[Kernel] Add B12X causal paged attention backend (#52017)
Signed-off-by: Luke Alonso <lalonso@gmail.com> Co-authored-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com>
This commit is contained in:
co-authored by
Lucas Wilkinson
parent
adebc41b7e
commit
0d4ad47981
@@ -34,6 +34,7 @@ def _import_submodule(module_name: str) -> ModuleType | None:
|
||||
_B12X_SUBMODULES = {
|
||||
module_name: _import_submodule(module_name)
|
||||
for module_name in (
|
||||
"b12x.attention.paged",
|
||||
"b12x.gemm.blockscaled",
|
||||
# TODO: Remove once B12X exposes the scale-swizzle API publicly.
|
||||
"b12x._lib.intrinsics",
|
||||
@@ -73,6 +74,10 @@ def get_b12x_fused_moe() -> ModuleType | None:
|
||||
return _get_submodule("b12x.moe.fused_moe")
|
||||
|
||||
|
||||
def get_b12x_paged_attention() -> ModuleType | None:
|
||||
return _get_submodule("b12x.attention.paged")
|
||||
|
||||
|
||||
def b12x_warmup_token_counts(
|
||||
*,
|
||||
max_tokens: int,
|
||||
|
||||
Reference in New Issue
Block a user