diff options
author | Tim Dettmers <tim.dettmers@gmail.com> | 2022-07-25 19:30:37 -0700 |
---|---|---|
committer | Tim Dettmers <tim.dettmers@gmail.com> | 2022-07-25 19:30:37 -0700 |
commit | 9268dc9d887a3d54cd1f008dcb628aaa5b5bd90a (patch) | |
tree | c557a57a63e627e6694bb383d0f7c8ade5f9071a /csrc/kernels.cuh | |
parent | 1e88edd8c096bde5202dd61411d3c8d7eda56645 (diff) |
Some progress on build script; added multi-cuda install script.
Diffstat (limited to 'csrc/kernels.cuh')
-rw-r--r-- | csrc/kernels.cuh | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/csrc/kernels.cuh b/csrc/kernels.cuh index cbfbeba..4e65e96 100644 --- a/csrc/kernels.cuh +++ b/csrc/kernels.cuh @@ -107,7 +107,7 @@ template<typename T, int BLOCK_SIZE, int NUM_VALS> __global__ void kPercentileCl __global__ void kHistogramScatterAdd2D(float* histogram, int *index1, int *index2, float *src, const int maxidx1, const int n); -template <typename T, int SPMM_ITEMS, int BITS> __global__ void kspmm_coo_very_sparse_naive(int *max_count, int *max_idx, int *offset_rowidx, int *rowidx, int *colidx, half *values, T *B, half *out, float *dequant_stats, int nnz, int rowsA, int rowsB, int colsB); +template <typename T, int SPMM_ITEMS, int BITS> __global__ void kspmm_coo_very_sparse_naive(int *max_count, int *max_idx, int *offset_rowidx, int *rowidx, int *colidx, half *values, T *B, half *out, float * __restrict__ const dequant_stats, int nnz, int rowsA, int rowsB, int colsB); template <int ITEMS_PER_THREAD, int SUBTILE_ROWS, int THREADS>__global__ void kdequant_mm_int32_fp16( int *__restrict__ const A, float *__restrict__ const rowStats, float *__restrict__ const colStats, |