In order to call a function multiple time at parallel (rendering function, very well parallelizable), but in order to push the CPU to it's limits I have to get some form of parallelization. Currently I'm using parallel foreach, which isn't very nice, and since it uses GC allocation that creates some performance impact from time to time.

On Saturday, 21 April 2018 at 02:08:24 UTC, solidstate1991 wrote:
> In order to call a function multiple time at parallel (rendering function, very well parallelizable), but in order to push the CPU to it's limits I have to get some form of parallelization. Currently I'm using parallel foreach, which isn't very nice, and since it uses GC allocation that creates some performance impact from time to time.

Did you attempt to create a memory pool with the GC? Any memory allocation will take time, whatever it's CG or manual.