Comments on: A case study in CUDA optimization http://blog.accelereyes.com/blog/2010/02/20/a-case-study-in-cuda-optimization/ Helpful posts about GPU computing. Discussion of Jacket and ArrayFire. Real speedups on real code! Wed, 02 Feb 2011 15:32:00 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: GPU MATLAB Computing » Median Filtering: CUDA tips and tricks http://blog.accelereyes.com/blog/2010/02/20/a-case-study-in-cuda-optimization/comment-page-1/#comment-158 GPU MATLAB Computing » Median Filtering: CUDA tips and tricks Fri, 05 Mar 2010 00:11:00 +0000 http://www.accelereyes.com/blog/?p=167#comment-158 [...] week we posted a video recording from NVIDIA’s GTC09 conference. In the video, I walked through median filtering, presenting [...] [...] week we posted a video recording from NVIDIA’s GTC09 conference. In the video, I walked through median filtering, presenting [...]

]]> By: malcolm http://blog.accelereyes.com/blog/2010/02/20/a-case-study-in-cuda-optimization/comment-page-1/#comment-157 malcolm Fri, 05 Mar 2010 00:06:17 +0000 http://www.accelereyes.com/blog/?p=167#comment-157 It's easy to force nvcc to use at most 9 registers, but then it'll just spill more into lmem -- there's no way to prevent it from spilling to lmem except changing the algorithm so it demands fewer registers. But I like your thinking, so I put together another set of experiments--some new, some old--and I got a little carried away so I made a <a href="http://www.accelereyes.com/blog/2010/03/04/median-filtering/" rel="nofollow">new post</a>. Thanks for the suggestion! -jm It’s easy to force nvcc to use at most 9 registers, but then it’ll just spill more into lmem — there’s no way to prevent it from spilling to lmem except changing the algorithm so it demands fewer registers.

But I like your thinking, so I put together another set of experiments–some new, some old–and I got a little carried away so I made a new post.

Thanks for the suggestion!
-jm

]]> By: Rohit http://blog.accelereyes.com/blog/2010/02/20/a-case-study-in-cuda-optimization/comment-page-1/#comment-110 Rohit Wed, 24 Feb 2010 02:01:56 +0000 http://www.accelereyes.com/blog/?p=167#comment-110 Great tutorial! I wish it was there 1.5 years ago when I was evaluating motion estimation algorithms on CUDA. Something that is missing from the evaluation is the effect of forcing the nvcc compiler to use 9 registers and prevent it from spilling the registers to memory and still using bubble sort. It would be interesting to see if that results in an improvement or degrades performance. Great tutorial! I wish it was there 1.5 years ago when I was evaluating motion estimation algorithms on CUDA.

Something that is missing from the evaluation is the effect of forcing the nvcc compiler to use 9 registers and prevent it from spilling the registers to memory and still using bubble sort. It would be interesting to see if that results in an improvement or degrades performance.

]]>