Added launch bounds to fix launch resource error for Lion.

0b2ebcda · Tim Dettmers · ed6f3eb1 · 0b2ebcda
Commit 0b2ebcda authored Apr 11, 2023 by Tim Dettmers
Hide whitespace changes
Inline Side-by-side

Showing with 1 addition and 0 deletions

csrc/kernels.cu csrc/kernels.cu +1 -0

No files found.
--- a/csrc/kernels.cu
+++ b/csrc/kernels.cu
@@ -1260,6 +1260,7 @@ kPreconditionOptimizerStatic8bit1State(T* p, T* __restrict__ const g, unsigned c

 template<typename T, int OPTIMIZER>
 __global__ void
+__launch_bounds__(1024, 1)
 kOptimizerStatic8bit1State(T* p, T* const g, unsigned char* state1,
                const float *unorm, const float max_unorm, const float param_norm,
                const float beta1, const float beta2,