SuperPi on GPU, were going CUDA

[Lestat]

i would simply like to see something like converting a standard 90 minute video standard video size or 720p done the standard good ol way and then done by GPGPU.
thats something that alot of people can relate to and could truly show the difference or advantage

[dinos22]

did you watch this video
because that's what it's showing in second part of presentation pretty much with a shorter clip obviously

http://www.youtube.com/watch?v=8C_Pj1Ep4nw

[MuffinFlavored]

and what did that email say?

Not what I thought it did.

[Charles Wirth]

I will have an update tomorrow afternoon.

[Luka_Aveiro]

I will have an update tomorrow afternoon.

That's good to hear, seems like the train is moving

[ORBR]

I will have an update tomorrow afternoon.

Wow THAT's GREAT

it's interesting to see gpu vs cpu at work

but if you can give us a CTM version, I'm sure the benchmark will be used in almost EVERY review of GPU's in future

Nvidia version agains ati's one is FAR MORE interesting and useful

[MuffinFlavored]

I will have an update tomorrow afternoon.

Any updates? I am rather excited for such application.

I looked over the CUDA SDK, and it is serious stuff.

[Trouffman]

did you watch this video
because that's what it's showing in second part of presentation pretty much with a shorter clip obviously

http://www.youtube.com/watch?v=8C_Pj1Ep4nw

Hey guys, I' haven't post this video regarding CUDA due I'm really busy since 2 Months...

Here it is : http://www.syndrome-oc.net/octv.php?episode=10 (as usual French + English Subtitles)

@FUGGER : looking forward for this new benchmark

[ZOMGVTEK]

I will have an update tomorrow afternoon.

I WANT THIS NOW!!!!!!!!!!!!!!11!!!!!!!!!!!!ONE!!!!!!!!!!SHIFT! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!11

[Trouffman]

hey stay cool man We're all exited about this new version !

[ZOMGVTEK]

hey stay cool man We're all exited about this new version !

111!!!!!!!!!!1111111111111111111111111111111111111 11111111111111111111111111111111111111111

I want SOMETHING to do with my video card.

F@H, SuperPI... whatever, WHEN DO I GET IT!??!?!?

[iop3u2]

http://techreport.com/articles.x/14968

Some new developments in the AMD side. Open source as well. Looks quite attractive.

[Charles Wirth]

Sorry for the delay, making exchanges working on getting the best working code.

Minimal progress so far.

[MuffinFlavored]

Sorry for the delay, making exchanges working on getting the best working code.

Minimal progress so far.

Good things should be done right the first time and take time doing it, I see it as there is no rush.

Thank you for the update.

[Horus´]

So if i read well it should be "Superpi on Nvidia gpu"
Ati won't be able to run this , am i right?

Sounds cool though

[Helmore]

So if i read well it should be "Superpi on Nvidia gpu"
Ati won't be able to run this , am i right?

Sounds cool though

That's right as far as I know, ATI has no intentions to run CUDA on their GPUs. They have more faith in the upcoming OpenCL programming language, which is currently being developed by Apple and others. Maybe SuperPi on OpenCL is the next step? Although I have no idea when the OpenCL standard is supposed to be 'standardized'. (little wrong wording there )

[Charles Wirth]

Progress!

Playing with threaded code now.

[BenchZowner]

Me wants some

[Lestat]

great to hear charles..

the community waits with baited breath

[Planet]

By threaded you mean it will take advantage of multiple display configurations.

[Charles Wirth]

No, it will thread itself to the cores without being prompted for number of cores.

[safan80]

will we still be able to do single core only runs on a multi core cpu?

[Lestat]

will we still be able to do single core only runs on a multi core cpu?

its not the cpu here its the GPU. gpu's are always multi threaded never single threaded (correct charles?)

and why would you want to do a single core run ?
if its for stability testing you can use orthos (SP2004) and set the affinity to the core you want.

[Charles Wirth]

For cuda enabled it is massive parallel.

[safan80]

and why would you want to do a single core run ?
if its for stability testing you can use orthos (SP2004) and set the affinity to the core you want.

thats true.. the big reason I'd do single core runs is to test if certain cores are running slower than others. I haven't had any problems it's just something I'm doing when I'm ocing.

For cuda enabled it is massive parallel.

ok... I'm just curious about what options will be enabled in the new versions.