• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

XP 2500+ says its a 1.83 ghz in system properties !?!?!?!

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Originally posted by: redhatlinux
AMd's PR department maybe good, but the PR rating has NEVER been based on a P4. Its based on AMD TBIRD. AMD some how 'scales' the power of the TBIRD and compares it with the power of the XP. IIRC they even have 'independant' auditors to verify the results.
Click to expand...
Redhatlinux, you are 100% correct in theory, but only semi correct in practice. Why?

1) Lets start simple. Suppose you had a 1000 MHz Tbird and you wanted to test a new processor and use the Tbird as a basis. To keep things simple I will start with one benchmark. Suppose the 1000 MHz Tbird got 100 fps in a popular game and the new processor got 150 fps. Clearly the new processor is 1.5 times as fast and it would be quite justified in giving it a PR1500 name.

2) Sadly nothing is that simple. Suppose instead the 1000 MHz Tbird got 100 fps in a popular game and the new processor got 155 fps. Marketing says you need to use round numbers. Do you call it a PR1500 or a PR1600? If this was all that was in the picture, I'd think you'd be safe and round down and just call it a PR1500. But what if it instead got 156 fps, or 157, or 159.9? At what point do you round up? As soon as you round either way it is no longer a perfectly based on the Tbird.

3) Ok now another step in complexity. Suppose two popular games are benched and the 1000 MHz Tbird got 100 fsp in each. Suppose the new processor got 150 fps in one and 180 fps in the other (an average of 165 fps). Now what do you call it? Do you use the lowest number to be really safe and call it a PR1500? Do you use the top number and call it PR1800? Or do you round? But remember marketing won't allow PR1650 so do you round to PR1600 or do you round to PR1700? I can see justification for PR1500, PR 1600, PR1700, and PR1800. Tricky isn't it? None will be a perfect answer, but all are decent.

4) But there is one more factor to consider. Lets use the exact same conditions in #3 above where the 1000 MHz Tbird got 100/100 and the new processor got 150/180. Suppose Intel has a 1.7 GHz P4 that scored 165 fps in each. Not considering the P4, any PR rating from 1500 to 1800 could be justified. But now considering the P4 doesn't the PR1700 seem like the best one to choose? Most people would say so - since the 1.7 GHz P4 and the new processor both averaged 165 fps, they should have the same sounding name. If you agree and use that to name your new processor you will be justified. However, now the specific name you selected is based on the P4!

See how AMD's PR ratings are based BOTH on the Tbird and the P4. Life gets a lot more complicated than what I mentioned though. AMD uses about a dozen benchmarks - giving a dozen possible names. Would the best name be the average of all benchmarks? Or is the best name the median? Isn't the P4 a good processor to help your choice?
 
Originally posted by: Insidious
As we see in many test that overall speed of the 2500+xp would likely better the 2500+ barton in quite a few apps
Click to expand...

I'm not dissin' the post, but if there are actual tests involved... how does the word likely fit into the conversation?

What did the test results say?

Again, this is not meant to be criticism... I truly am curious.

-Sid
Click to expand...


if you look at 2500+, the extra cache does absolutely nothing for gaming and on AVERAGE, it is about 3-4% faster than the throughbred core
 
Originally posted by: shady06
Originally posted by: Insidious
As we see in many test that overall speed of the 2500+xp would likely better the 2500+ barton in quite a few apps
Click to expand...

I'm not dissin' the post, but if there are actual tests involved... how does the word likely fit into the conversation?

What did the test results say?

Again, this is not meant to be criticism... I truly am curious.

-Sid
Click to expand...


if you look at 2500+, the extra cache does absolutely nothing for gaming and on AVERAGE, it is about 3-4% faster than the throughbred core
Click to expand...

It's not a matter of disbelief... I really would like to look.... can you linkify me?

 
Originally posted by: Superman9534
um, nobody has mentioned this, but his 2500+ IS underclocked. 1.83ghz is the speed of a 2200+, and unless he has a barton, its underclocked and should be over 2ghz...Try setting the FSB to 166 instead of 133.
Click to expand...
the first person to answer the original question...
 
I said likely cause I can't for certain they even made a 2500+xp...I know there is a 2600+ and 2800+ becuase of the famous paper launching of last fall, but I don't know if they filled in the gaps....


