Opteron (hammer) SPEC 2000 benchmarks.

[alexruiz]

AMD has posted a pdf document showing some benchmarks for an opteron 2 GHz.... no more details are provided about the test platform.

PDF document

They seem low in my opinion... I was expecting higher numbers.

If you find anything I missed, please post it here.

Thanks

 

[alexruiz]

Sorry, I forgot the numbers:

SPECint 2000: 1202 estimated (I assume this should be base)

SPECfp 2000: 1170 estimated (I assume this should be base)

What do you think??

 

[Diablo6178]

Take a look at the Intel Spec CPU 2000 Scores

I'd say that's great perfomance for a chip that is only at 2Ghz. Considering the Thuroughbred core is at 2.13 currently.

 

[alexruiz]

I agree that the numbers look good, but if you remember the hype, the clawhammer (a more humble CPU, because Opteron is sledgehammer) was expected to have a SPECint of 1350 for a clawhammer 3400+.... Well, maybe that was only that, hype....

 

[First]

Operton and ClawHammer won't be released for six months. A lot of performance tweaking can happen between now and then. Not just on the CPU side though, it can (will) happen on the chipset, motherboard, driver, etc. etc. side of things.

All in all these are very good numbers.

 

[Wingznut]

Originally posted by: Diablo6178
Take a look at the Intel Spec CPU 2000 Scores

I'd say that's great perfomance for a chip that is only at 2Ghz. Considering the Thuroughbred core is at 2.13 currently.

What's the point of comparing a server chip (Opteron) to desktop chips (P4 and T-bred)? They aren't in the same market.

Now, if you wanted to compare it to a Xeon or Itanium... That would make more sense.

Well, actually... A more accurate comparison would be to use future cpu's like Gallatin or Madison, instead of today's cpu's.

 

[JwLeonhart]

AMD better hurry up with any chip they make or else intel will crush them, and I hate intel. Intel will have a better P4 out before AMD can come out with anything. I just hope AMD makes a good chip...

 

[Diablo6178]

Originally posted by: Wingznut

Originally posted by: Diablo6178
Take a look at the Intel Spec CPU 2000 Scores

I'd say that's great perfomance for a chip that is only at 2Ghz. Considering the Thuroughbred core is at 2.13 currently.

What's the point of comparing a server chip (Opteron) to desktop chips (P4 and T-bred)? They aren't in the same market.

Now, if you wanted to compare it to a Xeon or Itanium... That would make more sense.

Well, actually... A more accurate comparison would be to use future cpu's like Gallatin or Madison, instead of today's cpu's.

Show me a material difference between a 2.4B Desktop P4 and a Xeon 2.4B excluding price. Comparing the Itanium to the Opteron isn't fair do to Price segments and server markets. Itanium is meant to compete with the IBM's Highend and HP's Alpha servers.

 

[Wingznut]

If you don't think the Opteron is meant to compete with Itanium, then you probably ought to notify AMD...

Originally posted by: AMD
The AMD Opteron processor is designed to provide flexibility, scalability and performance for demanding enterprise-class applications. It is planned to effectively compete against the Intel Xeon and Itanium processors.

EDIT: I personally agree that Opteron isn't competitive with Itanium (ESPECIALLY Itanium 2 and the next Itanium, codename Madison) and won't be used in the mission critical environment that Itanium is designed for... However, AMD obviously chooses to use it as a target.

 

[Diablo6178]

OK I stand corrected on that but I'll add this little foot note. To compete with both the Opteron is going to be aimed at the Xeon in Single and Dual CPU configuration and the Itanium in Quad and Hex configurations. Itanium is one exspensive chip, which is the main reason Intel is looking at x86-64 for Prescott.

 

[Wingznut]

Intel has already come out and said (mulitple times) that they aren't designing any kind of hybrid 32bit/64bit chip.

And remember that Itanium systems are significantly cheaper than the competition's 64bit high end servers (Sun/HP/IBM/etc). So while they may seem "expensive" compared to desktop units, they aren't when compared to their peers.

 

[jbond04]

There is actually a big difference between the P4 and the Xeon. It's called "hyperthreading". For server level applications (like what the Opteron and Xeon are targeted for), Hyperthreading provides are very real, tangible performance increase. In fact, hyperthreading (aka SMT) is going to be implemented on the 3GHz P4. It is THAT important.

So for server benchmarks, the Xeon can easily beat an equally clocked P4.

