• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

AMD DEMOed Barcelona Native Quad Core

ahock

Member
Have you guys watch the keynote in AMD website? AMD is projecting something like 70% anfd 40% in some benchmarks. I noticed though is they are comparing it to Intel Xeon 5300 which isn't it that Barcelona is an MP processor? Plus noticed that Xeon 5100 series were bested by AMD 2000 Opteron which based from most benchmarks Woodcrest is really the clear winner. What do you think guys?
 
SPECfp_rate has always run well on the Opteron platform. 1.4x the Xeon 5355 gives it a score of 140-150, 4 dual-core Opteron 8220 scores 175-178. As for the other benchmark, those types of tests almost alway favor Woodcrest by a significant margin, so its hard to say which test they used.
 
Originally posted by: anandtechrocks
Estimated Performance 1
Estimated Performance 2

Looks pretty good.


Thanks anandtechrocks. This is what I'm confused with. 5335 is a DP while Barcelona is an MP system 70% and 40% is something I think incomparable because they are 2 different systems. Although these are clovertown already but using 2 sockets vs 4 sockets for barcelona. I was kinna expecting more than 70% and 40% for these if they are to compare MP from DP.

Plus, are you guys agree that Opteron 2000 SE is ahead of woodcrest?
 
Its all marketing hype until we see real review sites...I never trust Intel's marketing BS either....

I am not overly surprised...I would hope their native quad core would have closed the gap....

 
Originally posted by: ahock
Originally posted by: anandtechrocks
Estimated Performance 1
Estimated Performance 2

Looks pretty good.


Thanks anandtechrocks. This is what I'm confused with. 5335 is a DP while Barcelona is an MP system 70% and 40% is something I think incomparable because they are 2 different systems. Although these are clovertown already but using 2 sockets vs 4 sockets for barcelona. I was kinna expecting more than 70% and 40% for these if they are to compare MP from DP.

Plus, are you guys agree that Opteron 2000 SE is ahead of woodcrest?

WTF are you saying? The 53xx series of Xeons are Quad cores though not native ones. It is a comparison of quad to quad...

 
Originally posted by: Duvie
Its all marketing hype until we see real review sites...I never trust Intel's marketing BS either....

I am not overly surprised...I would hope their native quad core would have closed the gap....

Yeah its all marketing but its all up to us to figure whether they are correct or not. I've seen AMD website benchmarks comparison that shows C2D lags in all front. This is another classic example that they are showing incorrect data, UNLESS I'm wrong.

Woodcrest from 3rd party sites is a clear winner but here they are showing that their SE opteron wins. Amd lastly, if its a 70% and 40% improvement is all they can offer I dont think they can win next year since these numbers are from DP to MP point of view. I guess it would be best to compare barcelona to Tulsa system as these benchmarks they are showing are for mostly an MP system benchmarks.
 
Originally posted by: ahock
Isn't it Caneland/Tigerton already been demoed? Meaning this is the real competition of Barcelona.....



I thought it was called cloverton...being Intels first native quad core...
 
Tigerton is the MP version of Cloverton, I think. So no surprises there.

A 70% increase in performance compared to an CURRENT Opteron is not all that great, the thing has 2x as many cores, for god's sake. That's basically close to what I'd expect without a redesign.

40% higher performance than a 2.66GHz cloverton does look pretty good--that would be the equivalent of a 3.76GHz Cloverton, assuming linear performance increases--but FP has always been AMD's strong point and we have yet to see AMD's integer performance.
 
Are you saying the % increase is compared to dual cores?? If so...SO!!! I mean as long as it beats the INtel quad.....
 
Originally posted by: Furen
Tigerton is the MP version of Cloverton, I think. So no surprises there.

A 70% increase in performance compared to an CURRENT Opteron is not all that great, the thing has 2x as many cores, for god's sake. That's basically close to what I'd expect without a redesign.

40% higher performance than a 2.66GHz cloverton does look pretty good--that would be the equivalent of a 3.76GHz Cloverton, assuming linear performance increases--but FP has always been AMD's strong point and we have yet to see AMD's integer performance.



well though there are 4 cores versus 2 I haven't seen much in terms of perfect scaling...i wouldn't have really expected 100% increase. PLus they really didn't list the speed of the barcelona....
 
Originally posted by: Duvie
Originally posted by: ahock
Originally posted by: anandtechrocks
Estimated Performance 1
Estimated Performance 2

Looks pretty good.


