***Official Opteron review thread***

[sirfergy]

Well, I am waiting for a Windows 2003 comparison. Windows 2003 supports the hyper threading much better, and it also supports some extra AMD features as well.

 

[PrinceXizor]

If this is the Official Opteron review thread, why can't I see a list of opteron reviews in post one....gee, I can add asterisks to my threads too...

P-X

 

[Macro2]

Nothing can touch the Opteron in 2 way and 4 way.

Looks like AMD has moved out into the passing lane.

Compiler tweaks will put the petal to the metal even more.

Mac...

PS...TOms review is "catch as catch can" at best.

 

[kuk]

Originally posted by: Macro2
Nothing can touch the Opteron in 2 way and 4 way.

Looks like AMD has moved out into the passing lane.

Compiler tweaks will put the petal to the metal even more.

Mac...

PS...TOms review is "catch as catch can" at best.

How do you know how 4-way servers benchmark?

 

[astroview]

Originally posted by: PrinceXizor
If this is the Official Opteron review thread, why can't I see a list of opteron reviews in post one....gee, I can add asterisks to my threads too...

P-X

Could you update the first post with all the reviews please?

 

[microAmp]

Originally posted by: astroview

Originally posted by: PrinceXizor
If this is the Official Opteron review thread, why can't I see a list of opteron reviews in post one....gee, I can add asterisks to my threads too...

P-X

Could you update the first post with all the reviews please?

I just sent a PM to him to do that.

 

[Rectalfier]

The 4 way opteron should pummel a 4 way Zeon into dust. If you look at a lot of the benchmarks, the Opteron is able to DOUBLE performance when going from 1 processor to 2. The Zeon can mange maybe 60-70percent at best. This is due to Hypertransport, and the 4 way should showcase this technology even further.

 

[HurleyBird]

Im thinking about 8-way... and after NUMA is resolved mem performance will be SO much better

 

[Macro2]

Because I've been shown this by someone who does benchmarking for Intel.

 

[kuk]

Originally posted by: Macro2
Because I've been shown this by someone who does benchmarking for Intel.

I should have made myself clearer ... I was talking about how do you know how 4-way Opterons benchmark ...

 

[NFS4]

http://www.hardocp.com/article.html?art=NDY1

 

[HurleyBird]

Opteron sure is screaming for more memory bandwidth though... Imagine super cooled memory running at 900Mhz dual channel (a dream, yes i know) first time in a long time that i can dream of memory running at the FULL SPEED OF THE PROCESSOR *drool* the platform should support rambus...

 

[Armitage]

Originally posted by: kuk

Originally posted by: Macro2
Because I've been shown this by someone who does benchmarking for Intel.

I should have made myself clearer ... I was talking about how do you know how 4-way Opterons benchmark ...

SPEC benchmarks

"Crush" does seem to be the proper adjective. Take a look at the SPECfp_rate link on that page. Note that these were run with MS Windows Server 2003 for the Opts, which apparently does a better job with the NUMA stuff. Also doesn't help that the XeonMP needed for > 2P stuff only does 2GHz (but with 2MB L2).

 

[HurleyBird]

went to the spec site... compared specint rate of 2.2 2.4 and 2.6 xeon systems to the spec int rate of opteron 1.4, 1.6 1.7

XEON:
2.2 100%
2.4 106%
2.6 113%

Opteron
1.4 100%
1.6 113%
1.8 126%

!!!!!!!!!!!

Opteron scales twice as rapidly as xeon performance wise!

This is all due to the integrated memery controller

Clock speed rises... FSB rises... Latency lowers...

Intel will now have to work twice as hard as AMD to scale their products!

 

[Nemesis77]

Originally posted by: HurleyBird
Clock speed rises... FSB rises... Latency lowers...

FSB does not increase, it stays put. And in case you are talking about the speed of the memory-bus: it stays the same as well.

 

[Wingznut]

Originally posted by: HurleyBird
went to the spec site... compared specint rate of 2.2 2.4 and 2.6 xeon systems to the spec int rate of opteron 1.4, 1.6 1.7

XEON:
2.2 100%
2.4 106%
2.6 113%

Opteron
1.4 100%
1.6 113%
1.8 126%

!!!!!!!!!!!

Opteron scales twice as rapidly as xeon performance wise!

This is all due to the integrated memery controller

Clock speed rises... FSB rises... Latency lowers...

Intel will now have to work twice as hard as AMD to scale their products!

The difference between a 2.2ghz/2.4ghz/2.6ghz is 100%/109%/118%

Wheras the difference between a 1.4ghz/1.6ghz/1.8ghz is 100%/114%/129%.

 

[HurleyBird]

I used the single proccessor dell machine for my intel comparrison... will have to check the opteron... were did you get your results? i used the spec site for one, and official AMD site for the other... one way or another the AMD chip still scales (performance wise) quite a bit better than the intel chip.

 

[Wingznut]

No, I'm saying that you aren't comparing apples to apples. When you increase a 2.2ghz to 2.6ghz it's an 18% difference in clockspeed. But when you increase from 1.4ghz to 1.8ghz, that's a 29% difference in clockspeed.

You can't say that one "scales" better than the other, when you aren't comparing similar clockspeed increases.

 

[MikeMike]

dont ever argue with wingznut about processors he knows much more than you about them.


However, with the 18% difference in clock speed for intel, the whole difference then would mean that they are losing 5% of the spec int for every 400mhz while AMD is only loosing 3% every 400mhz. (dont know how that will continues) but that means that amd gets 2% more specint rate than intel for every 400 if that continues that way.

so yeah

(i dont know much, but figured that might be assumable)

 

[HurleyBird]

Oops, stupid of me...

for every point-four GHz you notice around a 1% difference in the "doubling trend" so i would say that gives around a 2-4% less for the P4 score... unfortunantly their is no 1.6 GHz xeon...

so the 2.6 would prob be at around 116% IMO

 

[Wingznut]

Originally posted by: nourdmrolNMT1
dont ever argue with wingznut about processors he knows much more than you about them.

Well... that's definitely not always true. I make mistakes just like anyone else (as pm just pointed out about one of my posts from earlier this week. )

 

[HurleyBird]

well... my mistake didnt have anything to do with my knowledge of processors, just my math skills

 

[MikeMike]

Originally posted by: Wingznut

Originally posted by: nourdmrolNMT1
dont ever argue with wingznut about processors he knows much more than you about them.

Well... that's definitely not always true. I make mistakes just like anyone else (as pm just pointed out about one of my posts from earlier this week. )

unless the other person is **Not speaking for Intel Corp.** also, then dont argue with either of them.


What are you doing shouldnt you be working? no wonder intel had problems with the 800FSB p4!!

 

[Wingznut]

Originally posted by: nourdmrolNMT1
What are you doing shouldnt you be working? no wonder intel had problems with the 800FSB p4!!

Tomorrow is the start of my work week.

 

[UlricT]

wheres the anandtech review??????