One thing that everyone seems to be overlooking with the opteron vs. the xeon is:

Part of the reason AMD is not attempting to reach higher than 2Ghz right now is power consumption. If their CPU is out-performing the competition at a lower clock speed. Then they can work more closely on perfecting it and making it more server safe. Take into account that with a lower consumption of power, they can also reduce heat, which in turn increases CPU life. Remember, stability is far more important in server construction than speed. A server is no good to any company if it fails and has to be down for any amount of time. Also take into account, that regardless of how fast a server is, it still is restrained by network bottlenecks and bandwidth, along with being limited by the connecting workstation/pc.

Absolute Computer Solutions, Inc.
Andrew Sanders/ Owner

I hate to summon Shintai's ghost, but...

AMD said a 2GHz Barcelona gets a 28.7% higher score than an Intel Xeon E5345 at 2.33GHz in SPECfp_rate2006

I'm shocked, shocked that AMD's new CPU beats a 15%-higher clocked Intel chip in a benchmark where their old CPU beat the same higher-clocked Intel chip. Especially if they're doing the same stuff they did last time and using old Intel numbers.

One thing that is often missed by the people who pander to "zomg Intel is @ 4 GHz already!" crowd is this:

Given that they are already producing these as fast as they can to get some kind of inventory together, they've probably produced a million chips - including duds, previous samples and previous runs.

In order to showcase the "amazing power of the Phenom processor" at 3 GHz, they need 1 processor out of that million.

The chances are pretty high that of that million processors, one will run at 3 GHz and higher. You bet your bottom dollar that it's the very latest stepping, with a ton of process tweaks, and even then they're not exactly "common".

In summary, there are lies, lies, statistics, and presentations.

Exciting stuff. Looks like Intels simple shrink won't nearly be enough!

"AMD said a 2GHz Barcelona gets a 28.7% higher score than an Intel Xeon E5345 at 2.33GHz in SPECfp_rate2006."

With some clever multiplication, this appears to translate into almost exactly 50% better IPC... in that benchmark.

Am I missing something here? I would have thought a demo of a 3GHz quad core would have everyone drooling. Most of the reaction seems to be a big yawn.

Mostly because if they could actually produce it we'd be seeing press releases from AMD with choice lines such as, "Byaaaaaah!"

Instead we got a press release that said this:

"AMD still has no intention of releasing Barcelona chips clocked higher than 2GHz before the fourth quarter of this year. The company says the combination of low-power Barcelonas clocked at up to 1.9GHz and regular Barcelonas at up to 2GHz should take care of 94% of the market at launch, and that it will address the remaining 6% of shipments with Barcelonas running at "2.3GHz and above" in the fourth quarter."

Actions speak louder than words and all those words say is, "Need_more_spin! Activate the analyst dog and pony show!"

Am I missing something here? I would have thought a demo of a 3GHz quad core would have everyone drooling. Most of the reaction seems to be a big yawn.

It's not just a shrink... there are core enhancements too... besides Intel isn't just gonna slack around... for all we know, Intel may already have cooked up something by the time these products are launched by AMD.

I really really want to see some damn straight up standard benchmarks.

How does a 2ghz quad core one of these go vs a 2ghz quad core intel?
Is it 5% faster, 5% slower? 50% slower / faster etc?

Then once we know that, how well does it overclock? how much headroom is there on the chip?

THEN how much does it cost, is the chipset any good, does the chipset support pci-e 2.0 / ddr2 / ddr3 / dd9 etc

more info, clear info - we need a lot more and we need products.
I'm in no way an AMD fanboy but the news is so slow to come, the benchmarks no where to be seen, it's difficult to hold on and patiently wait when Q6600's are 266$ US and other C2D's are substantially cheaper.

To further emphasize all this, I mean if the info is taking this long to come out then one can only conclude that they are still a long way off from us walking into a store and picking up the product any time soon, right?

I was expecting benchmark figures or something (real ones, not the simulated kind).

100MHz is all that separates "low-power" from a standard CPU? They couldn't squeeze that extra 100MHz in without breaking the voltage bank?

AMD is saying 2ghz will take care of 94% of the market...

As ever, the issue is whether or not DAAMIT can really execute this plan effectively.

Since HT3.0 offers the option to split each link into 2x8b (rather than 1x16b), that means Sandtiger with its 4 ccHT links will be able to do directly-connected glueless 8S servers, for 32 cores (or more, if they go above quads at 45nm). With an IOMMU and nested page tables, that's going to make for some pretty compelling server offerings, especially for IT depts looking at consolidation.

Wow a Phenom @ 3 ghz - that should give existing C2D/C2Q a run for their money (even OC'ed). But I doubt it will hold up against Penryn :-/

Yawn, wake me for the hexadeca-cores. :P

No mention of EE for Phenom X2 in 2008? I recall hearing about the GE-6x00 series chips (with 45W TDP) said to be appearing in Q1.