Here is a link and look at the 2800+ beat the 3200+...and the 2400+xp beat the 2500+ barton...One could deduct from that, right???

Anandtech's 3.2ghz review 3D Rendering

Aginin in Lightwave the 2400+ beat the 2500+ with a 266fsb and less cache....

Anadtech's 3.2ghz review Lightwave 7.5

in this example at Toms look at cpu dependent app like mp3 encoding the 2600+xp beats the 2800+ Barton chip, and the 2700+xp with the 333fsb beats the Barton 2800+ in Divx encoding....The 2800+ not to far ahead of the 266fsb 2600+xp as well

Toms 3.2ghz review MP3 maker

here once again the 2700+xp beats the 2800+ barton and even hangs neck and neck with the 3000+ Barton which has same clock speed but the added stuff gives it less then 1% lead!!! Notice how the 2600+ 266fsb lays the smack down on the 2500+ Barton...NO 2600+barton woudl have hanged with the 2600+xp...

Toms 3.2ghz review MPeg2 encoding


I know these are not gaming apps but they are examples of real world apps ppl use with their desktop pcs...Not everyone games!!!


MORE!!!

Here the 2800+ beats the 3000+ Barton and the 2700+xp is right there in Windows Media Encoder 9....

Anandtech's 3.0ghz Review WM9

Again in another mp3 encoder Lame and another Divx test we see the 2600+ Tbred is about even with the 2800+ Barton and lays the smackdown on the 2500+ Barton leading you to deduct no 2600+ or 2700+ barton (if there is one) would beat it...Once again mhz wins and AMD pr ratings show the kinks in the armor...

The Tech Report Review of P4c's Audio/Video
 
Doh. Have not been reading much into your products have you? Or this might be just another sad stunt to show how AMD PR ratings are.
 
Gage8: I see quite a few people answering the question CORRECTLY rather than answering with something that is incorrect.

Windows does not report your processor name based on the speed. For that matter, the BIOS doesn't anymore. They report both the "name" that is written to the chip ID information, i.e., XP 2500+ or Pentium 4 2.4GHz, as well as the actual frequency, in this case 1.83GHz, or for the P4 it'd be the same as the ID. If you overclock, the BIOS and OS still report the stock ID information, but then also report the actual speed. It used to be different of course, the BIOS would report the name based on the speed, but then both AMD and Intel made it so they report the stock ID (and AMD would prefer the actual frequency never be displayed, it was one of the original specifications for Athlon XP BIOSes).

So if you overclock a Barton 2500+ to 2200MHz, the speed of a 3200+ Barton, the ID would still show 2500+ but then you'd also see the actual frequency.

In this case, his Barton is running at the correct speed of 1.83GHz, using a 166MHz bus with an 11X multiplier. If it was running on a 133MHz bus with a 13.5X multplier to get 1.83GHz and was switched to a 166MHz bus, it'd result in over 2200MHz, faster than the fastest available Barton (of course the bus would be slower than the 3200+).


The inanity of the PR system is obvious when you look at the current top of the line. From the Barton 2500+ to 2800+ is a 253MHz frequency difference, with the same bus speed, getting a 300 point higher rating. To the 3000+ with the same bus, the frequency is only 81MHz higher, but gets a 200 point increase. The rating increases by 2/3 compared to the previous jump, but the frequency only increases 1/3.

Moving to the 3200+, with a 400MHz bus speed, the frequency of the CPU only goes up 33MHz. Again less than half the frequency increase from the last rating, but a 20% bus speed increase, netting a 200 point PR increase.

Comparing to Thoroughbred: the closest to a 2500+ is the 2200+ Tbred with 133MHz bus. Only a 33MHz speed difference, but the extra cache and bus speed warrants a 300 point increase even though the cache only specifically helps certain applications.

So, the Tbred 2600+, the lowest with a 166MHz bus. 2.08GHz, comparable to the Barton 2800+. Same bus speed, same CPU speed, the increase in cache is accounting for a 200 point PR difference. The 2600+ of course has a 400 point lead over the 2200+, with only a 250MHz difference and a faster bus. Then to the 2700+, an 83MHz speed increase, netting only a 100 point PR increase. Surprisingly enough, the next rating of 2800+ is also an 83MHz increase, so it's somewhat believeable. Except that Barton is 166MHz slower but has the same rating just due to cache that isn't helpful at all times.

