Tuesday, October 24, 2006

AMD maintains massive performance lead

Integer performance, AMD leads by 22% to 100%.

AMD score: 175, or 279.

Intel score: 143

Floating point performance, AMD leads by 61%

AMD: 174, or 182

Intel: 108

DELL claims that AMD server consumes 20% less power than one with Intel's top end Xeon 7140 (16MB L3).

Intel only plays in the ultra low end with Woodcrest.

20 Comments:

Anonymous Graham said...

Who cares? Intel maintains a massive price/performance lead over AMD.

Link

6:54 PM, October 24, 2006  
Anonymous Anonymous said...

Xeon 7130M is a netburst based processor...of course amd will be faster than it....

also these tests were done on 4P setups and we already know AMD is faster at 4P and higher.

6:56 PM, October 24, 2006  
Anonymous Anonymous said...

Aren't those netwurst CPU's? Is that tulsa 4P or something?

7:29 PM, October 24, 2006  
Anonymous Anonymous said...

Am I reading the RAM spead correctly for the AMD DDR25300 and for the Intel DDR23200?

Wonder if that makes a difference?

7:37 PM, October 24, 2006  
Anonymous Anonymous said...

Cool, it means Woodcrest would be 20% less power than then AMD solution since Intel C2D is 40% lower power than the previous solution.

Now i know why the comparison was done against Tulsa, not woodcrest.

8:14 PM, October 24, 2006  
Anonymous Anonymous said...

Hey something smells fishy. The AMD rigs have 32G of memory while the Intel ones have 16G. Oh and by the way these are netburst chips which if I remember correctly were suppose to be like 40% slower than the K8s. Hmmm kind of interesting how Intel was still able to take back server market share away from Amd even with these dogs huh...LOL. Amd better get back in the game LIKE FAST.

8:15 PM, October 24, 2006  
Blogger core2dude said...


Intel only plays in the ultra low end.


Too bad that Intel totally dominates that ultra low end, and that ultra low end comprises 90% of server market.

Only if the MP mattered...

8:26 PM, October 24, 2006  
Blogger Greg said...

This really isn't an accurate look on a per processor basis. In 2p Intel would regain FP crown, since they wouldn't been so ridiculously constrained on their FSB. Their new Tigerton System will basically be the same thing as this though, since they're increasing the number of cores by the same factor they're increasing their fsb with the 2 extra connections.

8:29 PM, October 24, 2006  
Anonymous Anonymous said...

Actually, if you want to use spec, try using similar machines, ie HP vs HP same OS and such. AMD does do nicely on the FP as excpected but Intel has a slight edge in Integer for the MP space. I think most will agree Opterons are very good in the MP space.

In the 2P space its a different story. Intel has the lead in both Integer and FP.

MP

Integer Intel 163 AMD 160

http://www.spec.org/cpu2000/results/res2006q3/cpu2000-20060818-07059.html

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20060918-07377.html





FP Intel 105 AMD 140

http://www.spec.org/cpu2000/results/res2006q3/cpu2000-20060818-07047.html

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20060918-07373.html



Now in the 2P space

FP Intel 81.0 AMD 73.5

http://www.spec.org/cpu2000/results/res2006q3/cpu2000-20060626-06346.html

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20060918-07379.html


INT Intel 120 AMD 82.6

http://www.spec.org/cpu2000/results/res2006q2/cpu2000-20060515-05998.html

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20060918-07375.html

8:59 PM, October 24, 2006  
Anonymous Anonymous said...

BK here they go. I think you should take your beter's on Sharikou. You are the man with your PhD. Take the 100:1 odds for BK in 2008. It'll be like taking candy from a baby. With these benchmarks I say wager a thousand bucks. Man I'll even spot you a hundred bucks and you take care of the 900 bucks. We'll be eating steak and fine wine when we see INTEL BK in 2008.

Lets put the wager down PhD

9:49 PM, October 24, 2006  
Anonymous Anonymous said...

What Dell actually said was

"20 percent less power than previous-generation four-socket PowerEdge servers"

The previous is the important bit here as they are not comparing AMD to latest Intel server chips but the last generation I think.

10:24 PM, October 24, 2006  
Anonymous Anonymous said...


BK here they go. I think you should take your beter's on Sharikou. You are the man with your PhD. Take the 100:1 odds for BK in 2008. It'll be like taking candy from a baby. With these benchmarks I say wager a thousand bucks. Man I'll even spot you a hundred bucks and you take care of the 900 bucks. We'll be eating steak and fine wine when we see INTEL BK in 2008.

Lets put the wager down PhD


LOL ok ok Sharikou is joking...I hope

12:14 AM, October 25, 2006  
Anonymous Anonymous said...

Yeah but all of intel's processors in the 4P space are netwurst thus far, meaning that previous gen and current gen of intel in the 4P server space would be the same thing and power consumption between them would be similar, easily bested by AMD

12:25 AM, October 25, 2006  
Anonymous Anonymous said...

