Method identified to double computer processing speeds (sciencedaily.com) 11 points by _justinfunk 2y ago ↗ HN
[–] orf 2y ago ↗ They implemented a kernel driver that processes can submit work items to (multiply this vector, summing reduce et ). It balances the computation across available hardware.Seems really neat, if not a bit click-baity.Paper: https://dl.acm.org/doi/pdf/10.1145/3613424.3614285Code: https://github.com/escalab/SHMT [–] karmakaze 2y ago ↗ The work would have to be large enough to be worth the cost of kernel context switch IIUC. [–] westurner 2y ago ↗ src/Python/kernels/ground_truth_functions.py: https://github.com/jk78346/SHMT/blob/main/src/Python/kernels... :> [ blackscholes_2d, dct8x8_2d, dwt, hotsplot_2d, srad_2d, sobel_2d, npu_sobel_2d, minimum_2d, mean_2d, laplacian_2d, fft_2d, histogram_2d ]src/kernels: https://github.com/jk78346/SHMT/tree/main/src/kernels :> [ convolutionFFT2D, ]What is the (platform) kernel context switching overhead tradeoff for what size workloads of the already implemented functions?
[–] karmakaze 2y ago ↗ The work would have to be large enough to be worth the cost of kernel context switch IIUC. [–] westurner 2y ago ↗ src/Python/kernels/ground_truth_functions.py: https://github.com/jk78346/SHMT/blob/main/src/Python/kernels... :> [ blackscholes_2d, dct8x8_2d, dwt, hotsplot_2d, srad_2d, sobel_2d, npu_sobel_2d, minimum_2d, mean_2d, laplacian_2d, fft_2d, histogram_2d ]src/kernels: https://github.com/jk78346/SHMT/tree/main/src/kernels :> [ convolutionFFT2D, ]What is the (platform) kernel context switching overhead tradeoff for what size workloads of the already implemented functions?
[–] westurner 2y ago ↗ src/Python/kernels/ground_truth_functions.py: https://github.com/jk78346/SHMT/blob/main/src/Python/kernels... :> [ blackscholes_2d, dct8x8_2d, dwt, hotsplot_2d, srad_2d, sobel_2d, npu_sobel_2d, minimum_2d, mean_2d, laplacian_2d, fft_2d, histogram_2d ]src/kernels: https://github.com/jk78346/SHMT/tree/main/src/kernels :> [ convolutionFFT2D, ]What is the (platform) kernel context switching overhead tradeoff for what size workloads of the already implemented functions?
3 comments
[ 2.1 ms ] story [ 18.5 ms ] threadSeems really neat, if not a bit click-baity.
Paper: https://dl.acm.org/doi/pdf/10.1145/3613424.3614285
Code: https://github.com/escalab/SHMT
> [ blackscholes_2d, dct8x8_2d, dwt, hotsplot_2d, srad_2d, sobel_2d, npu_sobel_2d, minimum_2d, mean_2d, laplacian_2d, fft_2d, histogram_2d ]
src/kernels: https://github.com/jk78346/SHMT/tree/main/src/kernels :
> [ convolutionFFT2D, ]
What is the (platform) kernel context switching overhead tradeoff for what size workloads of the already implemented functions?