chenchenghao/DeepGEMM
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Refactor JIT compilation (+NVRTC support) (#94)

Browse Source 
* [wip] refactor: compile to .cubin

Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>

* refactor: compile to .cubin and add NVRTC option

Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>

* fix: compiler version

Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>

* feat: compat for old drivers

Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>

* feat: save kernel name to file

Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>

* feat: fix win compat

Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>

* fix: windows compat

Signed-off-by: Gabriel Wu <13583761+lucifer1004@users.noreply.github.com>

* feat: make API more general

Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>

* feat: drop support for CUDA<12.3

Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>

* doc: update README

Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>

* Some lints and refactor

* Refactor runtime

* Several fixes

* Refactor environment variables

* Code format

* Add a TODO

* Compatible with CUDA 12.3

* Fix indent

* Fix typing

* Drop support for Windows

* Add a TODO

---------

Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>
Signed-off-by: Gabriel Wu <13583761+lucifer1004@users.noreply.github.com>
Co-authored-by: Chenggang Zhao <chenggangz@deepseek.com>
This commit is contained in:
Gabriel Wu
2025-05-07 11:38:14 +08:00
committed by GitHub
co-authored by Chenggang Zhao
parent d374456787
commit bfe983c4c2
19 changed files with 909 additions and 660 deletions
Download Patch File Download Diff File Expand all files Collapse all files
deep_gemm/include/deep_gemm/scheduler.cuh
+3 -2
View File
@@ -46,7 +46,7 @@ struct Scheduler {
}
}
__device__ __forceinline__ bool is_tma_multicast_valid(const uint32_t& m_block_idx) {
__device__ __forceinline__ bool is_tma_multicast_valid(const uint32_t& m_block_idx) const {
if (num_blocks_in_group == 1)
return false;
if constexpr (kGemmType == GemmType::Normal or kGemmType == GemmType::GroupedMasked) {
@@ -63,7 +63,8 @@ struct Scheduler {
}
}
__device__ __forceinline__ void get_swizzled_block_idx(const uint32_t num_m_blocks, int block_idx, uint32_t& m_block_idx, uint32_t& n_block_idx) {
__device__ __forceinline__ void get_swizzled_block_idx(const uint32_t num_m_blocks, int block_idx,
uint32_t& m_block_idx, uint32_t& n_block_idx) {
DG_STATIC_ASSERT(kNum1DBlocksPerGroup % kNumTMAMulticast == 0, "Invalid group size");
// Swizzle for better L2 usages