• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Better OpenCL support in NVIDIA's CUDA SDK

humbug said:
Signing this benefits us all, the status quo benefits only Nvidia.
Click to expand...

I agree with the sentiment, however, Consoles going AMD could push OpenCL for hardware accelerated physics, that could actually be at a negative for Nvidia users since they'd be a loss of support.
 
humbug said:
AMD have put huge amounts of time and money into developing OpenCL for the benefit of everyone, and it is paying off with technological advancements.
The problem for Nvidia is increasingly you don't need Nvidia for GPGPU, whats more AMD's standard £350 GPU's have a lot more grunt than Nvidia's equivalent, if you want that level of GPGPU with CUDA it will cost you £3000 for an Nvidia Tesla GPU.
Click to expand...

No not really, and nowhere near what Nvidia have done with CUDA. OpenCL was an Apple initiative that AMD gave their 'Close to Metal' stuff to after letting it stagnate.

To be fair, it would cost you £800 for a Titan which offers far more performance than anything AMD have on the table, and if you only want SP then a 780 for ~£500 beats anything AMD have to offer when using CUDA. Those £3k cards you talk about support stuff like ECC (tell me, how much do Firepro W series cards offering this cost) which is an absolute must for any serious financial/scientific computing.

As it stands, they do support OpenCL (not as much as AMD fans would like), while offering support for their own mature GPGPU software platform as well. I'm not sure why some are getting worked up about it (well you can guess but lets be kind) if Nvidia keep it up then surely OpenCL will become far superior on AMD hardware, or are you worried that it cannot compete against the superior (but propriety) CUDA without assistance from Nvidia?

I cant help but feel that this is being used as another excuse for certain individuals to beat Nvidia with a stick. The constant arguing by a small group of people (no matter what card they happen to own) is making this forum an absolute chore to read. I wish the mods would get a bit more heavy handed like they once threatened.
 
Last edited:
humbug said:
AMD have put huge amounts of time and money into developing OpenCL for the benefit of everyone, and it is paying off with technological advancements.
Click to expand...

AMD have put a lot of time and money into talking about OpenCL... the people who have done most of the grunt work are the likes of intel and even apple. I'd almost go as far to say AMD have spent more time talking about the demise of CUDA than developing OpenCL (this isn't actually as wild an exaggeration even tho I'm exaggerating).

Lets just hope they don't drop the ball as badly with Adobe as they did with Bullet.
 
deuse said:
As I said the only people who want this is ATI people. All good 3D\Photo\Video software uses Cuda

As we all know hardware is faster then software. Why not ask ATI to get OpenCL working like Cuda does?

And as for "open standards are better for everyone" did that guy ever work in the network trade when it first started?

This just AMD fan boys wanting something for nothing AGAIN.
Click to expand...

Well the people who started this petition are NVIDIA users because the bigger picture its better for the developers and the PC community as a whole, the more users who can run it the better, they don't want to have to choose sides or write 2 versions.

http://www.linkedin.com/groups/Petition-put-back-OpenCL-samples-1729897.S.149050851
 
Last edited:
layte said:
No not really, and nowhere near what Nvidia have done with CUDA. OpenCL was an Apple initiative that AMD gave their 'Close to Metal' stuff to after letting it stagnate.

To be fair, it would cost you £800 for a Titan which offers far more performance than anything AMD have on the table, and if you only want SP then a 780 for ~£500 beats anything AMD have to offer when using CUDA. Those £3k cards you talk about support stuff like ECC (tell me, how much do Firepro W series cards offering this cost) which is an absolute must for any serious financial/scientific computing.

As it stands, they do support OpenCL (not as much as AMD fans would like), while offering support for their own mature GPGPU software platform as well. I'm not sure why some are getting worked up about it (well you can guess but lets be kind) if Nvidia keep it up then surely OpenCL will become far superior on AMD hardware, or are you worried that it cannot compete against the superior (but propriety) CUDA without assistance from Nvidia?

I cant help but feel that this is being used as another excuse for certain individuals to beat Nvidia with a stick. The constant arguing by a small group of people is making this forum an absolute chore to read. I wish the mods would get a bit more heavy handed like they once threatened.
Click to expand...

To be fair, it would cost you £800 for a Titan which offers far more performance than anything AMD have on the table, and if you only want SP then a 780 for ~£500 beats anything AMD have to offer when using CUDA.
Click to expand...

Well AMD don't have CUDA, you just stated the obvious, yet a 7970 using OpenCL is faster than a Titan using CUDA, for some tasks the 7970 is as fast as the £3000 Tesla using CUDA.

As it stands, they do support OpenCL (not as much as AMD fans would like),
Click to expand...

