In depth P4/Athlon performance article

[Guest]

For those of you who are really interested in a more in-depth analysis of FPU performance and other issue with these two chips, ACE's Hardware posted this article.
http://www.aceshardware.com/Spades/read.php?article_id=25000191

It's pretty technical, but it's pretty interesting and might give some good discussion points particularly on the impact of the branch prediction unit. But one thing, let's try not to start yet ANOTHER AMD vs Intel flame war over this.

 

[AmdMELTDOWN]

I doubt that the avg. Amd pup will understand this, all they care about NOW! they don't see the P4 as being way ahead of it's time, they really can't see past their freakn noses!

the P4 reminds me of a band called <A HREF="http://www.voivod.com/" target="_new">Voivod</A>, they were too way ahead of their times, these days ppl like to hear simple repeative beats played over and over.

I salute Intel for being "leaders of the pack" in this never ending computer frontier, and I give Amd the finger for being just another copy band.


"Amd cpu...Gone in 2 secs flat, it truly is a fast chip!"<P ID="edit"><FONT SIZE=-1><EM>Edited by AmdMeltdown on 03/08/01 01:57 PM.</EM></FONT></P>

 

[FUGGER]

Nice that someone posts the benchmarks where P4 dominates the Athlon.

http://www.aceshardware.com/Spades/read.php?article_id=25000197

 

[Sojourn]

P4 dominates? Did you <i>read</i> the article? Here are some quotes you must have missed...

It is still possible that new tests with new compilers will prove us wrong, but it very unlikely that any compiler will be able to make the Pentium 4 perform better in x87 code than the Athlon

Today's most used compilers (Visual C++ 5.x/6.x) do not optimize well for modern CPUs like the Athlon and Pentium 4. For some reason, the Pentium 4 will benefit more from better compilers than the Athlon. Or, in other words, the Athlon adapts better to poorly optimized applications than the Pentium 4.

Which means, given perfect operating conditions and completely optimized code, the P4 can make up for its branch prediction latency and poor FPU performance. This is almost never the case, or we wouldn't need the benchmarks, we could just crunch the numbers. Once again, clue in.

/Athlon-1.2GHz@1370MHz(137MHz*10)/Asus_A7V133/

 

[Guest]

Ok, since you like the second one, here is the first of the series.
http://www.aceshardware.com/Spades/read.php?article_id=15000196&alternate=true

Here is a quote from the conclusion: "The Athlon 1200 with DDR is a more balanced and less pricey solution than the Pentium 4 with Rambus. The Athlon 1200 DDR came in first or second place in every benchmark while the Pentium 4 was very capricious with some ups but more downs. Most people hate upgrading their favorite software all the time and the Athlon runs legacy applications faster. IOHO, the FPU of the Pentium 4 should have been made more powerful."

 

[Guest]

AMDMeltdown...

In another article they talk about slower software upgrade cycles in business. Frequent upgrades of software increase the total cost of ownership of the solution in question for the business making the purchase. I (and frankly most of the posters here) am certainly not arguing that The P4 does not have POTENTIAL. I am also NOT arguing that it does not have some areas where the performance surpasses anything else out there. But most people are buying a PC to do things today, not in 12-18 months from now. I also thought that the article was pretty interesting from the standpoint of the engineering sacrifices that Intel made (ie in FPU) and the issue of the impact on the deep pipeline on performance. If their branch predictor is NOT as good as they claim, then they are going to experience more processor stalls and, because of the pipelines depth take more time to recover.