You can see that the faster version of the algorithm (the second benchmark test) in WSL is almost twice faster as in its Windows version and is approximately as fast as Mathematica implementation. The screenshots below show (left to right) the speed of this convolution (two different algorithms) calculated by Julia running in WSL and Julia running in Windows, both versions 1.5.2 (and the same convolution made in Mathematica 12.1 for comparison purposes). MADEin4 project, we need to simulate a position-dependent blur, so the convolution speed is a very important parameter for us). Here are the results from a simple benchmarking of convolving a 512 x 512 image with a 48 x 48 blur kernel. Finally I’ve got some time to try it, and it worked even better than I expected! Namely, the benefit I’ve got is a faster version of Julia on my computer (not to mention enjoying 24K colours in REPL and normal work of all commands in the shell mode □). Publications praised the diversity of the distributives (you are not limited to Ubuntu anymore), a closer integration with Windows (you can call Windows programs from Linux, and the filesystems of Windows and Linux can be accessed bidirectionally), and I also saw some manuals how to install WSL on another drive. Recently, I’ve seen on Internet some news about WSL2, the second version of this Linux subsystem. I tried it, installed python via conda, watched how the free space on my fast but small system drive (SSD) quickly disappeared and forgot it. To get access to it and use it, one needed just to activate the developer mode and voilà - you typed bash and got Ubuntu 2016 (in terminal only). From its first days, Windows 10 provided a full-featured Linux (sub)system, called WSL.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |