Releases: xlite-dev/LeetCUDA
Releases · xlite-dev/LeetCUDA
v3.0.5
What's Changed
- [Misc] Automated submodule update by @DefTruth in #261
- Update README.md by @tpoisonooo in #264
- Update README.md by @DefTruth in #265
- bugfix: only export per token softmax kernels by @DefTruth in #266
- misc: update vllm latest slides by @DefTruth in #267
- feat: add triton vector_add kernel by @DefTruth in #268
- feat: add triton merge_attn_states kernel by @DefTruth in #269
- feat: add cuda merge_attn_states kernel by @DefTruth in #270
- feat: update cuda merge_attn_states kernel by @DefTruth in #271
- misc: dispatch CUDA merge_attn_states by @DefTruth in #273
- misc: add triton kernel index by @DefTruth in #274
- Fix mistake on mat trans 2d when init grid. by @bear-zd in #275
- misc: update cuda merge_attn_states kernel by @DefTruth in #276
- kernel: optimize merge_attn_states CUDA kernel dispatch by @DefTruth in #278
- feat: optimize merge_attn_states thread block dispatch by @DefTruth in #279
New Contributors
- @tpoisonooo made their first contribution in #264
Full Changelog: v3.0.4...v3.0.5
Contributors
tpoisonooo, DefTruth, and bear-zd
Assets 2
v3.0.4
What's Changed
- [Docs] Add vLLM + DeepSeek-R1 671B deploy blog by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/259
Full Changelog: DefTruth/CUDA-Learn-Notes@v3.0.3...v3.0.4
Contributors
DefTruth
Assets 2
v3.0.3
What's Changed
- [Misc] Automated submodule update by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/257
Full Changelog: DefTruth/CUDA-Learn-Notes@v3.0.2...v3.0.3
Contributors
DefTruth
Assets 2
v3.0.2
What's Changed
- Fix typo in block_all_reduce.cu by @wplf in https://github.com/DefTruth/CUDA-Learn-Notes/pull/247
- fix typo about enougth by @wplf in https://github.com/DefTruth/CUDA-Learn-Notes/pull/248
- [FFPA] Add FFPA tech zhihu blog by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/252
- [FFPA] Update FFPA(Split-D) blog title by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/253
- [Misc] Automated submodule update by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/254
New Contributors
- @wplf made their first contribution in https://github.com/DefTruth/CUDA-Learn-Notes/pull/247
Full Changelog: DefTruth/CUDA-Learn-Notes@v3.0.1...v3.0.2
Contributors
DefTruth and wplf
Assets 2
v3.0.1
What's Changed
- [README] Update README.md by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/240
- [Bugfix] remove some error comments by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/241
- Update README.md by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/242
- Update README.md by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/243
- Update README.md by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/244
- [misc] Add hgemm-tensorcores-mma submodule✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/246
Full Changelog: DefTruth/CUDA-Learn-Notes@v3.0.0...v3.0.1
Contributors
DefTruth
Assets 2
v3.0.0
What's Changed
- [README] Add cuffpa-py library News🔥 by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/216
- [README] Update cuffpa-py library News🔥 by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/217
- [README] Update README.md by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/218
- [README] Update README.md by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/219
- [Misc] fix ffpa-attn-mma bench links typo by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/220
- [Misc] fix ffpa-attn-mma bench links typo by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/221
- [Misc] fix ffpa-attn-mma bench links typo by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/222
- [README] Update README by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/223
- [README] Add 🤖ffpa-attn-mma D=512 bench by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/224
- [submodule] Add ffpa-attn-mma submodule by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/225
- [misc] Update ffpa-attn-mma & cutlass submodules by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/226
- [FFPA] add ffpa-attn-mma kernels to lists by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/227
- [Misc] Update ffpa-attn-mma submodule by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/228
- [README] Update README.md by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/229
- Update README.md by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/230
- [Misc] update ffpa-attn-mma submodule by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/231
- [Misc] update ffpa-attn-mma submodule by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/233
- [Misc] update ffpa-attn-mma submodule by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/234
- [Misc] update ffpa-attn-mma submodule by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/236
- [Release] Bump up to v3.0.0 (#237) by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/238
Full Changelog: DefTruth/CUDA-Learn-Notes@v2.6.15...v3.0.0
Contributors
DefTruth
Assets 2
🔥FFPA L1 release
What's Changed
- [README] Add
cuffpa-pylibrary News🔥 by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/214 - [README] Add cuffpa-py library News🔥 by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/215
