|
From: Sam B. <seb...@gm...> - 2015-05-01 20:44:00
|
Hi, Is there a way to enable sse4.2 vectorization while compiling a mex with gcc? I tried the option -O3 -mtune=native (my machine is i7 with sse4.2 support). The code is compiled and executes but is extremely slow. Compiling with MSVC2010 compiler I get the optimization working. I tried to manually edit mexopts.bat to add a -sse4.2 flag but it did not work. Thanks, Seb |