 Lu Fang
|
74aed78373
Replace c10::optional with std::optional in flash_attn
|
hace 4 días |
 Tri Dao
|
2ac6c986be
Fix Sm80 tile_count_semaphore, adjust test tolerance
|
hace 1 semana |
|
Tri Dao
|
e94f7e89dc
Always enable PackGQA is Split to reduce compilation and binary size
|
hace 1 semana |
|
Tri Dao
|
40fa35acd8
Always enable PackGQA if PagedKV to reduce compilation and bin size
|
hace 1 semana |
|
Tri Dao
|
a84a237d2a
Split bwd softcap compilation units for Sm80
|
hace 1 semana |
|
Tri Dao
|
1e3208566a
Tune tile sizes for compilation
|
hace 1 semana |
|
Tri Dao
|
84f1287e42
Rename bool_constant<true> to true_type, same w bool_constant<false>
|
hace 1 semana |
|
Tri Dao
|
df5fe55264
Change tile sizes for Sm8x to reduce stack frame
|
hace 1 semana |
|
Tri Dao
|
8dd0b479d5
Always enable PackGQA for Sm8x to reduce compilation and binary size
|
hace 1 semana |
|
Tri Dao
|
7bc3f031a4
Compile for both Sm80 and Sm90
|
hace 1 semana |
|
Tri Dao
|
7a802796e1
Big refactor and update
|
hace 1 semana |
 sclarkson
|
1feb711f46
Fix compilation with clang on ARM64 (#1285)
|
hace 1 mes |
 Kai Londenberg
|
b443207c1f
Paged Attention support for FA3 (#1268)
|
hace 2 meses |
 jayhshah
|
a5a75274bc
FA3 kvcache + split kv + gqa parallelization (#1236)
|
hace 3 meses |
 Ying Zhang
|
dff976a84a
fixes
|
hace 4 meses |
|
Ying Zhang
|
7b4e68e04f
hopper local attention
|
hace 4 meses |
|
Ying Zhang
|
db80387343
Add seqused_q in fwd / bwd and seqused_k in bwd.
|
hace 4 meses |
|
jayhshah
|
c92ca63268
FA3 FP8 qkv descales + restore max offset for h128 causal + added sync for producer WG (#1173)
|
hace 4 meses |
|
Tri Dao
|
5d5bfbb619
Remove contiguous checks
|
hace 5 meses |
|
Tri Dao
|
bafe253042
[FA3] Bwd
|
hace 5 meses |
|
jayhshah
|
5018ac6ac5
Fp8 kernel with "in-kernel" transpose of V in producer (#1100)
|
hace 5 meses |
|
Tri Dao
|
3aae9c18c1
Revert "Changes For FP8 (#1075)"
|
hace 5 meses |
 ganeshcolfax
|
1899c970c8
Changes For FP8 (#1075)
|
hace 5 meses |
 Ying Zhang
|
dfe1a59e4b
Add var-seq-len to FA3 fp16 / bf16 fwd (#1072)
|
hace 5 meses |
|
Tri Dao
|
74b0761ff7
[FA3] BF16 forward
|
hace 6 meses |
|
Tri Dao
|
7f67966cc7
FA3 initial code release
|
hace 6 meses |
