Thursday, April 13, 2006

Clovertown scores revealed

Clovertown (2GHZ, 65nm, 2x4MB L2, FB-DIMM) slower than Athlon 64 (2GHZ, socket 939, 512KB L2, DDR)

Intel showed off Clovertown quad-core server CPUs running on the Bensley platform with FB-DIMM memory at Spring IDF Taipei. Clovertown is basically two 65nm Conroe CPUs stacked together, with total of 8MB L2 cache. This page contained the benchmark scores for a 2P Clovertown. The reported clockspeed was 2GHZ (could be 2.13GHZ, same as Conroe E6400).



For single threaded test, the 2GHZ Clovertown got a Cinebench 9.5* score of 362. Daniel J. Casaletto, Intel Vice President, Digital Enterprise Group Director, Microprocessor Architecture and Planning, was running the demo. For 2P 8 cores, the score scaled to 1723, or 4.7x. Adding 7 cores led to 3.7x more performance. I think this is quite poor, you get only about half a core's worth when you add a core.

FSB (Front Side Bus) is an Intel bottleneck. Eight Conroe cores fighting for a 1066MHZ FSB is not a pretty sight: in 2P Clovertown each Conroe core gets only 133MHZ bandwidth or 1GB/s, not much better than a 80486 (Well, you may say there is the dual bus, so it should be 266MHZ, but 8 cores of cache coherence traffic must be considered). Fortunately, Cinebench doesn't put a lot of load on the bus, it spends most of the time digesting the data.

Let's pay more attention to this photo here, which shows the 2P Clovertown in action and is quite exciting. Look at the upper left corner, it reads Cinebench 64 Bit Edition. Finally, we can see Intel got 64 bit working, it's running the 64 bit version of Cinebench! My congratulations to the Israeli engineers for getting AMD64 figured out. Welcome to the exciting world of pervasive 64bit computing! Now, Microsoft will allow the world+dog to go 64 bit.

A reader kindly provided us the CINBENCH 64bit Edition result for a 2GHZ socket 939 setup. The spec: Athlon 64, Socket 939, 90nm, 2GHZ, 512KB L2. The score: 370. <--click to view screen capture.

On my old Athlon 64 2800+ (1.8GHZ, Socket 754, 130nm, 0.5 MB L2 cache), I got a 64 bit Cinebench 9.5 score of 294. My ClawHammer is a bit slower than Conroe CORE, but only a little. If you consider my CPU is only 1.8GHZ and only uses single channel DDR, and my old PC only has integrated S3 UniChrome graphics which eats some memory, it's quite good. I managed to overclock it to 1.9GHZ and got a score of 312. I expect the three year old ClawHammer to get a score 0f 294*2/1.8= 327 at 2GHZ, within 10% of the future Conroe. So I feel I made a good investment buying AMD* - I paid less than $100 for the CPU+MB.

Clock for clock, the performance of Intel CORE (Merom/Conroe) is very close to socket 939 Athlon 64 . Mooly Eden definitely over-exaggerated Conroe performance.

So far, Intel is trying to follow AMD's footsteps. Conroe is still a few months away, and AMD is a moving target. Dirk Meyer said AMD will soon release a higher capability product besides changing from DDR to DDR2. With the new process technologies jointly developed with IBM, AMD can boost clockspeed by an upward of 40%.

The Conroe performance analysis is here. I pointed out that when working set is larger than Conroe's unified cache (4MB), Conroe performs slower than Athlon64. Conroe only shines in simple and single threaded tests where the whole working set sits inside the 4MB unified cache. The Cinebench 9.5 needs over 150MB to run. Clovertown's 8MB cache has some positive benefit, as Cinebench works on the scene top down, only a slice of the scene is being worked on at any moment of time . But still, the effect of the 2x4MB cache is much smaller than the cases where the working set sits inside the cache all the time.

In other news, AMD subpoenas Microsoft in its anti-trust lawsuit against Intel. From the documents it's seeking, AMD is clearly suspecting that Intel hindered the development of Windows for AMD64.

* Cinebench is essentially a CPU/Memroy performance test, it can be downloaded from cinebench.com .

* I bought the CPU+MB (with S3 IGP) combo for $79 at Frys, nowadays, such deals are impossible to find.