 

[Diablo6178]

Look HERE to see just how mcuh it helps out. at best about 17% on a dual CPU setup less on a single and from looking at it against the competition it's just catching up. As for HT on the Desktop when they impliment it isn't there even less of a difference then? I'll agree that HT will provide an increase but in that test it did no better then catch up with the Athlon MP or just surpase it. Read Anand's conclusion,

"But the real kicker here is what happens when Hyper Threading is enabled on the dual 2.2GHz Xeon platform. The resulting 17% improvement brings the new Xeons back into the game and puts them less than 3% away from the fastest dual Athlon MP 2000+ setup. This is by far the largest tangible performance improvement we've been able to produce in a real-world setting on Intel's 2.2GHz Xeons. For being the first true incarnation of a Hyper Threading enabled processor, we can say that we're impressed with the technology at this point. The performance boost not only resulted in a noticeable improvement in DB processing power but it also made the Xeon competitive. Without Hyper Threading, the Xeon would be in some serious trouble in the server world from a performance standpoint. With Hyper Threading enabled however the 2.2GHz Xeons can keep up with AMD's Athlon MP 2000+ CPUs. Oh, how times have changed; it used to be that AMD would never have been taken seriously in the enterprise market, and now we're talking about Intel striving to be competitive enough with the AMD's enterprise level performance."

Intels now up to 2.8 Xeon but they are running on the same bus and core that the 2.2's were. So while the perfomance will scale it's not going to be earth shattering considering a speed increase of 27%. I'm willing the bet that the 3.0Ghz P4 will be released at the same time as the 3.0Ghz Xeon because they will effectively be the same part. I'd also be willing to bet that HT is on the Northwood core it's just disabled.

*I just realized I'm arguing with an Intel employee*

 

[ErikaeanLogic]

*I just realized I'm arguing with an Intel employee*

LOL

 

[jbond04]

Apparently, Intel is said to have improved their hyperthreading implementation on the new 3.06GHz P4 processors...but this is speculation. And you're correct in that the Northwood core does have HT on it's die--all P4's do. However, it's disabled at the fab...so that's a moot point.

The crux of this whole argument is that HT does provide a significant enough performance increase to differentiate the Xeon line from the P4 line. How much faster is the Opteron than the Clawhammer? Is it enough to call the two the "same" thing, just as you lumped the Xeon and P4 together?

 

[alexruiz]

Going back into topic, remember that SPEC is heavily dependant on the compiler (that is why the P4 get higher marks than the Athlon in SPECfp). It seem these tests were done using a 32 bits Intel compiler, and Fred Weber states that using a x86-64 compiler can improve the scores by 20%.... not that bad.

But I was still expecting those 1350 SPECint for the clawhammer....

 

[Wingznut]

Originally posted by: Diablo6178
*I just realized I'm arguing with an Intel employee*

Nah, we're not arguing.

Besides, if you are referring to your HyperThreading discussion, that's not even with me.

But we were discussing Xeon vs Northwood SPEC scores... To be honest, I'm not sure how the Xeons score in SPEC. Anybody happen to have a link of some scores for a 2.8 Xeon w/HT?

 

[First]

Originally posted by: jbond04
Apparently, Intel is said to have improved their hyperthreading implementation on the new 3.06GHz P4 processors...but this is speculation. And you're correct in that the Northwood core does have HT on it's die--all P4's do. However, it's disabled at the fab...so that's a moot point.

The crux of this whole argument is that HT does provide a significant enough performance increase to differentiate the Xeon line from the P4 line. How much faster is the Opteron than the Clawhammer? Is it enough to call the two the "same" thing, just as you lumped the Xeon and P4 together?

No, this is not speculation, it's fact. It's been confirmed to us that the 3.06GHz P4's and up utilize a new stepping of HT which, from the demos we saw of 3.06 HT CPU's at IDF, greatly improve performance...

 

[IdahoB]

Hmm... this may sound dumb but every one of the benchmarks of spec1350 for the hammer said they were for the clawhammer 3400+ - it never said what clock speed that would actually run at - maybe it's a little faster than 2GHz, hence the higher results.

I agree though - AMD had better pull back the performance crown from Intel or most of the enthusiast market will abandon them.

 

[chizow]

I'm going to wait for Megatron, I never liked the Autobots much.

Chiz

 

[Kell]

Originally posted by: Wingznut
If you don't think the Opteron is meant to compete with Itanium, then you probably ought to notify AMD...

Originally posted by: AMD
The AMD Opteron processor is designed to provide flexibility, scalability and performance for demanding enterprise-class applications. It is planned to effectively compete against the Intel Xeon and Itanium processors.

EDIT: I personally agree that Opteron isn't competitive with Itanium (ESPECIALLY Itanium 2 and the next Itanium, codename Madison) and won't be used in the mission critical environment that Itanium is designed for... However, AMD obviously chooses to use it as a target.

Funny, the Opteron sample appears to defeat Itanium2 (and everything else) in SPEC integer performance by a hefty margin. FP performance isn't as good, but Opteron makes a pretty decent showing overall.

Plus, unless Intel has actually gotten away from their shared SMP bus architecture, Itanium2 won't scale nearly as well multiprocessor-wise as Opteron. The terribly inefficient power consumption on IA64 is another major tripping point...

Originally posted by: alexruiz
I agree that the numbers look good, but if you remember the hype, the clawhammer (a more humble CPU, because Opteron is sledgehammer) was expected to have a SPECint of 1350 for a clawhammer 3400+.... Well, maybe that was only that, hype....

According to the only rumors we have, the Hammer is supposed to achieve a clock speed of around 2.6GHz to get anywhere close to a 3400+ rating.