Do you get an 'XP rating' applied when you o/c?

[eldren]

Originally posted by Austin
Well it does show the XP2500+ as XP3200+ if you switch from 11x166 to 11x200 but that's all.

My Gigabyte 7NNXP detects my JIUHB (1700+ for those that don't know) as a XP2700+ when clocked at 200x11 (At 200x11.5, it gives the speed). Odd. Unless it correctly factors in the Thoroughbred/Barton differences in P-rating.

Either way, I think AMD should do away with their equivalency ratings. Only way it would happen, though, is with the mass re-education of all those people that just look at processor speed as a true indication of a PC's speed. Kinda like the P133 vs. the P166 - in 99.999% of cases, you'd never notice the difference.

Why go for higher processor speeds then? Well, I do enjoy the ability to encode mp3s at 20x.

[Austin]

Unfort there's just too many people who think the ghz is all important and would never dream an AthlonXP @ 2.2ghz (XP3200+) could beat a P4 3ghz in practically everything. Using the rating system saved AMD's bacon, it's a big shame we don't keep it when we o/c with all mobos. AMD have been (reportedly) looking into some kind of benchmark suite to give a PC an overall perf rating making things much better for the consumer (if it is taken on as an industry std and people don't 'cheat'). That would avoid all those mobos using super CPUs but with slow RAM, gfx etc. However it was supposed to be finalised long before Athlon64 came around ... guess it fell through. Big shame.

I would say there was a very noticable diff between P133 and P166 (at the time), however when it came to things like P133 (66mhz FSB) vs P150 (50mhz FSB) things were definitely very grey. MMX was a big thing when that was introduced ... many people didn't grasp that the P166mmx was faster in almost everything than the P200 for example.

[eldren]

Originally posted by Austin
I would say there was a very noticable diff between P133 and P166 (at the time)...

I meant P133 vs. P166 non-MMX. Should have been more specific about that.

[dgr]

my 2800+ barton reads a 3200+ at 11*200...

[hotpurple]

Originally posted by Flanderz
How does the XP rating work anyways?

Or does it mean to take, The Pentium 4 equivalent in mhz? Eg the XP 3200 is equal to a P4 3.2GHz. Thats the way I have always understood it. I'm up for correction ...

It's supposed to mean the equivalent speed a thunderbird (non xp athlon in flip chip design) would have to go to provide the same performance.

Back to the question though, I've got an asus A7N8X Dlx and I get the xp rating on the start up screen.

Chris

[hotpurple]

Actually, I'd agree with austin on the P133 vs P166 argument. The P166 was significantly faster in most things, especially in games.

Chris

[Austin]

The XP rating in reality does indicate the AthlonXP's perf relative to P4 although to avoid any legal strifes AMD claim it's all to do with the Thunderbird Athlons LOL!

eldren meant to say that mmx was more important than pure clock speed rather than the P133 vs P166. As in the 133mmx would be faster than the P166 (although the mmx started from 166mhz IIRC).

[hotpurple]

mmx itself was a bit of a let down really. It only made any difference on integer calculations so it was pretty useless for most games. The reason the mmx processors were faster than the non-mmx processors was that intel added more l1 cache (32k instead of 16k iirc).

Chris

[Alex]

So is an XP2500+ @ 2.2GHz actually a faster processor than a P4 3GHz running at stock? Or are the other factors involved?

[hotpurple]

not really, the xp ratings are getting completely wrong.

An xp2500+ at 2.2GHz is really an XP3200+, which gets trounced by the P4 3.0GHz in most situations.

EDIT: Realised I'd not taken into account overclocking

[Austin]

The XP rating is alittle generous to the Barton core and when using 400FSB but it's still very accurate when comparing the AtlonXP to a 533FSB P4. The AthlonXP3200+ 2.2ghz 400FSB (inc XP2500+ at this speed) is quite often faster than the P4 3.0ghz 800FSB and generally a fair deal faster than the P4 3.06ghz 533FSB. A lot depends upon specifics but it is a very good example of why ghz has not been a good indication of PC or even CPU perf. Just look at what the Athlon64 can do at 2.0-2.2ghz and that's without 64bit OS, apps or optimisations!

[Lexeus]

My MSI K7N2 Delta-ILSR shows the XP rating !
+! point for MSI

[Saracen]

Originally posted by Austin
The XP rating in reality does indicate the AthlonXP's perf relative to P4 although to avoid any legal strifes AMD claim it's all to do with the Thunderbird Athlons LOL!

