single-core 939s. give them the $40 3600+ x2s and 99% of them will upgrade. some will likely even go for the higher SKUs, like me.
i know there still are occasional 939 x2s that you can buy here and there, as well as the 939 dualcore opterons. but believe me - supply is really low, making prices unreasonable compared to am2 chips, and also while they may be possible to find in the usa, uk or central europe, there are none at all available in my own country in eastern europe. i really want to find one, been looking for a long time now, and the only option i got is to import, further driving the price up - its just not worth it compared to am2 as it is.
13Gigatons - Tuesday, September 11, 2007 - link
That is very disappointing....duploxxx - Tuesday, September 11, 2007 - link
Sry to say anand, but i would advise to remove the review. If people would have a decent view on how this benchmark was done they would understand that this will be nowhere near the performance of the real phenom but it is a good indication. (and that was the whole intention why you made this review) However most only look at the benchmarks and think it will never be any faster than this.To my opinion you are killing a product before it is even launched and a bad name for you're site.
BaronMatrix - Monday, September 10, 2007 - link
Maybe someone should try putting these in a QFX mobo (well THE QFX mobo) and see what happens. They should OC pretty good and that would also raise the RAM clock and take away the ECC penalty.eRacer - Monday, September 10, 2007 - link
Were any tests done to compare a single dual-core K8 Opteron versus a single quad-core K10? In previous reviews single- and dual-threaded benchmark performance was actually hurt by running dual A64 FX CPUs instead of a single A64 CPU at the same frequency.For example, in this review K10 was 16.2% faster than two dual-core Opterons at the same frequency in Oblivion. However, a previous Anandtech review showed that a single 3GHz A64 X2 6000+ was 17.1% faster than two 3GHz A64 FX-74 CPUs in a QuadFX system.
http://www.anandtech.com/cpuchipsets/intel/showdoc...">http://www.anandtech.com/cpuchipsets/intel/showdoc...
So is it possible the 2x K8 vs. 1x K10 comparison is telling us as much about the shortcomings of using dual-socket AMD systems in desktop benchmarks as it is about the strength of K10 vs. K8?
MadBoris - Tuesday, September 11, 2007 - link
See my post above showing some K10 to K8 comparison with single and dual thread from techreport, the K10 advantage seems pretty insignificant on single threaded or more serial based applications.
This will not look so good for the consumer desktop sector with so many applications/games that are still serial in their nature(even if they are multi-threaded).