Too bad I found only one C2D based test in the latest Cint2006 benchmakrs and no 2P 3GHz K8 ones.

AMD Opteron 2220SE, SPECint_rate_base2006 = 40.2
Intel Xeon 5160, SPECint_rate_base2006 = 45.2

Good thing is they both had identical OSes and other settings. Only CPU's, RAM and motherboards were different.

Assumint perfect linear scaling, 3GHz K8 would get a score of about 43. Not that much slower than the Xeon.

12:41 AM, October 25, 2006  
Anonymous Anonymous said...

I'm sorry Intel FANATIC boys but doesnt:
- INTEL sell and manufactures the XEON 7130M?
- Doesnt that processor released 2 months ago, so its a brand new processor from Intel?
http://www.eweek.com/article2/0,1895,2009916,00.asp

So the processor is bad, and thats Dell, IBM, AMD, ... fault not Intel.

"The previous is the important bit here as they are not comparing AMD to latest Intel server chips but the last generation I think. "

They are COMPARING DELL previous (2 moths OLD) servers to their NEW release ONES. Both lines will exist. The AMD is superior so superior MODEL NUMBER.

AMD is PowerEdge6950
INTEL is PowerEdge 6850

http://biz.yahoo.com/bw/061023/20061023006128.html?.v=1

3:35 AM, October 25, 2006  
Anonymous Anonymous said...

Now in the 2P space

FP Intel 81.0 AMD 73.5

http://www.spec.org/cpu2000/results/res2006q3/cpu2000-20060626-06346.html

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20060918-07379.html


INT Intel 120 AMD 82.6

http://www.spec.org/cpu2000/results/res2006q2/cpu2000-20060515-05998.html

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20060918-07375.html

Come on... You got a very poor result from Opterons FP...
2218@2.6GHz:
Base with Intel C++ Compiler: 71.7
Base with PathScale: 82.6
Base with Sun Studio: 101
Peak with Sun Studio: 117

How can you say using ICC for Opteron is fair if for Opterons it performs 40% less than Opteron max and almost max and almost max for woodcrest?

http://www.spec.org/cpu2000/results/res2006q3/cpu2000-20060901-07162.html
http://www.spec.org/cpu2000/results/res2006q3/cpu2000-20060901-07161.html
http://www.spec.org/cpu2000/results/res2006q3/cpu2000-20060815-06997.html

4:04 AM, October 25, 2006  
Anonymous Anonymous said...

Fanboys, FLies.. Spot Sharikou the 1000 bucks.

I want to wager the Pretender and his hovering flies on INTEL BK in 2008 Q2 is that right?

100:1 odds...

Where are you betters...

What you not confident INTEL is going BK?

5:38 PM, October 25, 2006  
Anonymous Anonymous said...

Interesting choices.

With no effort at all, I find in Q3 the Acer Altos R910 4 CPU 3.4 GHz Intel Xeon Processor 7140M, which has integer of 163 and FP 105 , better than the SunBlade X8400 at 156 but worse than the FP of 182. HP and Dell models have equivalent performance. Not quite the big difference as illustrated for integer. And you have to know that Intel will have a Woodcrest-based MP offering, but that Intel spends extra time validating MP chips. When it comes out, will you post the new scores if they are less than favorable for AMD?

As far as the 279 for an 8 CPU version, now you talking about the price and power-insensitive big iron space. Want to compare other 8 CPU options? Lets see:

16 core, 8 CPU Itanium box from SGI: no integer score, FP 486
Unisys: 247 int, no FP

16 core, 8 CPU SunFire X4600:
279 int, 231 FP.

Great Integer score, about 15% faster. Need FPU, about more than 2X slower (SGI 210% faster! That's a fragging!!)

How about UltraSparc IV+?
AMD and Intel both wipe the floor with it.

IBM Power5 scores for 8 way aren't shown, but would at least be competitive with both.

8:09 PM, October 25, 2006  
Anonymous Anonymous said...

You fools.

Placing a wager/bet on a prediction is stupid.

They're predictions from the current statistics which he analyses and presents them to us for open discussion.

Grow up. Do something with your life. Graham, keep investing in Intel shares, who really cares anyway?

You're making a profit, so like it matters what Sharikou has to say?

12:54 AM, October 26, 2006  
Blogger Marc said...


>AMD Opteron 2220SE, score = 40.2
>Intel Xeon 5160, score = 45.2
Good thing is they both had identical OSes and other settings. Only CPU's, RAM and motherboards were different.
Assumint perfect linear scaling, 3GHz K8 would get a score of about 43. Not that much slower than the Xeon.


Ah! These SPECint_rate_base2006 scores are interesting... Because it means that the
fastest Opteron 22xx is 11% slower than the fastest Woodcrest 51xx, which is not
much. But it also means that K8L only needs to be 11% faster than K8 in order to regain
the performance lead in SPECint_rate_base2006 over Woodcrest. We live in interesting times !

1:28 PM, October 26, 2006  

Post a Comment

Links to this post:

Create a Link

<< Home