CUDA Factorial Benchmark from TOC.ru

[[XC] riptide]

http://total-oc.ru/download.php?id=101



Show me what you got. Interested in high end NVIDIAs. Remember... shader and on card memory clocks are essential.

[Chumbucket843]

you might want the problem size to be bigger, just a thought. i know you already have seen my results. im just trying to get this thread goin. i wonder if anyone on XS has access to a tesla system. memory is at 1200 and shaders are at 1500.

[[XC] riptide]

Its fine. You can do any problem size you want. Your GFX is a 260? Is that 216 shaders?

[poke349]

I wonder what algorithm it uses...

750000! can be done well under a second with Mathematica 6.0 single-threaded on any i7.

Yet it takes 1+ minutes using CUDA?



*Sorry for being off topic.

[[XC] riptide]

I wonder what algorithm it uses...

750000! can be done well under a second with Mathematica 6.0 single-threaded on any i7.

Yet it takes 1+ minutes using CUDA?



*Sorry for being off topic.

Have you got mathematica? Can you show us the result? We already know there are insanely fast algorithms and approximations that can be used http://www.luschny.de/math/factorial/Benchmark.html .... but we also know there are many algorithms for calculation Pi to 1million places.. all with diff times.

[poke349]

Have you got mathematica? Can you show us the result? We already know there are insanely fast algorithms and approximations that can be used http://www.luschny.de/math/factorial/Benchmark.html .... but we also know there are many algorithms for calculation Pi to 1million places.. all with diff times.

Timing[750000!;]
{0.405, Null}

i7 920 @ 3.5 GHz

Mathematica can't multi-thread high-precision arithmetic. So there's no way see how much better it can go.


On the other hand... I can mod the program in my siggy to do multi-threaded factorials (using a sub-optimal algorithm)... And I'm certain I can beat 0.405 seconds.

But again... off topic.

[[XC] riptide]

Timing[750000!;]
{0.405, Null}

i7 920 @ 3.5 GHz

Can you export all the digits? I mean does it deliver all the digits? There should be 4080578 digits.

[poke349]

Can you export all the digits? I mean does it deliver all the digits? There should be 4080578 digits.

That run of 0.405 seconds did only binary digits. Printing it out in decimal requires an expensive conversion.

750000!
\!\(\*
TagBox[
RowBox[{"2646896442810456334473283390526976189442958803731348335812907\
9334567747113504796887022327350144664381155203676817108918748679291696\
6443372148573575453227479621798163102781469763477812875007762400556456\
3838296982600913849826449820515029294880777450379489322119687361868491\
51503071358153700424169800424565", 
RowBox[{"<<", "4079973", ">>"}], 
    "00000000000000000000000000000000000000000000000000000000000000000\
0000000000000000000000000000000000000000000000000000000000000000000000\
0000000000000000000000000000000000000000000000000000000000000000000000\
0000000000000000000000000000000000000000000000000000000000000000000000\
0000000000000000000000000000"}],
Short[#, 5]& ]\)

Took roughly 6 seconds including conversion.

[Talonman]

(Thanks for the Benchmark link.)

I have a question. When I ran the CUDA Factorial Benchmark program, I set it up for 900,000 and 4 threads @ 3.81GHz on my Q6600.
My GPU is listed as GTX 295, and is running 2.5 times faster than my CPU. Just for the record, that GPU test is only calculating on 1/2 of my 295, correct?
(Basically, running on the equivalent of a single GTX 275.)


CPU takes 3 minutes, 20.828s
GPU takes 1 minute, 21.580s
2.5 x faster...


Note that if I set the benchmark for 999,000 my GPU moves up to 3.1 X as fast as my CPU. The bigger the order, the more the GPU appears to gain.
CPU takes 4 minutes, 23.231s
GPU takes 1 minute, 27.306s
Checksum: 578712543720173939

I would love to find the app, that checks to see if you have more than 1 GPU in your system, then use them all.
3 instances of folding, can load up 3 GPU's... (But that is 3 separate programs running...)
A game with PhysX can run graphics in SLI, and PhysX on another... (Still, that's Graphics on 2 or more, and PhysX on the other...)
But still not 1 benchmark program to use all GPU's in your system.

To be fair, I think CUDA apps couldn't grab cards in SLI until a recent release... I believe?
I would love to see this CUDA Factorial benchmark, grab all available GPU's in your system, in a later update.

If 1/2 of my 295 can be 3 X as fast as my Q6600, I have to wonder how many times faster it would be with both my 295 and 280 in on the deal...

I imagine DX11 will also use just 1 GPU for video transcoding, not all GPU's in your system?

[Chumbucket843]

^^ you realize we came to the conclusion cpu was faster. just read the posts above.

[Talonman]

Is that using Mathematica, and a different equation?

