Tags: xlite-dev/ffpa-attn
Toggle v0.0.2.post4's commit message
Toggle v0.0.2.post3's commit message
[docs] Add FFPA(Split-D) tech blog link (#77 )
* Update README.md
* Update README.md
Toggle v0.0.2.post2's commit message
[tests] rename test.py -> test_ffpa_attn.py (#72 )
* Rename test_fake_fused_mla.py to test_fused_mla.py
* Rename test.py to test_ffpa_attn.py
* Update README.md
* Update README.md
* Update README.md
* Update README.md
* Update README.md
* Update README.md
* Update README.md
* Update README.md
* Update README.md
* Update README.md
* Create fused_mla_api.cc
Toggle v0.0.2.post1's commit message
[feat] support ffpa-l1 registers double buffers (#70 )
* Update README.md
* Update README.md
* Update env.py
* Update prefill.cuh
* Update ffpa_attn_templates_L1.cuh
* Update launch_templates.cuh
* Update README.md
Toggle v0.0.2's commit message
[Release] Bump up to v0.0.2 (#61 )
* Update setup.py
* Update version.py
Toggle v0.0.1.post3's commit message
[bench] update perf plots for qkv swizzle (#40 )
Toggle v0.0.1.post2's commit message
[misc] fix bench link typos (#35 )
Toggle v0.0.1.post1's commit message
[FFPA] fix some macro typos (#21 )
* Update faster_prefill_attn_F16F16F16F16_L1.cu
* Update faster_prefill_attn_F32F16F16F32_L1.cu
Toggle v0.0.1's commit message
You can’t perform that action at this time.