Where are all the Pentium 4

This Website contains a growing list of CPU-Benchmarkresults done with the free raytracing Program POVRAY. But I'm missing the current P4 CPU's.
Where are they? Everyone is welcome to post a Result.
The Benchmark show's how fast the P4 can be, if the SW is optimized.

Time in seconds to render the Example chess2.pov:

P4 1500 MHz
non optimzed 689 Sec.
optimzed 386 Sec.

You can download the optimzed Version from
Povray Recompilation Experiment

Thanks in advance,
Tilly

I'm not sure exactly what renderer you used, but I downloaded the "Povray for windows 3.1" and used the 4 different renderers found at <http://students.washington.edu/sschmitt/pov/> I get the following results with chess2:

P3_pov: 385 secs.
P3_new: 286 secs.
P4_pov: 378 secs.
P4_new: 244 secs.

Unoptimized (just running the app.): 492 secs.

Wow, I guess those new P4 compiler optimizations really do work. My time was virtually cut in half. Nice System specs in signature.

Thanks...

Thanks Turbo.
Can you tell what kind of Operatingsystem you have installed?

Re: Thanks...

Tilly said:

Thanks Turbo.
Can you tell what kind of Operatingsystem you have installed?

Click to expand...

No problem My bad, win98.

I always thought that the object of CPU-Benchmarking to compare the results of similar systems? ..... However the Pentium 4 results here are obviously done using optimzed software which does not display the real performance of the CPU or allow a real comparision, it only displays how well the software is written and optimized for a specific system.
Is there also a 'specific' version of this software optimized for the AthlonXP ?
imho this is 'similar' to when Intel using optimized MPEG codecs to show the performance of the CPU which are not available to the general public......... lets compare like with like and get the real facts regarding performance out there

I like Sandra's benchmark. Not sure if I want to risk running
a P4 optimized benchmark program.....

As you can see at www.tabsnet.com the Athlon or PIII profit from the optimization in the same way. So I think we can compare the results.

Tilly

I did the "pawns" test they had at the site you suggested. With the P4_pov renderer I got a time of 60 seconds. With the P4_new it only took 37 seconds. Now that's what I call an improvement allright. Even the P3_new only took 50 seconds to complete.

Tilly said:

As you can see at www.tabsnet.com the Athlon or PIII profit from the optimization in the same way. So I think we can compare the results.

Tilly

Click to expand...

One of the websites you quote ( http://students.washington.edu/sschmitt/pov/ ) the following statement is made

" I've been fairly amazed by the attention that our little project has received. The results are fairly clear: all modern x86 CPUs gain a large amount from the use of a good optimizing compiler -- Intel's C compiler. The Pentium4 in particular, gains proportionally more from the recompile effort than other "

I read this as saying that all CPU's gain by optimized code which is a know fact, however it follows by stating that the Pentium 4 gains more from the recompile than other CPU's

the following statement was made later in the page :-

" After the initial publication of the binary and results, I was contacted by an Intel Engineer by the name of Eric Moore, who had a variety of suggestions for improving the compiled performance of povray with the P4. Doing so produced a binary that is 27% faster than the previously optimized version on my Athlon alone, "

I also read this as Intel being involved in the creation of the new optimized codec which also benifited the Athlon.. this could be due to the fact that the code was poor origionaly and even if this was not the case having a CPU benifit from better code is not the same as having code optimized for it, and do you honestly consider Intel were concerned about being fair or the Athlon? .. No of course not, the code created was to enhance the performance of the Pentium 4 it was just luck that the Athlon also obtained gains.

the following statement was made later in the page :-

"Several people have reported that the most recently updated Pentium4 optimized version produces buggy output for some files. "

the reasoning for the new version was stated on the page as being :-

" Here's the binaries that resulted from our experiment with recompiling Pov-ray with Intel C v5.0. This is the fruit of a discussion on Anandtech's BBS about the benefits of recompiling apps for P4,"

The above information suggests that it is a specific Pentium 4 optimized version and I can not find information about any Athlon optimized version which take advantage of AMD specific features, which in turn imho makes the information obtained by this benchmark worthless when comparing Intel to AMD. Once again all this proves is how good the P4 actually is if the whole software market write specifically for it.. A fair comparison it is not

UnseenMenace:

I guess, you misunderstood me. I don't want to say Intel-CPU's are better than AMD-CPU's or AMD-CPU's are better than other CPU's. That is an endless and worthless discussion. I just want too ask for some results of Intel-CPU's. Because there are many current AMD-CPU's at the list and I have no change to compare them with the current Intel-CPU's.

I manage a lot of PC's and SUN Workstation all over the World and I think every System (CPU) have Advantages and Disadvantages. Too find out the right System or CPU it is important to know what you want and need. And, of cause to have enough money for that peace of Hardware .

UnseenMenace wrote:

I always thought that the object of CPU-Benchmarking to compare the results of similar systems? ..... However the Pentium 4 results here are obviously done using optimzed software which does not display the real performance of the CPU or allow a real comparision, it only displays how well the software is written and optimized for a specific system.

Click to expand...

What do you mean with "similar Systems"? Same MB, Graphiccard, Harddisk, etc. ?
What's about the www.spec.org ? Isn't it the same? You can get the Source of the SPECfp2000, SPECint2000, etc. and compile it with all optimisations you can. And I think, that is the only way to compare CPU's of different concepts and manufactures like SPARC, PowerPC, x86, Alpha, MIPS, etc. .
The official Version of POVRAY was compiled with a Watcom compiler and it is 3 Years old.
If we run only that Version to Benchmark a CPU, we just test how well the Code was generated for the CPU's of 1998.
And that is not what I want. I want to see what the new features can do for me.

Example:
If you render a little spot with, let me say, a length of 5 Seconds and 25 Pictures per Second, you need to render 125 Images. And let us say every Picture needs 2 hours (that is not much for a complex Image) with the not optimised Version. It takes 250 hours to render the whole spot. And now, you can save 30% of time by using the optimised Version .