4 ms·
Intel's pushing OpenCL, which seems to me gives most of the advantages of HSA. OpenCL 2.0 includes unified CPU & GPU virtual addresses (SVM - shared virtual me
by bjwbell 11y ago
Intel's pushing OpenCL, which seems to me gives most of the advantages of HSA.
OpenCL 2.0 includes unified CPU & GPU virtual addresses (SVM - shared virtual memory).
- pmalynin 11y agoAnd so does CUDA, which usually out-performs OpenCL, plus the tooling for CUDA is just so much better with the Visual Studio integration.
- glogla 11y agoYes, but CUDA is Nvidia only - another trick in their anti-competitive bag. I always get disappointed when I see something that CUDA specific. "Hey that's interesting machine learning package! Oh wait, that's Nvidia specific." With OpenCL, you can develop on laptop, test on workstation and deploy on actual servers. With CUDA, you're trapped.
- m_mueller 11y ago> With OpenCL, you can develop on laptop, test on workstation and deploy on actual servers. With CUDA, you're trapped. GPGPU developers usually prefer hardware with Nvidia GPUs inside, so you can do all this too. I wouldn't call Nvidia anti competitive, just like I wouldn't use that word for Apple. They saw a niche (HPC/smartphones) when no one thought that market to be attractive, jumped in with proprietary technology that others later tried to reproduce (OpenCL/Android) and they kept the market leadership in terms of profits by reiterating their product.
- pmalynin 11y agoWell I'm developing on a laptop with GTX 860M so I can still develop on a laptop, test on a workstation (GTX 970) and deploy to actual servers (Tesla)