Skip to content

Instantly share code, notes, and snippets.

View interestingLSY's full-sized avatar

Shengyu Liu interestingLSY

View GitHub Profile
@interestingLSY
interestingLSY / element_stride.cu
Created August 15, 2025 08:08
Meaning of "elementStride" in TMA descriptor
#include <cstdio>
#include <cuda_runtime.h>
#include <cuda.h>
#include <cute/arch/copy_sm90_tma.hpp>
#include "utils.h"
__global__ void test_kernel(const __grid_constant__ CUtensorMap tensor_map) {
__shared__ int data[16][16];