Message boards :
Number crunching :
OpenCL: democracy for GPU Computing
Message board moderation
| Author | Message |
|---|---|
|
Send message Joined: 13 Mar 09 Posts: 59 Credit: 324,366 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]()
|
Interesting article about the ratification of the OpenCL standard http://www.behardware.com/articles/744-1/opencl-democracy-for-gpu-computing.html |
ZydorSend message Joined: 8 Feb 09 Posts: 252 Credit: 1,309,451 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]()
|
Nice find - thank you for posting it. Its a good "real world" overview and strongly makes the point that talk of a "brave new world" of OpenCL computing with all chatting happily to each other is a ways off yet. At the same time it was bullish about the eventual outcome. If the internal thoughts of the various OpenCL teams are similarly bullish and long-term cooperative minded, its looking good. Regards Zy |
Paul D. BuckSend message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
It also brings out the point I made some time ago, writing in CUDA or Brook /CAL is going to be more efficient than OpenCL and that there will be lag in support for newer cards. That said, it will still allow more portable code for those projects without the resources to create versions for the two MFGR's APIs ... |
|
Send message Joined: 1 Feb 09 Posts: 139 Credit: 575,023 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]()
|
As i read this i think there is a super long road to go before its gonna be usefull. So i guess its gonna be cuda and cal along this long road |
Paul D. BuckSend message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
As i read this i think there is a super long road to go before its gonna be usefull. There will be a place for CUDA and CAL (or their successors) forever ... OpenCL will complement them, never replace them ... I suspect only those projects that cannot afford to invest in the "hand-coding" needed to use the CUDA or CAL APIs will use OpenCL. |
©2025 Universitat Pompeu Fabra