picture of 4-way opteron system

[Soulkeeper]

mmmmmmm tasty
just found it
also they got a P4 2.8 oc'd to almost 4ghz with super cooling
looks nice but i can't read their language

 

[rIpTOr]

Hmm that looks like it can take 20GB's of RAM there.

I also saw this http://www.muropaketti.com/artikkelit/sekalaista/amd_hammer/index.phtml
6.4GB/sec I sure would like an Opteron (drools) It looks like Intel might get a run for their money in the server market.

 

[SSXeon5]

Originally posted by: rIpTOr
Hmm that looks like it can take 20GB's of RAM there.

I also saw this http://www.muropaketti.com/artikkelit/sekalaista/amd_hammer/index.phtml
6.4GB/sec I sure would like an Opteron (drools) It looks like Intel might get a run for their money in the server market.

Too bad the Itanium 2 has 6.4gb/s bandwidth 128-bit 400Mhz fsb, Quad channel DDR200 and 3MB L3 ondie cache They smack SUN's/Alpha's/IBM's chips and at a super low cost. IMHO I think the opteron will compete with Xeon MPs. They will go to the northwood core in 2003 and get up to 2GHz+ speeds with 2MB L3 ondie cache. Still it could compete with the Itaium 2/3 because of Jerry saying so much that the Opterons are faster then the Itaniums, we will see But even by the time 4-way Opterons come, Madison will launch with 6MB L3 ondie, and ranging from 1.2Ghz-1.6GHz in core speed. Im guessing that is what the Opteron will launch at, 1.6GHz. And yes Madison will be 4way and possibly 8way and prob a faster Bus speed, Packing 500m transistors too

Soulkeeper good link, crazy people with there LN2 That is a tight picture with the opterons, good to see it in 4-way setups. Sad there isnt any more info or pictures of clawhammer or sledgehammer yet.

SSXeon

 

[rIpTOr]

Yep Intel never lets anyone get the best of them for too long. I think The Opteron was made to compete with Xeon... Itanium is almost a whole different game.

 

[SSXeon5]

Originally posted by: rIpTOr
Yep Intel never lets anyone get the best of them for too long. I think The Opteron was made to compete with Xeon... Itanium is almost a whole different game.

How do you figure? The Itanium is Pure 64-bit, and unlike AMDs 32bit with 64bit extensions helps amd want to be "faster" in the 64-bit department. I mean Jerry went off on how stupid the Itanium is and how much the hammer can out perform it. LOL!! When he retires he might see how much of a stupid little school girl he was when he said all that.

SSXeon

 

[andreasl]

Opteron will beat Itanium 2 in Integer workloads for sure. Hell even the P4 and XP does that now. Once IA64 moves to 0.13 things will get interesting though. Integer performance will probably be pretty evenly matched while the Itanium (3?) will walk all over Opteron in float performance.

 

[BD231]

Just wondering, who would use such a configuration? It dose not seem like a cost effective investment, and I'm sure that thnig will make an outrageous amount of heat. I cant even see a large corp. buying such a config.

 

[tritium4ever]

Originally posted by: bdog231
Just wondering, who would use such a configuration? It dose not seem like a cost effective investment, and I'm sure that thnig will make an outrageous amount of heat. I cant even see a large corp. buying such a config.

Large corporations are Sun and Intel stalwarts anyway. Getting em to switch to a total n00b player in the server market like AMD would be one hell of a task for the sales force.

 

[Jigman]

Originally posted by: SSXeon5

Originally posted by: rIpTOr
Hmm that looks like it can take 20GB's of RAM there.

I also saw this http://www.muropaketti.com/artikkelit/sekalaista/amd_hammer/index.phtml
6.4GB/sec I sure would like an Opteron (drools) It looks like Intel might get a run for their money in the server market.

IMHO I think the opteron will compete with Xeon MPs. They will go to the northwood core in 2003 and get up to 2GHz+ speeds with 2MB L3 ondie cache.SSXeon

They do look like good competition

 

[Kell]

Originally posted by: SSXeon5

Originally posted by: rIpTOr
Yep Intel never lets anyone get the best of them for too long. I think The Opteron was made to compete with Xeon... Itanium is almost a whole different game.

How do you figure? The Itanium is Pure 64-bit, and unlike AMDs 32bit with 64bit extensions helps amd want to be "faster" in the 64-bit department. I mean Jerry went off on how stupid the Itanium is and how much the hammer can out perform it. LOL!! When he retires he might see how much of a stupid little school girl he was when he said all that.

SSXeon

Ja, but that's just because Jerry Sanders (and Craig Barrett, too) have a bad habit of shooting their mouths off prematurely. Probably no one in their right mind is going to take Opteron over Itanium2 just for its performance. Opteron over Itanium, maybe.

But that isn't going to be the point of Opteron anyways...everyone's pretty much agreed on that. It's pitched as a transitional chip more than anything.

 

[pm]

Ja, but that's just because Jerry Sanders (and Craig Barrett, too) have a bad habit of shooting their mouths off prematurely.

I can't think of a single instance of Craig Barrett shooting off his mouth about anything. He's pretty even-keeled. Can you cite an example of Craig shooting his mouth off? McNealy, Sanders, and Ellison are all famous for various utterances - and we can't forget infamous TJ Rodgers (Cypress Semiconductor) - but Craig is not well known for saying things without consideration.