And presumably they fake all the benchmark figures that they derive the PR ratings from as well, then?

[Austin]

Originally posted by Saracen
And presumably they fake all the benchmark figures that they derive the PR ratings from as well, then?

Are you seriously trying to suggest that the 'XP rating' is remotely linked to the old Thunderbird Athlon? The fact that the XP rating ties in almost exactly with the relative P4 at the time is just a coincidence? If you truly think that I'd have to say you're in a very small minority (who can always be right of course).

Please show us any figures that show how much faster the AthlonXP is compared to the original Athlon. Then show how the AthlonXP fairs against the P4's out at that time (400FSB P4 to early AthlonXP and then 533FSB P4 to the rest). Which would you say is closer?

[dgr]

austin - maybe true for some cpus, but i would say most CERTAINLY not with the 3000 and 3200+...

dgr

[Saracen]

Originally posted by Austin
Are you seriously trying to suggest that the 'XP rating' is remotely linked to the old Thunderbird Athlon? The fact that the XP rating ties in almost exactly with the relative P4 at the time is just a coincidence? If you truly think that I'd have to say you're in a very small minority (who can always be right of course).

Please show us any figures that show how much faster the AthlonXP is compared to the original Athlon. Then show how the AthlonXP fairs against the P4's out at that time (400FSB P4 to early AthlonXP and then 533FSB P4 to the rest). Which would you say is closer?

Yes, that's exactly what I'm saying.

The PR rating is based on an amalgamation of a series of benchmarking results. If I remember correctly, AMD ran about a dozen industry standard benchmarks and then combined them into one overall rating. I sat down in a meeting with AMD's European Marketing Manager (Richard Baker) and went through these figures. The benchmarks used, the specification of the machines on which they were run, together with the exact timings achieved by each of the individual benchmarks, AND the method for combining them to reach the PR rating were all published on AMD's website at the time, for anybody that cared to go and look. They may well still be there for all I know.

I'm certainly not about to say that the PR rating is perfect, because it isn't. One of the problems is that it is an artificial measure based on a scale relative to what is now quite an old chip, and the process does not scale especially well. Personally, I think we are pushing it seriously hard when we get above 3000, and possibly before. The one thing it has in its favour is that it is still, IMHO, a better basis for comparing CPU's than megahertz is.

The PR rating was at least in part, and probably in LARGE part, a marketing exercise. AMD had a problem. Their CPU's real-world performance against competition, either from Intel or from generations of their own CPU's, was not at all adequately reflected by megahertz, yet Joe Public tended to buy believing that 'megahertz is king'. He was likely to walk into a shop, see an 1800Mhz P4 and a 1500Mhz AMD and ASSUME that the P4 was necessarily faster, whether that was true or not. This very dubious assumption was playing right into Intel's hands and AMD were not happy about it.

AMD's overt intent in the PR rating was to get people to be aware that this was not the true situation. When Palomino technology was released, THAT made the point beautifully. A Palomino Athlon (XP) was quite a bit faster than the old Thunderbird Athlon running at the same clock speed, simply because the CPU internals were better. Therefore you cannot rely on megahertz as a simple rating of comparative CPU performance if the underlying engines were different. The days when AMD chips were effectively the same chip as Intel have gone, and relying on megahertz and an accurate guide to comparative CPU performance went with it.


What the industry needs is an objective, independent test procedure that has the support of both manufacturers, but a well-respected independent third party doing the tests and publishing the results. At the moment, we aren't likely to get it simply because relying on megahertz STILL reflects well on Intel, so why would they want to change?

These arguments are pretty well known on technical bulletin boards, and among technically savvy users, but the VAST majority on consumers are not that technically savvy. They are going to walk into PC World (or similar) to buy a machine for their kids to do homework on, play a game or two, and maybe for mummy and daddy to be able to web-browse or collect their email, and ASSUME that bigger megahertz is better. Breaking that fixation is what PR ratings are all about. They are NOT aimed at the people that read hardware forums like this. People here are very likely to have seen individual benchmark tests (or go and route them out) by Ziff Davis Labs, or whoever, and be aware that megahertz is not a good guide to comparative performance, and even that certain makes of chip are good, or bad, at specific types of real-word job ...... and buy accordingly.

I sat in a room when Richard Baker specifically and explicitly stated that the PR rating was not a comparison with P4's, and then went through the benchmark figures in some detail to prove it. Was he lying? Were the figures all faked?