[Mesa-dev] [RFC] ralloc: use jemalloc for faster GLSL compilation

Marek Olšák maraeo at gmail.com
Sun Oct 2 14:51:50 UTC 2016

On Sun, Oct 2, 2016 at 4:03 PM, ⚛ <0xe2.0x9a.0x9b at gmail.com> wrote:
> Hello Marek
> instead of utilizing jemalloc, I strongly recommend you take a look (for
> example) at:
> classes such as ir_copy_propagation_elements_visitor
> file glcpp-parse.y
> and to put optimizations and better algorithms in there.
> Eliminating the causes of malloc-like calls will result in 2x faster shader
> compilation in the case of Alien Isolation - instead of the 1.1x speedup you
> achieved with jemalloc.

Can you prove your claim?

BTW, I recommend not to use callgrind, as it's not accurate (it
doesn't measure the real CPU time).


More information about the mesa-dev mailing list