CUDA Leaks: Information Leakage in GPU Architectures — mentioned by inetknght

Mon, 15 May 2017 23:38:23 GMT

Yup, I'd strongly consider something similar. But I'd have to seriously consider the security implications: giving Windows access to the graphics card could end up being more hassle than it's worth. GPUs' memory management isn't exactly security friendly. -> https://arxiv.org/pdf/1305.7383.pdf

Paper and discussion history. A mention is not an endorsement.

CUDA Leaks: Information Leakage in GPU Architectures — mentioned by paulmd

Thu, 27 Apr 2017 18:54:09 GMT

There's two killers, compatibility and launch overhead. CUDA is the most commonly-used compute language but it's NVIDIA-only. OpenCL is "universal" (sort of) but - being a general-purpose framework for heterogeneous compute rather than a task-specific GPGPU language - it's much more complex and lacks the libraries/etc that are available in CUDA. NVIDIA's OpenCL performance is all over the board, sometimes good sometimes fairly terrible (half of CUDA performance or less). Other users have nothing at all. So inherently - any GPU acceleration targets only a subset of your users. The other killer…

Paper and discussion history. A mention is not an endorsement.

CUDA Leaks: Information Leakage in GPU Architectures · HackerPapers discussions

CUDA Leaks: Information Leakage in GPU Architectures — mentioned by inetknght

CUDA Leaks: Information Leakage in GPU Architectures — mentioned by paulmd