The real issue is that AMD's ratings only apply in certain cases. In other applications, the rating is stupidly off. They naturally choose applications that make it look good, and say that they're adjusting the rating based on the overall performance not the peak performance in certain apps, but of course they're going to make it look as good as they can.

dullard: one thing to remember, performance doesn't scale linearly with clock speed. So an XP that got 150fps compared to a tbird 1000 getting 100fps would actually be performing like a tbird 1.8GHz or higher. 🙂 Just increasing the clock speed of a tbird 50% wouldn't result in 50% better performance.

But despite everything, it's still just plain stupid to claim to be comparing to such an old chip, with different bus speeds, higher frequencies, more cache, more optimized pipelines, et cetera. How exactly do they come up with numbers for what a tbird would perform like if it ran at 2.5GHz? And even if they could get an accurate estimate of that, what is the point of comparing their current processor to their old ones? Do they think consumers are actually looking at it and comparing it to previous AMD processors? The tagline for the XP line should then be "faster than our old technology!". Comparisons to a previous technology is trying to make a comparison in a closed system that doesn't involve anyone else, and becomes an arbitrary number. In order to give it any validity, it has to have something external to compare to, and it has that in the P4, but AMD just can't say that.
 
Originally posted by: ElFenix
i wonder if the rating is still lined up with a t-bird, like it was supposed to be
Click to expand...
If so, that would mean that a P4 is just as fast as a T-bird, clock-for-clock. And in many cases lately, significantly faster.

If I tried to sell that the P4 is clock-for-clock as fast (or faster) than a T-bird, I'm pretty sure you guys wouldn't be buying. 😉
 
Just to make things fun, here's an app that I actually have an older version of: trueSpace 5.1 benchmark at LostCircuits I have version 4.3. This app was the particular reason I've had three dualies in the past, and the benchmark clearly shows why real SMP still has its place. Nothing like watching a render job crawl to make you want all the CPU power you can lay your hands on! 😉
 
Originally posted by: Lord Evermore
Gage8: I see quite a few people answering the question CORRECTLY rather than answering with something that is incorrect.

Windows does not report your processor name based on the speed. For that matter, the BIOS doesn't anymore. They report both the "name" that is written to the chip ID information, i.e., XP 2500+ or Pentium 4 2.4GHz, as well as the actual frequency, in this case 1.83GHz, or for the P4 it'd be the same as the ID. If you overclock, the BIOS and OS still report the stock ID information, but then also report the actual speed. It used to be different of course, the BIOS would report the name based on the speed, but then both AMD and Intel made it so they report the stock ID (and AMD would prefer the actual frequency never be displayed, it was one of the original specifications for Athlon XP BIOSes).

So if you overclock a Barton 2500+ to 2200MHz, the speed of a 3200+ Barton, the ID would still show 2500+ but then you'd also see the actual frequency.
Click to expand...

In the case of the P4, yes, you are correct. Mine reports a P4 2.4GHz, with a clockspeed of 2.93GHz.

In the case of my Palamino and TBred Athlon XP's, then you are incorrect. My TBred B 1700+ (1.47GHz) is running at 2.08GHz, XP2600+ levels, and Windows XP shows it as a XP2600+ running at 2.08GHz, not a 1700+ running at 2.08GHz. The same is true for my Palamino core XP1800+ (1.53GHz), when I overclocked that to 1.63GHz (XP2000+ levels) it shows as a 2000+ rather than an 1800+.

The Barton cores, may do the same as the P4, but I have not seen them myself, so am unable to comment on that. I presume that the same occurs as does with the Palamino/Tbred.


Also, the XP's are based on a suite of about 12 tests, to give a PR value compared to a TBird. As others have mentioned, a "rounded" figure may lay about the same as a just released P4, but then that is marketing. The race to 1GHz got a lot of people talking, and back then, AMD were "GHz matters" as they beat Intel to a widely available 1GHz CPU. Now that the P4 has ramped up in clock speed, they need to be seen to be keeping up, or if they can, getting ahead, by justifying their PR values according to where the P4's are.

However, I think that they are putting too much emphasis on the L2 cache of the Barton cores, and should have kept the clock speed higher also, and gain a lead over Intel, rather than appearing to fall behind in peformance due to their over-rated PR values.


