True, but an engine doesn't make a car.

This CPU will have to have the right support chips. There's a world of difference between a workstation chipset, and a server chipset.

The fear is that you will end up pushing it, if its made with the quality control VIA, and AMD have become known for.

Just like the Itanium, I bet these chips won't be sub $1,000 a peice, and if they are it will be just barely. So like the Itanium, no one will even want to play with it because of the price except for maybe the rich kids. But then again, aren't they the ones who drive Porsche?

We'll see what happens with price.

the workstation version will be cheap. the server version I think will be 400 - 500$ per CPU

GERMAN TECH MAGAZINE c't has published a set of benchmarks in the latest paper edition of its magazine which some subscribers have received today.

The magazine, in an article by Georg Schnurer, compares a prototype 1.2GHz Athlon64, a 1.2GHz Athlon XP and a Pentium 4 running at 2.2GHz.

Of the benchmarks, the most important noted by our German readers are the Bapco Sysmark 2002, SPECfp base 2000, and SPECint base 2000.

The magazine also publishes a number of comparisons between these three different CPUs using different games.

The Athlon 64 (Clawhammer) is shown as having 1MB of cache L2 cache, the XP 256K, and the 2.2GHz Pentium 4 512K, with the different processors using PC2700, PC2700 and PC2100 memory types in order.

Here's a brief snap of what the paper magazine publishes:

Bapco Sysmark
Athlon 64 178, Athlon XP 145, Pentium 4 217

SPECfp base 2000
Athlon 64 674, Athlon XP 504, Pentium 4 677

SPECint base 2000
Athlon 64 739, Athlon XP 532, Pentium 4 774

Now, with the P4s scaling like crap over 2.5 Ghz from what I hear, and the hammers *supposedly* scaling like champs and being introduced at 2 Ghz, it will be interesting to see what will happen with comparisons then, with the hammer running on a 64b optimized OS and running on a mature chipset.

It will be even more interesting to see what happens with the 2.4 and 2.8 Ghz models that are supposed to follow soon after the 2.0, and what happens when they're running at 200 FSB instead of 166.

I wish they would have run the P4 on something higher than PC2100 ... it invalidates the benchmarks completely, though it does give a small idea of what is happening with the hammers.

Also note: this is the 1MB L2, not the 256K model. I imagine that difference will make THE difference in speed with these CPUs.

Some progress/news:

AMD's Athlon MP 2100+ vs. Intel's Xeon 2.4GHz - Database Server Comparison

AMD Athlon MP 2200+ Server Roundup - September 2002

So... if Anandtech is running with dual AMDs, and AMDMB is doing the same... Why can't ProCooling do it too?:p

AMD's "Venus" SledgeHammer looms into view

AMD forecasts prices for Hammer (Opteron) servers

I hate to hear that my Barton won't be running at 3000+: I guess I'll have to give it a nudge in the right place;) . On the plus side, now I can consider a couple of Bartons on a Tyan board:D

ProCooling.com is P4 powered on the server ( a P4 1.7Ghz)

my home machine is a Dual AMD.

If I had my way everything would be Dual Xeon :)

One of the problems with those test is that a lot of database style work depends on the disk subsystems, different chipsets, different drives, different interfaces etc.. Furthermore, when you speak of web servers as well, networking speeds will play a role as well. All of which should be very obvious, CPU benchmarks should stick to just benches that fairly provide testing methods that do not rely on any other factors other than the CPU.