I can see a lot of people in these forums complaining that, for example, an AMD 3200+ isn't equivalent to a P4 3.06GHz. I understood that the official AMD line was that the 3200+ indicates performance equivalent to the original Athlon running at 3.2GHz NOT ANY of the P3 or P4 series. Is this correct and how close do they come to actually filling this claim?
My opinion is that you are missing the point here. First of all, let me clear up that I am currently using and I've been using for the past 3 years AMD processors. Now about your question. It doesn't really matters if that rating is trully equivalent to what you say since there are no original Athlon CPUs clocked at 3.2GHz, so the comparison is useless. Instead, at those speeds P4s are available so it is quite natural for someone to compare those ratings towards P4 performance. And deffinetely, after the release of Barton core CPUs, those ratings don't justify the performance of newer AMD CPUs, when compared to the P4.
Maybe what you are saying IS correct but that's simply a marketing trick the way I see it from AMD. They could even release a CPU and rate it 5000+ because it would perform like an original Athlon at 5GHz. But if this processor is slower to a 3GHz P4 then its rating has no meaning.
My point was that possibly these ratings, whilst a useful marketing tool, would/should allow a user to reasonably accurately compare processors within the Athlon range. My question was are these rating actually accurate when considered in this way or have they been degraded into a mere marketing tool?
It matters because if I am going to upgrade my Athlon based processor to "the next one up" I want to be sure that I am actually getting a processor with better performance, even if it has a lower clock speed.
Maybe if I REALLY understood how CPUs worked I could do it from a technical spec sheet, but until then I am going to have to make do with those ratings (and obviously the reviews from TH).
Like Pitsi said, before the release of Barton core, Athlon XP PR naming system is pretty much accurate. After the release of Barton core, because of the inferior clock speed, e.g. 1.83GHz for 2500+ is nowhere close to Pentium IV 2.53GHz, the 700MHz difference does matter a lot. So the current situation is if you want to purchase Barton core AMD CPU, don't even bother the PR naming.
OK, if you have to built a new computer from scratch would say get Intel lineup of 'C' core 800MHz FSB CPU and i865/875P chipset mobo for superior performance, the 2.4C~2.8C are on the way, provide you a better bargain. If you ony want to upgrade your existing Athlon XP CPU, a good way is get a 2800+ Thoroughbred-B (or Barton if you want 512k cache but slower clock speed), it's cheaper and good at OC.
depends what youre doing, gaming, the 2500 exceeds the 2.53 ghz in almost everyway with the exception of quake 3 based games
multimedia, like converting MP3's, rendering, clock cycles are pretty much all that matter, athlons are competitive, but intel definetly beats them by a fairly significant margin, so it all depends what you do and what you can afford
personally i thought that the 2500xp and the 2700 xp were rated about right, but the 2600 2800 and above seem a bit iffy, and the 3200 is flat out lying, and even AMD sort of Admits this, i saw a comparison of the 3000, the 3.0c and the 3200 from AMD in gaming performance (a very strong point of AMD) that showed the 3000 at 98% the 3.0C at 100% and the 3200 at 103%, so you tell me? does a 5% percent performance gain justify a 200 point increase? probably not, this thing should have been called the 3025XP
"Going to war without France is like going hunting without an accordion."
- Gen. (Ret) Norman Schwartzkopf
No Koriel , that's not right. The "PR" rating is based on the intel clock speed if it were placed in the real world right now . It is still accurate because most users will shell out the big bucks to get a new rig ,but wont spend the 3 grand to buy the software it needs to be optimised.About 80% of the programs out there right now wouldn't know SSE2 if it bit thier bun.
amds official stance on the pr rating is and always has been that they correspond to an equivalent tbird core athlon.
its become accepted however by most of the hardware community that the pr rating are a guide to how it should perform against an equivalent mhz p4 and this was probably amds intention anyway. the tbred B cores do live up to their pr ratings on average but most people agree that the barton core doesnt.