Its not about what AMD fans would like, I use OpenCL for rendering, encoding ecte... regular with great performance, I don't need CUDA for anything.
The performance on an increasing number of games for you with your Nvidia GPU is junk! while its great for me because my GPU has far superior open standard compute.
That will only get worse for you as the new Game Consoles take effect.

I don't get you Nvidia fans, it seems to me you would rather cut your nose off to spite your face.

Instead of just using things like this to big up Nvidia and weirdly defending the indefensible would you rather not have better compatibility performance with your own GPU?

I / We AMD fans as you put it, don't need Nvidia to stop behaving like infants, OpenCL is working very well for us and unlike you we don't have to pay a premium for it.

your the one who needs this more than we do so your games don't grind to a stuttering halt, so you can use mainstream and free accelerated applications, how many Open Source Apps use CUDA? none!

Hey singe it if you want, but don't do it for me, do it for yourself, or cut your own nose off. the choice is yours.
 
Last edited:
humbug said:
Well AMD don't have CUDA, you just stated the obvious, yet a 7970 using OpenCL is faster than a Titan using CUDA, for some tasks the 7970 is as fast as the £3000 Tesla using CUDA.



Its not about what AMD fans would like, I use OpenCL for rendering, encoding ecte... regular with great performance, I don't need CUDA for anything.
The performance on an increasing number of games for you with your Nvidia GPU is junk! while its great for me because my GPU has far superior open standard compute.
That will only get worse for you as the new Game consoles take effect.

I don't get you Nvidia fans, it seems to me you would rather cut your nose off to spite your face.

Instead of just using things like this to big up Nvidia and weirdly defending the indefensible would you rather not have better compatibility performance with your own GPU?

I / We AMD fans as you put it, don't need Nvidia to stop behaving like infants, OpenCL is working very well for us and unlike you we don't have to pay a premium for it.

your the one who needs this more than we do so your games don't grind to a stuttering halt, so you can use mainstream and free accelerated applications, how many Open Source Apps use CUDA? none!

Hey singe it if you want, but don't do it for me, do it for yourself, or cut your own nose off. the choice is yours.
Click to expand...

Titan with CUDA against 7970 with OpenCL: http://www.anandtech.com/show/6774/nvidias-geforce-gtx-titan-part-2-titans-performance-unveiled/3

As for your overly emotional response. I'd be interested if you could point out where the performance in games is increasingly junk. Some specifics would be interesting. Even Tomb Raider with it's AMD developed (and OpenCL using) TressFX runs better on my supposedly poorly supporting of open standards Titan than it does on any GPU AMD have on the market right now. So while all or nothing benchmarks show that Nvidia's performance is lacking, real world numbers do not back this up.

Of course I would like better performance, who wouldn't. But right now it is of very little consequence to the vast majority of people. Big GPGPU people have bespoke systems they can develop to suit their needs, business is well served by a number of CUDA using solutions and while consumer level application performance is lacking, it is hardly the indefensible affront you are trying your best to make it out to be. I'd imagine if it becomes the problem you are making it out to be already (it's not) then Nvidia will do something about it, as if they don't they will lose business.
 
Last edited:
layte said:
Titan with CUDA against 7970 with OpenCL: http://www.anandtech.com/show/6774/nvidias-geforce-gtx-titan-part-2-titans-performance-unveiled/3

As for your overly emotional response. I'd be interested if you could point out where the performance in games is increasingly junk. Some specifics would be interesting. Even Tomb Raider with it's AMD developed (and OpenCL using) TressFX runs better on my supposedly poorly supporting of open standards Titan than it does on any GPU AMD have on the market right now. So while all or nothing benchmarks show that Nvidia's performance is lacking, real world numbers do not back this up.

Of course I would like better performance, who wouldn't. But right now it is of very little consequence to the vast majority of people. Big GPGPU people have bespoke systems they can develop to suit their needs, business is well served by a number of CUDA using solutions and while consumer level application performance is lacking, it is hardly the indefensible affront you are trying your best to make it out to be. I'd imagine if it becomes the problem you are making it out to be already (it's not) then Nvidia will do something about it, as if they don't they will lose business.
Click to expand...

Anands bench there is very synthetic, its like using WIE or IBT to measure performance.
Toms Hardwear use real world applications, if you look through that its very much Titan in one and the 7970 in the other, depending on which application.
http://www.tomshardware.com/reviews/geforce-gtx-titan-opencl-cuda-workstation,3474-23.html

As for Tomb Raider



That's the £300 7970 Ghz faster than the £900 GTX Titan.

To match the GTX 680 all that is needed is a 7870.
 
Tomb Raider isn't a great way to compare compute ability, TressFX has been (and its a valid way to do it) implemented in a way that suits superscalar architectures i.e. what AMD use and even GCN still inherently works well with that kind of workload. Its not really representative of overall compute performance.
 
You must log in or register to reply here.
Back
Top Bottom