*I am interested in seeing some Clovertown and Sempron socket 939 comparisons. If you have such a machine running Windows x64, please submit your results in the comments. Don't under estimate AMD desktop CPUs, check out this Athlon 64 and Xeon comparison.

There are several people keep posting 32 bit benchmarks for Athlon 64. Please note Intel was doing a 64 bit Cinebench 9.5 . That's why Intel got a high score of 362 at 2GHZ. As I can see from posts by users on the internet, a Conroe at 2.4GHZ gets about the same score. So, please read the benmark condition: 64 bit edition of Cinebench 9.5.

34 Comments:

Anonymous Anonymous said...

You could say Conroe's credibiilty is hanging by a thread

6:08 AM, April 14, 2006  
Anonymous Anonymous said...

very funny...

maybe bad drivers, they are far form shiping them.

sure many months to go, but i wont for a sec think think intel got a better cpu then amd coming down the lineup to i SEE it for my self!
intal rigged and plays dirty.

7:49 AM, April 14, 2006  
Anonymous Anonymous said...

it looks like Cinbench is 64-bit. does it mean that intel demonstrated conroe's 64-bit capability since Kentsfield is consisted of two conroe's core?

9:38 AM, April 14, 2006  
Blogger Sharikou, Ph. D. said...

This was fresh stuff, the benchmark was done by Dan Casaletton, Director o Intel's Digital Enterprise Group. We noticed two things, frist single thread performance of the CORE is slower than 2GHZ Opteron 246, secondly the scale up factor of 8 cores is only 4.7

10:16 AM, April 14, 2006  
Anonymous Anonymous said...

Will any of these "subpoenas" really help AMD against a mafia run company ?? I think AMD should put these subpoenas in the trash and begin to operate intel style ..for starters they can bribe Microsoft to exclude Intel procs out of their Vista for Desktop and Mobile space ..like the intel-skype deal ..that would be serious fun then !!

11:44 AM, April 14, 2006  
Blogger Sharikou, Ph. D. said...

I updated the score of my Socket 754. During the previous test, the PC was having a lot of stuff running (Yahoo, MSN, Skype), I stopped some of them and now get a score of 294. If I scale the score to a 2GHZ CPU, I would get 294 * 2/1.8 = 327. Within the striking distance of a 2GHZ Conroe.

This links to the screen capture of my bench.

4:29 PM, April 14, 2006  
Anonymous Anonymous said...

You didn't mention if your Cinebench 9.5 test was 32bit or 64bit. If 32bit then your not comparing apples to apples. Also you dont know that scaling to 2.0Ghz is linear. Why not overclock your Athlon to 2.0Ghz, then see what your score actually is?

5:00 PM, April 14, 2006  
Blogger Sharikou, Ph. D. said...

You can see from the screen capture (click the links), I am also running 64 bit Cinebench 9.5. I overclocked the CPU to 1.9GHZ and got a score of 312. Couldn't push it higher, the RAM is too cheap. I have another socket 939 PC, but there is only one piece of RAM inside. I will do a test with dual channel DDR after I get two pieces of RAM for the 939 box.

The beauty of the Cinebench is that it's basically a CPU/memory test. Other parts of the system don't involve much.

5:07 PM, April 14, 2006  
Anonymous Anonymous said...

I have 371 with venice 2ghz ddr500

5:37 AM, April 15, 2006  
Anonymous Anonymous said...

AMD Opteron 146 @ 3ghz
Cinebench 9.5 32bit

Cinebench 9.5 64bit: 502

6:45 AM, April 15, 2006  
Blogger Sharikou, Ph. D. said...

Please note Intel was using a 64 bit benchmark. Therefore, for apples to apples comparison, the Athlon 64 results must also be 64 bit under Windows x64.

Since my old socket 754 Athlon 64 (130nm, 1.9GHZ, 0.5MB L2, single channel DDR, no SSE3, rev CG) is within 10% of the Clovertown/Conroe. I am quite sure that a modern Socket 939 Sempron with dual channel DDR should be able to defeat the Clovertown(65nm, 8MB L2, FB-DIMM) at the same clockspeed.