Opteron will beat Itanium 2 in Integer workloads for sure. Hell even the P4 and XP does that now. Once IA64 moves to 0.13 things will get interesting though. Integer performance will probably be pretty evenly matched while the Itanium (3?) will walk all over Opteron in float performance.

Aside from the obvious unfairness of comparing a currently shipping 0.18um product against the estimated performance of a product that's at least several quarters off, I'd like to note that the fastest currently shipping Althon XP does not beat the Itanium 2 in integer performance using the industry standard SPECInt benchmark. Nor, AFAIK, in other major predominantly integer benchmarks such as tmpC and OpenSSL speed.

 

[andreasl]

Aside from the obvious unfairness of comparing a currently shipping 0.18um product against the estimated performance of a product that's at least several quarters off, I'd like to note that the fastest currently shipping Althon XP does not beat the Itanium 2 in integer performance using the industry standard SPECInt benchmark. Nor, AFAIK, in other major predominantly integer benchmarks such as tmpC and OpenSSL speed.

pm,

I was merely comparing the expected IA64 performance to Opteron when Opteron is released. AFAIK the 0.13 version of Itanium isn't expected until H2 next year right? This will give Opteron a couple of months headstart. And you don't need to convince me that comparing a 0.13 chip to a 0.18 chip is unfair from an architecture point of view. But there are more ways to make a comparison. And one of these is to compare the best products currently shipping. Intel shipped Northwoods long before AMD shipped Tbreds and I didn't see alot of people complaining when comparisons were made there.

And as far as SPECint. The new XP2600 scores 813/839 which is a hair above Itanium 2. The P4 2.8 scores in the region of 970 (don't think these has been published yet). I know the XP2600 isn't availible JUST YET. But unless Intel ships a 1.1GHz Itanium 2 within the next few weeks I don't think my comparison was that off...

Basically my future comparison of these chips was made with these few assumptions which I consider very resonable:

1) Opteron ships before IA64 reaches 0.13 if only by a few months.
2) Opteron will perform higher than the fastest shipping Athlon XP at that time.

Bt the way, substitute Opteron with the Athlon K8 if you want and it becomes even more clear as these are based on the same core.

If these assumptions become true then Opteron WILL indeed beat Itanium 2 in SPECint. But with IA64 hitting 0.13 just shortly after that things will get interesting again. It may end up beating Opteron. That would not be a very unresonable claim. We'll know in about a year

 

[Kell]

Originally posted by: pm

Ja, but that's just because Jerry Sanders (and Craig Barrett, too) have a bad habit of shooting their mouths off prematurely.

I can't think of a single instance of Craig Barrett shooting off his mouth about anything. He's pretty even-keeled. Can you cite an example of Craig shooting his mouth off? McNealy, Sanders, and Ellison are all famous for various utterances - and we can't forget infamous TJ Rodgers (Cypress Semiconductor) - but Craig is not well known for saying things without consideration.

Back when all the Rambus controversy was at its height, Craig Barrett had a bad habit of letting anti-Rambus comments slip. He would say something negative about Rambus, then reporters (especially techie scandal-rags like the Register) would latch on to it and run, then Intel's spin doctors would have to deal with it. Not the kind of emotional outbursts you get from Sanders, but Sanders actually has reason to be emotional. There's a lot of bad blood between Sanders and the old Intel management.

 

[dannybin1742]

i think the best selling point will be the price, itanium 2s are more than 10grand a peice, and how much will opterons be? no more than 1k

but intel does make some sweet A$$ server processors

 

[pm]

I hadn't see the Spec scores on the new Athlons. I'd say "I stand corrected" except that I did say "currently shipping" in my posts just on the chance that the newer CPUs were, in fact, faster. Thanks for the information, andreasl.

Back when all the Rambus controversy was at its height, Craig Barrett had a bad habit of letting anti-Rambus comments slip. He would say something negative about Rambus, then reporters (especially techie scandal-rags like the Register) would latch on to it and run, then Intel's spin doctors would have to deal with it. Not the kind of emotional outbursts you get from Sanders, but Sanders actually has reason to be emotional. There's a lot of bad blood between Sanders and the old Intel management.

I don't know that I would consider the very mild responses that Craig would give regarding future memory roadmaps to be shooting off his mouth. He'd be asked very pointed questions about future DDR support after various speeches and he would respond by saying that there would be future chipsets that supported DDR and the press would latch onto this and print articles about Intel abandoning RDRAM. I see what you are saying, but I still don't think Craig says much without due consideration. Having said that, there is always Andy Grove who has always spoken his mind without mincing words.

Sanders actually has reason to be emotional.

I'd say that "Sanders actually thinks he has a reason to be emotional". And being the person that he is, he'd probably be saying similar things even if the past hadn't played out like it did.

itanium 2s are more than 10grand a peice.

$10k? That is an interesting number. How did you come up with that? According to the publicly released pricing at the time of the Itanium 2 launch, the 1GHz Itanium 2 with 3MB on-die cache is $4,226. The 1GHz Itanium 2 with 1.5MB is $2,247 and the 900MHz Itanium 2 with 1.5MB cache is $1,338. Prices are, as usual, in quantities of 1,000.

and how much will opterons be? no more than 1k

This is an interesting number as well. I doubt even AMD management would commit to pricing at this stage, but certainly they have priced CPU's above $1000 in the past. We will see what they sell for when they are released.