update warp size

Author: SlyEcho

ggml-cuda.cu

@@ -132,7 +132,7 @@ static_assert(sizeof(block_q8_0) == sizeof(ggml_fp16_t) + QK8_0, "wrong q8_0 blo
 
 #define CUDA_MUL_BLOCK_SIZE 256
 #define CUDA_DEQUANTIZE_BLOCK_SIZE 256
-#define CUDA_DMMV_BLOCK_SIZE 32 // dmmv = dequantize_mul_mat_vec
+#define CUDA_DMMV_BLOCK_SIZE 64 // dmmv = dequantize_mul_mat_vec
 
 static __global__ void mul_f32(const float * x, const float * y, float * dst, const int kx, const int ky) {
     const int i = blockDim.x*blockIdx.x + threadIdx.x;