I don't know if we can do a valid comparison that way...

http://www.hearne.com.au/products/mathematica/pricing/
Looks pricey!!

[Chumbucket843]

cpu's and gpu's use different algorithms to get the best performance. it would be an invalid comparison if we used the same algorithm. both of these processors have advantages and drawbacks.

[poke349]

Is that using Mathematica, and a different equation?

I don't know if we can do a valid comparison that way...

http://www.hearne.com.au/products/mathematica/pricing/
Looks pricey!!

Mathematica uses GMP - which is open sourced and free, so price doesn't matter.
You'd get the same fast timings using GMP directly than through Mathematica.

Obviously this isn't a fair comparison at all.
GMP uses state-of-the-art algorithms which are much faster, but probably not as easily paralleled as whatever this CUDA benchmark uses.

[Talonman]

Thanks for the info guys..

[OverFoxtrot]

Hi ALL.

I am from www.total-oc.ru and can answer any your question about this benchmark

Now it use only one GPU from all, but we will work to make it universal. So this test can use all of CPU and CPU core simultaneously or in any combinations.

And about speed of calculation - it more slowly because is applied uniform algorithm of calculation on the processor and GPU that there was a comparability of results. Aim not to receive as much as possible fast algorithm, and universal.

This is my world record on GPU in this test - 4.331
On CPU in this test world record make community XtremeLabs.org - 6.888

Discussion n our forum in English and some results here.

I am ready to answer all interesting questions under the test

[[XC] riptide]

Hi ALL.

I am from www.total-oc.ru and can answer any your question about this benchmark

Now it use only one GPU from all, but we will work to make it universal. So this test can use all of CPU and CPU core simultaneously or in any combinations.

And about speed of calculation - it more slowly because is applied uniform algorithm of calculation on the processor and GPU that there was a comparability of results. Aim not to receive as much as possible fast algorithm, and universal.

This is my world record on GPU in this test - 4.331
On CPU in this test world record make community XtremeLabs.org - 6.888

Discussion n our forum in English and some results here.

I am ready to answer all interesting questions under the test

Welcome.

I haz a question however. There seems to be a discrepancy once you go above ~>3000n between the CPU result and the GPU result. Is there an approximation that is diff from the CPU and GPU? A different algorithm?

[HeUeR]

My Core 2 Duo E6400 + XFX 8800GTX

System:

CPU: Core 2 Duo E6400 (2.13Ghz) @ 3.20Ghz
Video: XFX 8800GTX @ standart
Ram: 4x512 OCZ Platinium @ 400Mhz

Results:

CPU: 5m 46.113s
GPU: 3m 4.773s



How can you see, my E6400 @ 3.20Ghz is little bit too week against 8800GTX, so need some Quad power

[[XC] riptide]

My Core 2 Duo E6400 + XFX 8800GTX

System:

CPU: Core 2 Duo E6400 (2.13Ghz) @ 3.20Ghz
Video: XFX 8800GTX @ standart
Ram: 4x512 OCZ Platinium @ 400Mhz

Results:

CPU: 5m 46.113s
GPU: 3m 4.773s



How can you see, my E6400 @ 3.20Ghz is little bit too week against 8800GTX, so need some Quad power

Dude... there's plenty of mileage left in that GTX.... Push it!

[HeUeR]

[XC] riptide maybe you can help, and little bit tell me about 8800GTX card OC and such things ? Pm me with more info, im intrested to oc my card to check how far it is going :

[OverFoxtrot]

[XC] riptide
This is nuances working of algorithm on GPU and CPU. If you calculate <=3000!, GPU not working, but CPU is.
I improve my personal record on CPU - 7.188

check file
generate with help AntiCheat TOC 0.9.8.3

[HeUeR]

Hi again,

got new score in gpu test with my XFX 8800GTX

OLD Result:

GPU: 3m 4.773s

NEW Result:

GPU: 2m 54.065s (~10s faster)

[Talonman]

Hi ALL.

I am from www.total-oc.ru and can answer any your question about this benchmark

Now it use only one GPU from all, but we will work to make it universal. So this test can use all of CPU and CPU core simultaneously or in any combinations.

Outstanding!! Thanks for the post. GPU's rule.

[OverFoxtrot]

HeUeR
Calculate 250000! on your video please. We get statistics about any hardware
Talonman
Our project develops more test packages of comparison of speed of video cards and processors and useful utilities. I shortly will place here a theme that we could test speed and in them.

[OverFoxtrot]

[XC] riptide
Please fix title of theme from TOC.ru to Total-OC.ru

[HeUeR]

Hi,

OverFoxtrot i got some score for you


750000 bench:

GPU: 2m 45.140s




250000 bench:

GPU: 27.903s



Both results ar with same clocked 8800gtx and same clock cpu, just in 750000 bench i forgot to take photo from gpu-z