 Jay Shah
|
eb9c0ee22a
add rmem -> gmem for fp8
|
vor 2 Monaten |
|
Jay Shah
|
551b91f4c9
uniform notation
|
vor 2 Monaten |
|
Jay Shah
|
7169b23399
unify rmem -> gmem methods
|
vor 2 Monaten |
|
Jay Shah
|
ab5d336e61
better writeout logic with vectorization
|
vor 2 Monaten |
|
Jay Shah
|
d437d3dd5c
remove smem usage for when rmem -> gmem epilogue is used
|
vor 2 Monaten |
 Ganesh Bikshandi
|
e49cb5f77c
passes except for hdim=256.
|
vor 2 Monaten |
|
Ganesh Bikshandi
|
dc2c952f37
compiles and builes. Not validates.
|
vor 2 Monaten |
|
Ganesh Bikshandi
|
a075e769fb
handle gqa_parallel with rmem-to-gmem. Not validating yet.
|
vor 2 Monaten |
|
Jay Shah
|
8f45a8cfa2
tests passing now for non-gqa impl
|
vor 2 Monaten |
|
Ganesh Bikshandi
|
f0b49460ec
changes to use tiledcopy (still not passing).
|
vor 2 Monaten |
|
Ganesh Bikshandi
|
8fbefa8ac4
adding rmem to gmem. (Not validating yet).
|
vor 2 Monaten |
|
Jay Shah
|
0a1a0c22b6
refactor for split kv
|
vor 3 Monaten |
 Tri Dao
|
bafe253042
[FA3] Bwd
|
vor 4 Monaten |
|
jayhshah
|
5018ac6ac5
Fp8 kernel with "in-kernel" transpose of V in producer (#1100)
|
vor 4 Monaten |
|
Tri Dao
|
3aae9c18c1
Revert "Changes For FP8 (#1075)"
|
vor 4 Monaten |
 ganeshcolfax
|
1899c970c8
Changes For FP8 (#1075)
|
vor 4 Monaten |
 Ying Zhang
|
dfe1a59e4b
Add var-seq-len to FA3 fp16 / bf16 fwd (#1072)
|
vor 4 Monaten |
 Cameron Shinn
|
cb516f855b
Remove torchlib dependency from cpp files (#1083)
|
vor 4 Monaten |
|
Tri Dao
|
7f67966cc7
FA3 initial code release
|
vor 5 Monaten |
