Dom

A hardware web-site x86-secret.com has revealed a formula describing performance rating of AMD Athlon 64 processors. If you still do not understand how a 1.60GHz Athlon 64 processor is brand-named as Athlon 64 2800+, then the formula will show you how such things may happen.

Apparently, performance rating of AMD 64-bit desktop processors looks as follows:

Performance Rating = (CPU Frequency x 3)/2 + 400
e.g. 2800+ = (1600MHz x 3)/2 + 400

The first Athlon 64 processors to appear this year will be models 3100+ and 3400+ with 1.80 and 2.0GHz core-clock respectively and 1MB of L2 cache. As we said earlier, in Q1 next year AMD will launch the last and final Athlon 64 processor on ClawHammer core with 2.20GHz frequency and 3700+ performance rating. Additionally, the company will offer a 0.13 micron Paris processor with 256KB of L2, 2.0GHz core-clock and 3100+ performance rating. You can guess that AMD will use another performance rating formula for its value 64-bit desktop chips what is a rather positive sign.

AMD Athlon 64 CPUs made using 90nm fabrication process are now set to come starting from the second quarter 2004. The expensive higher-end 2.40GHz (4000+) and 2.60GHz (4300+) Athlon 64 (San Diego) CPUs will emerge in the second and the third quarters respectively. Also in the second quarter next year AMD will issue bunch of new 90nm mainstream/value 64-bit chips based on Victoria core. 2.0GHz (3100+), 2.20GHz (3400+), and 2.40GHz (3700+) models will come straight away and 2.60GHz (4000+) processor will be available in the third quarter of 2004.

In fact, AMD Athlon 64 “ClawHammer” is not really successful for AMD. There will be only three models of such CPUs on the market and they will be quite expensive. As a matter of fact they will hardly become massive. Furthermore, as I have learnt by now, only one PC3200 memory module will be supported by AMD Athlon 64 ClawHammer. It means you will hardly be able to install even a Gigabyte of RAM into a personal computer based on the first ever 64-bit desktop CPU from AMD. The Athlon 64 processors coming out this year will have 1.50V Vcore and up to 89W Maximum Thermal Power. Hot babes, aren’t they? The 90nm San Diego Athlon 64 processors should be less hot in terms of temperature and support more PC3200 memory.

Manufacturing of AMD Athlon 64 3100+ and 3400+ is set to begin on the 14th of June and on the 16th of July respectively.

Watch out, they are coming here!

______________________
Source: X-BitLabs

WaltC

I must be missing something here...this just looks like somebody found out (or thinks) that a 1.6GHz A64 will carry the 2800 + performance rating, and simply worked backwards from that information to come up with this "formula"--which explains nothing about why AMD thinks consumers will compare, and should compare, an A64 @ 1.6GHz to a P4 running at 2.8GHz (Which is really the only commercial value the PR rating has for AMD.)

I rather think it's because AMD's tests and benchmarking show a performance equivalency between them. Seems logical to me...

Ah, xBit....wasn't it just a few days ago they revealed ATi would soon stop making ATi-branded products? (Which ATi quickly denied.)

But this does raise an interesting point that it may be harder for AMD to publish a comparative performance rating for A64, because what specifically do you compare it to? Xeon, P4, Madison, or Intel's upcoming .09 micron P4's (which are rumored to ship some time after the A64)?

mike2h

their performance rating is already in the toilet. their 3200 can not match even a p43000c. & it costs more. i think waltc is right about the way they figured out that rating. jmo

__________________

AlternateVirus

omg, dudes (waltc and mike2h, THE PR system IS NOT MEANT TO COMPARE TO INTEL PROCESSORS AT SAME SPEED. its to compare thunderbird at same speed. 3200+ is meant it should perform just like thunderbird at 3.2ghz

JavaFox

That's what they SAY, Alternate, but it's not true. A 1.533GHz Palomino (1800+) and a 1.533GHz TBird perform almost identically. OTOH, a 1600+ and a 1.6GHz P4 ARE pretty close. With the exception of the recent Northwood-Cs, AMD has been pretty much on the money with their performance ratings.

If a 3.2GHz TBird and a 3200+ Thoroughbred were pitted against each other, the TBird would outperform the TBred by an absolutely fantastic margin.

Actually, a 3.2GHz TBird would probably look like this, though..

WaltC

I know that's what AMD maintains, but think about the fact that people shopping for these cpus don't know anything about a T-Bird (been discontinued for awhile) , and that what they are comparing it to is the competition--in this case the P4 versus the AXP. The PR rating really would have no meaning at all if AMD intended it to relate to the T-bird--it would be useless since customers would have no basis of comparison with Intel chips.

Intel has similar problems--Centrino runs at a slower MHz than P4 and so does Itanic...but Itanic is definitely much faster than P4 and Centrino should do better than a Celeron. If Intel doesn't get onboard with a PR of its own they'll have problems, especially with Itanic...

But the A64 is fully SSE2 supporting and will do 64-bit calculations as well--so what does AMD compare it to? My guess is still the P4, which won't do 64-bit processing at all (which is fair because Athlon XP doesn't do SSE2.)

Ideally Intel and AMD would get together and establish a single PR standard for all of their products based on *something*--I wouldn't care if it was a T-Bird or a P4. But this is the only way consumers will readily grasp what's generally what in the performance arena.