Thanks anandtechrocks. This is what I'm confused with. 5335 is a DP while Barcelona is an MP system 70% and 40% is something I think incomparable because they are 2 different systems. Although these are clovertown already but using 2 sockets vs 4 sockets for barcelona. I was kinna expecting more than 70% and 40% for these if they are to compare MP from DP.

Plus, are you guys agree that Opteron 2000 SE is ahead of woodcrest?

WTF are you saying? The 53xx series of Xeons are Quad cores though not native ones. It is a comparison of quad to quad...

They are quad to quad but Barcelona being used for these benchmarks are MP system as what AMD is trying to show and clearly 16 cores in task manager. 53xx series are quad core BUT on DP system. That is why I'm trying to clear whether my understanding is correct or not. If my assumptions are correct then 70% and 40% I think is not enough. AMD is using 4 socket to get 70% and 40% to beat cloverton 2 sockets only.

 
Originally posted by: ahock
They are quad to quad but Barcelona being used for these benchmarks are MP system as what AMD is trying to show and clearly 16 cores in task manager. 53xx series are quad core BUT on DP system. That is why I'm trying to clear whether my understanding is correct or not. If my assumptions are correct then 70% and 40% I think is not enough. AMD is using 4 socket to get 70% and 40% to beat cloverton 2 sockets only.
I think it's most likely a comparison of two-socket systems with a 40% and 70% edge over the Opteron dual-core system.
 
No problem I do that as always as you do....

Although these are preliminary but for me these numbers are not very impressive as compared to the preliminary data being showed when Intel previewed C2D.
 
Originally posted by: Accord99
Originally posted by: ahock
They are quad to quad but Barcelona being used for these benchmarks are MP system as what AMD is trying to show and clearly 16 cores in task manager. 53xx series are quad core BUT on DP system. That is why I'm trying to clear whether my understanding is correct or not. If my assumptions are correct then 70% and 40% I think is not enough. AMD is using 4 socket to get 70% and 40% to beat cloverton 2 sockets only.
I think it's most likely a comparison of two-socket systems with a 40% and 70% edge over the Opteron dual-core system.

That's what I initially concluded but when I read and watched the video, somehow they are showing is an MP system 😀. For me I think they should have compared this with Tulsa and project more that 100% increase as Tulsa is still a netburst derivatives. This makes me think that Tulsa is really somewhat competitive performance wise.
 
I don't know if anyone noticed this, but the video told us that quad cores will work with socket AM2! This text from AMD's site confirms it too:

Additionally, 1P systems using AMD Opteron processor for socket AM2, support unbuffered memory at frequencies up to DDR2-800, which can further benefit application performance. In 2007 the same DDR2-based platforms will be able to upgrade to Quad-Core AMD Opteron processors, without requiring thermal or cooling infrastructure changes, for significantly more performance.

I know that there has been a lot of talk on AM2 and quad core support and a lot of people didn't believe it before this. This should settle that discussion.

Should be fun to drop one of those into my system in about 9 months. May go with a dual-core, though, since quads are bound to be very expensive.
 
Am I the only one missing something out here? Aren't these numbers forecasted and virtualized? Meaning they still don't have a physical live sample of their new Opteron even made yet?

If so, this is the same crap we've been getting for over a month now. Though it's funny to see AMD on the opposite side trying to strengthen their marketing campaign on a lower performing part. I can see them in the meeting now discussing this over and over "lets talk about our strengths". I must of heard the word direct connect and "leading industry performance" about a thousand times. At least the chief executive financial and marketing officer made clear that innovation is crucial. Though its how they plan to "innovate" that makes me worrisome. The whole torrenza initiative and direct connection platform sounds great and all but when you are getting outperformed by a "20 year old" platform it just makes you look bad.
 
Do you think FSB/Direct connect do have merit or purely marketing scheme? I ask because AMD is stil hyping the said advantage wherein Intel clearly shown that they can still push FSB above its current performance as in the case or C2D. AMD still hyping it again in Quad FX which shows no performance increase as compared to Kentsfield. Scalability wise as reported by Anand article shows that they are not scaling as suppose to be.

This is again being touted by AMD in the native quad core
 
Well, we're watching an advertisement, what did you expect?

Does anyone else find it funny that AMD just got its ass handed to it by a "legacy frontside-bus based" architecture?
 
Funny in the short-term amusement sense, disheartnening in the "Intel has less incentive to drop their prices or roll-out higher performing products" sense.

Healthy competition is good, not the kind that leads to lay-offs and folks not making their mortgage payments, but good old-fashioned market-space competition is good.

I will find it equally not funny if R600 does not pwn G80 in some sense when AMD releases in spring. $650 G80's need some competition.
 
Back
Top