Hi, I am trying to improve the memory management of some complex iterative procedure. Basically the procedure could be synthesized to for ii = 1:Nt do some partial arrays value update Compute some stuff (include fft ffti and other) end I have some question about partial assignment of CuArrays…

[image] dgasteau: Every computation step is made inplace, all CuArrays are created at the beginning of the program, though I still have allocations. You are looking at CPU allocations. Use CUDA.@time to report GPU allocations.

Inplace array modification performances

Specific Domains GPU

maleadt July 12, 2022, 6:35am 4

You are looking at CPU allocations. Use CUDA.@time to report GPU allocations.

2 Likes

Topic		Replies	Views
Calculate FFT on GPU for every row of a 2D array Performance gpu	2	1047	August 14, 2018
Dot-product of CuArray views is slow GPU performance , memory-allocation , views	10	1570	May 11, 2021
How do I get allocations down but keep code speed? GPU	9	360	January 28, 2023
What is the optimal way of updating CuArray? GPU cudanative	7	1553	July 5, 2018
Performance of view with cuArrays GPU	11	2708	November 11, 2018

Inplace array modification performances

Related topics