Article: AMD's Analyst Update
By: Erwin Coumans (erwin.coumans.delete@this.gmail.com), February 9, 2012 8:39 pm
Room: Moderated Discussions
OpenCL is slowly catching up, C++ templates are an extension at the moment, see http://developer.amd.com/sdks/AMDAPPSDK/assets/cplus_kernel_language.docx
Bryan Catanzaro (bryan.catanzaro@gmail.com) on 2/9/12 wrote:
---------------------------
>David Kanter (dkanter@realworldtech.com) on 2/9/12 wrote:
>---------------------------
>>pk (pk@notmyrealaddress.com) on 2/9/12 wrote:
>>---------------------------
>>>David, the quote above is quite a heavy statement. Could you elaborate?
>>>
>>>Nvidia would certainly argue that CUDA gives you the absolute best performance
>>>for the highly parallel applications often run in HPC environments, and those people
>>>benefit greatly from it. Are you arguing that CUDA will just >stay confined to the HPC space?
>>
>>What is the performance of CUDA code on an AMD GPU? Or Knight's Corner? I'm pretty
>>sure the performance is very bad compared to the alternatives : )
>>
>>I agree that CUDA is acceptable where you want to be tied to NV's hardware. For HPC, that might be fine.
>>
>>But for consumer applications, you need GPU acceleration on ALL hardware platforms.
>>So if you want CUDA, you probably also need OpenCL, etc.
>>
>>At that point, why would you ever bother with CUDA in the first place?
>>
>>My point is that proprietary programming models that are tied to hardware with
>>small marketshare are stupid where alternatives exist. Initially there was no alternative to CUDA. Now there is.
>>
>>David
>
>Have you used OpenCL? CUDA has a lot of important features that OpenCL does not,
>C++ templates being the biggest (which are incredibly useful in parallel programming),
>and CUDA is also extraordinarily simpler to use. Hello World in OpenCL is about 1000 lines of code. In CUDA it's 10.
>
Bryan Catanzaro (bryan.catanzaro@gmail.com) on 2/9/12 wrote:
---------------------------
>David Kanter (dkanter@realworldtech.com) on 2/9/12 wrote:
>---------------------------
>>pk (pk@notmyrealaddress.com) on 2/9/12 wrote:
>>---------------------------
>>>David, the quote above is quite a heavy statement. Could you elaborate?
>>>
>>>Nvidia would certainly argue that CUDA gives you the absolute best performance
>>>for the highly parallel applications often run in HPC environments, and those people
>>>benefit greatly from it. Are you arguing that CUDA will just >stay confined to the HPC space?
>>
>>What is the performance of CUDA code on an AMD GPU? Or Knight's Corner? I'm pretty
>>sure the performance is very bad compared to the alternatives : )
>>
>>I agree that CUDA is acceptable where you want to be tied to NV's hardware. For HPC, that might be fine.
>>
>>But for consumer applications, you need GPU acceleration on ALL hardware platforms.
>>So if you want CUDA, you probably also need OpenCL, etc.
>>
>>At that point, why would you ever bother with CUDA in the first place?
>>
>>My point is that proprietary programming models that are tied to hardware with
>>small marketshare are stupid where alternatives exist. Initially there was no alternative to CUDA. Now there is.
>>
>>David
>
>Have you used OpenCL? CUDA has a lot of important features that OpenCL does not,
>C++ templates being the biggest (which are incredibly useful in parallel programming),
>and CUDA is also extraordinarily simpler to use. Hello World in OpenCL is about 1000 lines of code. In CUDA it's 10.
>
Topic | Posted By | Date |
---|---|---|
Fused Multiply-Add and HSA | Richard H. | 2012/02/07 06:29 PM |
Fused Multiply-Add and HSA | David Kanter | 2012/02/07 10:54 PM |
"CUDA is on its way out" | pk | 2012/02/09 04:37 AM |
"CUDA is on its way out" | David Kanter | 2012/02/09 10:49 AM |
"CUDA is on its way out" | Bryan Catanzaro | 2012/02/09 11:44 AM |
"CUDA is on its way out" | David Kanter | 2012/02/09 04:48 PM |
"CUDA is on its way out" | Anon | 2012/02/09 06:11 PM |
"CUDA is on its way out" | David Kanter | 2012/02/09 11:00 PM |
"CUDA is on its way out" | Michael S | 2012/02/10 03:54 AM |
"CUDA is on its way out" | Anon | 2012/02/10 05:42 PM |
Proprietary SW interfaces and hardware vendors | David Kanter | 2012/02/10 08:07 PM |
Proprietary SW interfaces and hardware vendors | Anon | 2012/02/11 03:08 AM |
Proprietary SW interfaces and hardware vendors | Anon | 2012/02/11 03:12 AM |
Proprietary SW interfaces and hardware vendors | Del | 2012/02/11 07:39 AM |
"CUDA is on its way out" | Del | 2012/02/11 07:33 AM |
"NVIDIA is on its way out" | Brendan | 2012/02/10 09:42 PM |
"NVIDIA is on its way out" | anonymous | 2012/02/11 10:37 AM |
"CUDA is on its way out" | Del | 2012/02/11 07:22 AM |
"CUDA is on its way out" | Anon | 2012/02/11 03:11 PM |
ispc on GPUs | Matt Pharr | 2012/02/12 09:02 AM |
ispc on GPUs | Anon | 2012/02/13 02:34 PM |
"CUDA is on its way out" | Erwin Coumans | 2012/02/09 08:39 PM |
"CUDA is on its way out" | Brett | 2012/02/10 11:54 AM |
"CUDA is on its way out" | David Kanter | 2012/02/10 03:00 PM |
"CUDA is on its way out" | Anon | 2012/02/10 05:45 PM |
"CUDA is on its way out" | Brett | 2012/02/10 07:20 PM |
"CUDA is on its way out" | David Kanter | 2012/02/10 07:58 PM |
"CUDA is on its way out" | Brett | 2012/02/11 06:15 AM |
Macroscalar | none | 2012/02/11 07:40 AM |
"CUDA is on its way out" | Anon | 2012/02/10 05:39 PM |
"CUDA is on its way out" | David Kanter | 2012/02/10 08:19 PM |
"CUDA is on its way out" | Anon | 2012/02/11 03:33 AM |
"CUDA is on its way out" | NN | 2012/02/14 03:24 PM |
"CUDA is on its way out" | Anon | 2012/02/15 03:16 PM |
"CUDA is on its way out" | bakaneko | 2012/02/11 04:31 AM |
"CUDA is on its way out" | Aaron Spink | 2012/02/09 11:29 PM |
Fused Multiply-Add and HSA | EduardoS | 2012/02/08 02:11 PM |
Fused Multiply-Add and HSA | Foo_ | 2012/02/09 05:24 AM |