Please capture your screen and upload to http://www.imagethrust.com

10:34 AM, April 15, 2006  
Blogger Sharikou, Ph. D. said...

I have 371 with venice 2ghz ddr500

Could you please provide a screen capture?

Thanks.

3:59 PM, April 15, 2006  
Anonymous Anonymous said...

venice 2000mhz ddr500 370 cinebench64bit

http://img122.imageshack.us/img122/482/cine6tm.gif

sciencemark64bit

http://img105.imageshack.us/img105/7158/science0rw.gif

7:16 PM, April 15, 2006  
Blogger Sharikou, Ph. D. said...

venice 2000mhz ddr500 370 cinebench64bit
Could you download the newer Cinebench 9.5 from www.cinebench.com and rerun the bench?

Thanks.

10:01 PM, April 15, 2006  
Anonymous Anonymous said...

cinebench64 9.5 venice 2000mhz ddr500
http://img117.imageshack.us/img117/7957/cine21wg.gif


cinebench64 9.5 venice 2700mhz ddr490
http://img70.imageshack.us/img70/7785/cine39pq.gif

6:01 AM, April 16, 2006  
Anonymous Anonymous said...

ran cinebench 9.5 64 bit edition on a Athlon64 X2 4200+ (2.2 GHz) with dual 7800 SLI

http://i3.tinypic.com/vpiya0.jpg

Max was 370 but average was around 363. (we ran it 5 times)

And that's 200 MHz faster then the Intel variant.

Scale the X2 down to 2.0 GHz and you'd get around 330ish

Core has a 8.8% performance advantage and that's WITH the bottlenecked FSB.

11:22 PM, April 16, 2006  
Blogger Sharikou, Ph. D. said...

1333MHZ FSB more than enough to run a single instance of Cinebench.

We should conclude that Conroe and Athlon 64 are within a few percent of each other at the same clockspeed. As we can see from independent ScienceMark and Cinebench 9.5 results, except the pathological case of MolDyn where Conroe led by 33%, which I concluded to be a result of the boundary effects of the 4MB cache.

This is definitely an improvement from Pentium 4.

On desktop, what we will have next then is a GHZ war. Who commands the clockspeed advantage will win the day.

With Stress Memorization on dual stress linear strained silicon on insulator technology, AMD is supposed to jack up clockspeed by 40% from current levels to 4GHZ.

10:46 AM, April 17, 2006  
Anonymous Anonymous said...

I am glad someone else noticed this less than stellar performance :) If you have a minute take a look at my site, particularly here :

http://renderfred.free.fr/cb95vs2003.html

Click on the small screenshot and you'll see the performance of a dual Opteron 285 running at 2.8 GHz on CineBench 95 (64 bit). So you can imagine what an 8-core Opteron will do next year...

11:15 PM, April 17, 2006  
Anonymous Anonymous said...

Yep, in the up coming opteron they have made quite a few changes to improve the memory controler to reduce the cache coherency traffic. Was done to make the 8 ways opterons improve things when running with 16 cores, though as they are they are pretty darn good. Will help the quad cores quite a bit I suspect as well.

12:05 AM, April 18, 2006  
Anonymous Anonymous said...

Geez! This site really seems paid by AMD to spread out false claims and rumors against Intel at any cost, uh ?
What are you talking about ?
Which benchmarks ?
Benchmarking what ? 4-core and 8-core Intel CPUs that don't even exist yet ?

What a childish attitude, indeed.

3:06 AM, April 18, 2006  
Anonymous Anonymous said...

Here is the benchmark from my machine. Duel AMD Opteron 275s in a stock HP xw9300 workstation running windows XP 64 Pro.

http://s10.photobucket.com/albums/a134/rivieracadman/?action=view¤t=bench.png

7:50 AM, April 18, 2006  
Anonymous Anonymous said...

Here's my rig! Athlon X2 4400...

Link

2:09 PM, April 18, 2006  
Anonymous Anonymous said...

I ran the 64-bit version on my desktop and this is what I got.

CINEBENCH 9.5
****************************************************

Tester : K.P.