Full Changelog: DefTruth/CUDA-Learn-Notes@v2.6.14...v2.6.15
Contributors
DefTruth
Assets 2
QKV Fine-grained Tiling
What's Changed
- [ELU] support ELU F32/F16 kernel✔️ by @southkarl in https://github.com/DefTruth/CUDA-Learn-Notes/pull/194
- [HARDSHRINK][FP16] support HARDSHRINK F32/FP16 kernel by @southkarl in https://github.com/DefTruth/CUDA-Learn-Notes/pull/195
- [swizzle] update smem swizzle layout tools✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/196
- [swizzle] update smem swizzle layout tools✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/197
- [swizzle] add padding -> swizzle layout tools🎉 by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/198
- [HGEMM] HGEMM TN A&B SMEM Swizzle✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/199
- [HGEMM] HGEMM TN A&B SMEM Swizzle✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/200
- [FA2] shared-qkv + HMMA F32F16F16F32✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/201
- [HARDSWISH] HARDSWISH F32/F16 kernel✔️ by @southkarl in https://github.com/DefTruth/CUDA-Learn-Notes/pull/202
- [FA2] kOStorageAccFloat32 flag -> shared-qkv✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/203
- [FA2] kOStorageAccFloat32 -> share-qkv✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/204
- [FA2] kOStorageAccFloat32 -> share-qkv✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/205
- [FA2] share-kv + MMA F32F16F16F16F32✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/206
- [FA2] share-kv + MMA F32F16F16F16F32✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/207
- [FA2] tiling-kv + MMA F32F16F16F16F32✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/208
- [FA2] tiling-qkv + MMA F32F16F16F32✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/209
- [FA2] tiling-qkv + MMA F32F16F16F32✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/210
- [FA2] flash-attn-mma fully tiling-qkv🎉 by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/211
- [FA2] flash-attn-mma fully tiling-qkv🎉 by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/212
- [FA2] tiling-qkv F32/F16 + swizzle q/qk/qkv🎉 by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/213
New Contributors
- @southkarl made their first contribution in https://github.com/DefTruth/CUDA-Learn-Notes/pull/194
Full Changelog: DefTruth/CUDA-Learn-Notes@v2.6.13...v2.6.14
Contributors
DefTruth and southkarl
Assets 2
FA2 QKV SMEM Swizzle✔️
What's Changed
- [Doc] Refactor README for better readability✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/186
- [FA2] shared-kv + fully smem swizzle✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/187
- [FA2] tiling-qk + fully smem swizzle✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/188
- [FA2] shared-qkv + fully swizzle q/qk/qkv✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/189
- [FA2] hotfix launch setting -> shared-kv d=256✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/190
- [FA2] support shared-qkv + O s2g kernel✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/191
- [FA2] Update RTX 3080 Laptop performance✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/192
- [Doc] Refactor README for better readability✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/193
Full Changelog: DefTruth/CUDA-Learn-Notes@v2.6.12...v2.6.13
Contributors
DefTruth
Assets 2
🎉FA2/HGEMM SMEM Swizzle
What's Changed
- [FA2] split-q + tiling-qk D=512 performance🎉 by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/178
- [FA2] split-q + tiling-qk D=512 performance🎉 by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/179
- [FA2] split-q + tiling-qk D=512 performance🎉 by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/180
- [Doc] Refactor README.md to improve readability✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/181
- [Doc] Refactor README.md for better readability✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/182
- [FA2] flash-attn-mma 3080/L20/4090 bench✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/183
- [FA2] flash-attn-mma 3080/L20/4090 bench✔️ by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/184
- [FA2] fa2/hgemm manually smem swizzle🎉 by @DefTruth in https://github.com/DefTruth/CUDA-Learn-Notes/pull/185
flash_attn_mma_stages_split_q_tiling_qk_swizzle_kernel
void flash_attn_mma_stages_split_q_tiling_qk_swizzle_kernel<512, 16, 8, 16, 8, 1, 8, 1, 1, 16, 1, 64, 2, 0, 0, 8>(__half *, __half *, __half *, __half *, int, int) (8, 48, 1)x(256, 1, 1), Context 1, Stream 7, Device 0, CC 8.9
Section: Command line profiler metrics
------------------------------------------------------------------ ----------- ------------
Metric Name Metric Unit Metric Value
------------------------------------------------------------------ ----------- ------------
sm__sass_l1tex_data_bank_conflicts_pipe_lsu_mem_shared_op_ldsm.avg 0
sm__sass_l1tex_data_bank_conflicts_pipe_lsu_mem_shared_op_ldsm.max 0
sm__sass_l1tex_data_bank_conflicts_pipe_lsu_mem_shared_op_ldsm.min 0
sm__sass_l1tex_data_bank_conflicts_pipe_lsu_mem_shared_op_ldsm.sum 0
------------------------------------------------------------------ ----------- ------------Full Changelog: DefTruth/CUDA-Learn-Notes@v2.6.11...v2.6.12
Contributors
DefTruth