Confused
 
Maybe a fair solution would be for AMD to benchmark the P4's and give them PR ratings too 🙂 Call me crazy...
 
Originally posted by: Lord Evermoredullard: one thing to remember, performance doesn't scale linearly with clock speed. So an XP that got 150fps compared to a tbird 1000 getting 100fps would actually be performing like a tbird 1.8GHz or higher. 🙂 Just increasing the clock speed of a tbird 50% wouldn't result in 50% better performance.
Click to expand...
Yes things are a whole lot more complicated than what I posted above. Because performance doesn't scale linearly with clock speed, are they basing the PR on the slower Tbirds or the faster Tbirds? That is a very interesting thought if you sit down and list the consequences.

There is also a whole slew of other questions to be asked as well. The biggest is: why does the CPU rating include motherboard and video card efficiencies as well?
 
Originally posted by: Gage8
Originally posted by: Superman9534
um, nobody has mentioned this, but his 2500+ IS underclocked. 1.83ghz is the speed of a 2200+, and unless he has a barton, its underclocked and should be over 2ghz...Try setting the FSB to 166 instead of 133.
Click to expand...
the first person to answer the original question...
Click to expand...

Actually, that would be correct if he didnt have a Barton. But he has. So we have almost all been answering the original question mate.
 
Weird. My Palomino 2100+ always reads as a 2100+ in both Windows and the BIOS, and then lists the actual clock speed. Even when it's set to an overclock just enough to reach 2200+ speeds.
 
BWAHAHAAHA.

http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_756_3734%5E3876,00.html

Q: What specific applications benefit from the performance of the AMD Athlon XP processor 3200+?

A: The real benefit of the QuantiSpeed architecture that powers the AMD Athlon XP processor 3200+ is that it can deliver better performance on a broad range of applications. AMD?s customers demand that all their applications perform well, including office productivity, digital media and 3-D gaming. For benchmarks and other specific information please see:http://www.amd.com/products/cpg/athlon.
Click to expand...

So...the benefit of making up a number to call the processor other than the actual speed means it can perform better. Better than what? When do we get flame-stripes on the CPU package?

Q: How do you expect the AMD Athlon XP processor 3200+ to perform against the 3.0GHz Intel Pentium® 4 processor with an 800 FSB?

A: The AMD Athlon XP processor 3200+ is the world?s highest-performing PC processor on a variety of real-world, industry-standard benchmarks for office productivity, digital media and 3-D gaming. It outperforms its closest competitor by an average of 6 percent on a variety of industry-standard benchmarks.
Click to expand...

Hmmm...3GHz + 200MHz = 6.67% increase in speed. AMD thinks that a P4 will get a near linear performance increase from increased clock speed. Why didn't they call it a 3400+, since it would most likely take more than 6% clock speed increase to get 6% performance increase? Oh yeah, because they know it's a farce and nobody would even take them the lease bit seriously.

Q: What does the 3200+ model mean?

A: This is a model number. AMD identifies the AMD Athlon XP processor using model numbers, as opposed to megahertz. Model numbers are designed to communicate the relative application performance among the various AMD Athlon XP processors. As additional evidence that performance is not based on megahertz alone: the AMD Athlon XP processor 3200+ operates at a frequency of 2.2GHz yet can outperform an Intel Pentium® 4 processor operating at 3.0GHz with an 800 FSB and HyperThreading on a broad array of real-world applications for office productivity, digital media and 3-D gaming.
Click to expand...

Hmmm, so, now the model numbers are only a relative number amongst all the XP line? Convenient how they just happen to increase at a rate that keeps up with the P4 speeds, and adjustments are made to keep the rating up without actually increasing the processor speed. It's also nice that they're willing to compare it to a P4. Especially when you imagine that the XP isn't going to get much faster but a P4 could be ramped up to 4GHz.
 
hmm... lots of posts in this thred. Remember, when AMD started the PR thing, Willy P4s were soooo slow, but with high clockspeeds. At the time I thought the PR rating was pretty fair. I guess AMD could just say 'We spent ALL our development $'s on 64 bit, our 32-bit line is soooo old now so we're just giving up'. Not likely. The reality is that Athlon IS old, the short 'fat' pipe just won't clock at the same rate as the long 'thin' pipe of a P4. What do I think of the PR ratings today, bogus, but without the competition from AMD, Intel would be asking a lot more for P4's.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top