Intel Optimises Tom's MPEG4 Benchmarks

[BFG10K]

Tom has just put up his third installment of the Pentium 4 review. Intel engineers spent all night optimising the test bench for the Pentium 4 and have now recompiled it. Needless to say the Pentium 4 is doing really well now.

What's interesting to note is that when running the new version of the program the Pentium 3 and Athlon scores were boosted as well, suggesting the original benchmark was written in a crappy way.

Tom's Hardware

 

[GL]

Only Tom could get Intel to release an optimized version of a tool which many people are using to rip copyright material off DVDs I luv it...now where can I get my hands on this version of the software? I wonder how much Intel hates Tom?

BTW isn't Tom violating the GNU license in some way...shouldn't the code be available for download since the original work was covered by the GNU?

-GL

 

[Leo V]

Gotta give it to Tom, he has really improved his journalism since the days of Q3demo benchmarking. His quick reaction to Intel's new code--in spite of the resulting misspellings--speaks volumes about Tomshardware's integrity in the face of politics. Not to mention Tom's dedication to his audience in spite of powerful faschists like the MPAA...

 

[GL]

It's not new...I've been following Tom since the days of his monster truck madness benchmarking and telling everybody about the Asus T2P4's 75 and 83MHz bus settings. He's always first out with a new slant on an issue.

-GL

 

[Leo V]

GL, I don't think there are any GPL issues. If Tom isn't distributing the program, he doesn't owe anyone the code.

 

[GL]

That Intel guy gave Tom the binaries...he must also give Tom the source right? Intel would kill this 'Alex' guy if the source was release though...I understand why they'd be pissed. The P4 in fast enough versions could rip DVDs in realtime How's that for a selling point.

-GL

 

[Ben]

Interesting. I like this sentence


<< For the time being Intel is in the same situation AMD used to be with K6-2 and 3D Now! Without proper support of the software industry it will be hard to make a product such as Pentium 4 successful. >>


Now there's a role reversal.

 

[AngelOfDeath]

Well bla bla bla now the P4 can decode a DVD. Woohooo I'm gonna buy meself a P4 now...Ahhh shut up. I don't see the point in that bench since it's still a horrible cpu at the moment.

Stupid article totally unneccesary :|

AoD

 

[JimMc]

Far too many people have rushed to judgment on the P4. Due to it's very different core and SSEII, it will be some time before we have a good idea of how well it can perform, we already know it doesn't do particularly well on PIII/Athlon oriented benchmarks, time will tell on the P4.

My guess is the P4 will do very well on optimized benchmarks, and these optimizations will eventually start showing up in everyday software. How well the P4 will compete with future offerings from AMD, particularly on a price/performance basis, is less clear, particularly once AMD has implemented SSEII.

 

[LocutusX]

This doesn't surprise me at all. Flask is a poorly optimized program in the first place. There are already MPEG2-to-MPEG4 encoding tools that can perform at real-time on MY machine. (Pentium 3 600E @ 810).

 

[xtreme2k]

I must say the P4 today isnt a good solution, due to POOR optimisation and performance. But I think this WILL change in the future. This Flask optimisation is one perfect example of what we are missing. Most of us are being too short sighted when saying the P4 is rubbish. I agree there is NO reason to buy a P4 now, but it might not be so slow when programs started to optimise for it, and the MHz goes up.

 

[Kwad Guy]

So when is this new version of FlaskMPEG (or equivalent) going
to be made available to the world at large. I imagine there's
now going to be GREAT pressure/demand to optimize the program...

Wow...That'll be great...I am already getting 8fps on my dual
processor machine. Looks like I'd be able to get within shouting
distance of realtime with the Intel mods...! (Well, at least 16fps,
anyway).

Kwad

 

[Mday]

i like sse2. but i don't like the p4.

i don't even like the p3, but i like sse.

 

[Venomous]

Well gee.. why dont we put up 4 or 5 more IMPROVED versions to suit the P-4. Thats pretty sorry man.. Need optimization to run decent. Look at the Athlon, it has very very little opts and still hangs with the big boys. Without opts, Intel has no merit. Pretty pathetic.

 

[Wingznut]

VERY well put, xtreme.

The scene is changing, and you guys don't see it. Soon software, benchmarks, etc, will be optimized for the P4, and you'll see how powerful the P4 really is.

Sure, it's not so special on yesterday's applications... But just be patient and you'll see.

 

[Soccerman]

I'd like to add a comment..

wouldn't it be nice if AMD's Athlon could take more advantage of 3DNow!, and SSE 1 and 2?

it appears the Athlon core isn't made to work well with SIMD instructions like 3DNow and the like. it would be nice if AMD incorperated SSE into it though!

we barely know anything about the sledghammer core don't we?? I gotta know now!

 

[Hulk]

I think Tom did a great job with the original article and the follow ups. All are very impartial, he's just reporting the facts. And I like the fact that he made Intel scramble!

Don't admonish the bearer of the news. Just read it and make your own judgements.

I only support faster and cheaper processors. I really don't care who makes them. Make sure you can make that claim before you start making comparisions.

 

[OneEng]

Yes, Yes, Yes, there will soon be optimised benchmarks that make P4 look better like the one at Tom's. It will be quite some time (about a year) before there is signifigant support in released software.

P4 in this date and time is rather poor all the way around (except for bandwidth). No one is going to go out and purchase all new software to make their new computer perform better.

How many people benefit from a benchmark program that runs quickly?

 

[BFG10K]

A lot of those old programs might not even have to be re-written at all. All it might take is a simple recompilation by one of Intel's new compilers. Apparently Intel's newest compilers are some of the best in existance at extracting parallelism from normal source code and converting it to SSE/SSE2 style code.

 

[DeathroweR]

-
I'm not sure if you got the original point, made public long before P4's release:

Intel wants to REPLACE x87 with SSE2.

Intel sais x87 must die.

 

[erub]

If optimizing software for SSE2 is this easy, and Microsoft wants some good publicity, wouldn't it be nice to see them recompile Windows 2000/ME for SSE2? A little windows update (ok, about 100 mb probably), and wham, 3x the speed.

 

[DeathroweR]

Windows doesn't use FPU that much, it's mostly integer calculations, AFAIK.

 

[LocutusX]

The bottleneck in a &quot;mainstream Windows setup&quot; isn't the processor anyways, and hasn't been for quite some time now.

 

[JeffreyLebowski]

LocutsX


<< There are already MPEG2-to-MPEG4 encoding tools that can perform at real-time on MY machine >>



What and where do we get these encoding tools?


We damaned freeware!

 

[Floyd]

To the best of my knowledge, MPEG2AVI is the fastest MPEG2->MPEG4 encoder. On my Celeron 850MHz it encodes at approx 18fps whereas a code-optimized derivative of Flask (PX3) encodes at only 8fps, which amounts to a 225% improvement. The differential is probably even larger for the standard FlaskMPEG release.

Best regards,
Floyd