CUDA: remove incorrect precision check (ggerganov#7454)

fairydreaming · May 22, 2024 · 95fb0ae · 95fb0ae
1 parent 3e5faa8
commit 95fb0ae
Showing 1 changed file with 0 additions and 3 deletions.
diff --git a/ggml-cuda/fattn-tile-f32.cu b/ggml-cuda/fattn-tile-f32.cu
@@ -286,9 +286,6 @@ void ggml_cuda_flash_attn_ext_tile_f32(ggml_backend_cuda_context & ctx, ggml_ten
     const ggml_tensor * KQV = dst;
     const ggml_tensor * Q   = dst->src[0];
 
-    const int32_t precision = KQV->op_params[2];
-    GGML_ASSERT(precision == GGML_PREC_DEFAULT);
-
     if (Q->ne[1] <= 16) {
         constexpr int cols_per_block = 16;
         constexpr int parallel_blocks = 4;