Skip to content

Improve launch performance of kernels

Churavy, Valentin requested to merge vc/launch_perf into master

Created by: vchuravy

Before:

[ Info: Ka Launch
BenchmarkTools.Trial: 
  memory estimate:  1.73 KiB
  allocs estimate:  32
  --------------
  minimum time:     9.702 μs (0.00% GC)
  median time:      10.228 μs (0.00% GC)
  mean time:        11.063 μs (0.00% GC)
  maximum time:     68.124 μs (0.00% GC)
  --------------
  samples:          10000
  evals/sample:     1

After:

[ Info: Ka Launch
BenchmarkTools.Trial: 
  memory estimate:  816 bytes
  allocs estimate:  9
  --------------
  minimum time:     4.653 μs (0.00% GC)
  median time:      5.074 μs (0.00% GC)
  mean time:        5.444 μs (0.57% GC)
  maximum time:     317.282 μs (97.24% GC)
  --------------
  samples:          10000
  evals/sample:     7

Merge request reports