Processor : Athlon 64 X2
4400+
Motherboard : Asus A8N SLI
Premium
MHz : 2.2 GHz
Number of CPUs : 2
Operating System : WIndows XP Pro 64-Bit

Graphics Card : nVidia 7800GT
Resolution : 1280x1024
Color Depth : 32-bit

****************************************************

Rendering (Single CPU): 367 CB-CPU
Rendering (Multiple CPU): 680 CB-CPU

Multiprocessor Speedup: 1.85

Shading (CINEMA 4D) : 357 CB-GFX
Shading (OpenGL Software Lighting) : 1420 CB-GFX
Shading (OpenGL Hardware Lighting) : 3531 CB-GFX

OpenGL Speedup: 9.89

****************************************************

The system also has 3GB of RAM.

6:26 AM, April 19, 2006  
Anonymous Anonymous said...

Just tested home server with X2-3800 on SuperMicro H8SSL-i and 2GB DDR400. Running Server 2003 x64, no overclocking.

32-bit 1 Core: 295
32-bit 2 Cores: 548
64-bit 1 Core: 329
64-bit 2 Cores: 609

9:41 AM, April 19, 2006  
Anonymous Anonymous said...

I ran Cinebench 64 and 32 on my 148 opteron, at 2.0, underclocked I got 339, in 64bit benchs at stock 2.2 I got 369 at my normal speed 2.92 I got 491
http://www.planetamd64.com/index.php?showtopic=21104&view=findpost&p=211748

1:29 PM, April 19, 2006  
Anonymous Anonymous said...

I just ran Cinebench 9.5 on WinXP64 using an x2 3800+ clocked to 2.4GHz.

Scores are:
402
756
1.88x

Not bad for a 300 buck processor you can buy today :-)

4:18 PM, April 20, 2006  
Anonymous Anonymous said...

1333 FBS = 10.6 gigs of bandwith actually :)

11:19 AM, May 01, 2006  
Anonymous Anonymous said...

Conroe 64 bit results @2978 MHz:

1 core: 608
2 cores: 1118

Screenshot


Overclocked from 2.13 GHz, stock voltage, aircooling.

4:41 PM, May 12, 2006  
Anonymous Anonymous said...

It doesn't seem like the core architecture gets a lot of performance boost of a higher FSB. Especially beyond a given level. See for yourself when Merom with high multi and lower FSB is challenged by Conroe with lower multi and higher FSB:

http://www.xtremesystems.org/forums/showthread.php?t=99466
as you see there isn't a big difference in Cinebench.

4:35 AM, May 14, 2006  
Anonymous Anonymous said...

In other words, Core isn't really bandwidth starved.

Also, its performance scales well with clock speed.

12:19 PM, May 16, 2006  
Anonymous Anonymous said...

Has anyone tried to run Cinebench on 64bit Vista (beta 2)?
Have to download it tonight and give it a try!
Let's see how my San Diego 3700+ and 3GB of DDR performs

2:57 AM, June 27, 2006  
Anonymous Anonymous said...

Ok here the results!

My system (no OC!):
A64 San Diego 3700+ 2,2Ghz, 3GB DDR400 (3,3,3,8, 2T), ATI X1800XL

Running it under Vista beta 2 in 64bit gave me a score of 307. The graphics benchmark crash under Vista.

http://img426.imageshack.us/img426/6205/cinebench64bitvistabeta28zg.jpg

Think this is not a bad result considering the beta stadium of the OS (running Aero Glass) and a single core CPU. Would really like to see benchmark results of Conroe under the future OS. XP 32bit was yesterday, 64 bit computing starts NOW!!

I noticed on the 2.9 Ghz oced Conroe screenshot that cinebench 2003 had been used. Is there a big difference to version 9.5? If it's like 3DMark then please post a screenshot running 9.5

12:49 PM, June 28, 2006  
Blogger ellan said...

Hi everyone, I am running a 64-bit Win Xp pro, and faced issue with software. There is very few applications that are trully native. So I've been following www.64xsoft.com and some other blogs and websites to search for the software. Do you have any other ideas of where I can find it? Thanks,

Regards,
Jotheph Nurtes.

3:24 PM, August 27, 2008  
Blogger kenwood said...

The most undesirable way to approach the Salesforce CRM tool would be to implement the CRM solution and waiting for success to come on its way. I Salesforce training in Hyderabad

11:30 AM, October 25, 2020  

Post a Comment

<< Home