Michaël Touati

Results 6 comments of Michaël Touati

Vectorized algorithms not implemented for 1D and 2D geometry (even with the non-adaptive method)

I'm unlucky. I was interested in 1D Cartesian expensive simulations on a desktop using the threadripper AMD technology (64 cores). I asked the question to know if the developers who...

Vectorized algorithms not implemented for 1D and 2D geometry (even with the non-adaptive method)

Re-bonjour, "There is a reason why it has not been done yet. We estimate that the performance improvement would probably not be significant. We could be wrong of course." The...

Vectorized algorithms not implemented for 1D and 2D geometry (even with the non-adaptive method)

Coucou Micka, J'espère que tu vas bien. Yes, it would be wonderful if it is implemented in 1D. I'm running expensive 1D-3V PIC simulations of TNSA using SMILEI resolving the...

Vectorized algorithms not implemented for 1D and 2D geometry (even with the non-adaptive method)

Hello Xxirii, Thank you very much for all these important pieces of information. It is very clear and, without it, it would have been impossible for me to implement it....

Vectorized algorithms not implemented for 1D and 2D geometry (even with the non-adaptive method)

Merci beaucoup

Vectorized algorithms not implemented for 1D and 2D geometry (even with the non-adaptive method)

Thank you very much @xxirii. I will test it this week and/or next week using GNU and OpenMPI compilers together with AVX and AVX2 vectorization compiling options on my threadripper...