[Perf][Model] Qwen4Exp QSA: sm_90 tuning table for _select_config (#57273)
Signed-off-by: Shuolei Wang <shuoleiwang123@gmail.com> Signed-off-by: Shuolei Wang <948904026@qq.com> Co-authored-by: Roger Wang <hey@rogerw.io>
This commit is contained in:
co-authored by
Roger Wang
parent
bd2d7ae7c1
commit
7bbce752b8
@@ -22,6 +22,12 @@ def _is_sm120() -> bool:
|
|||||||
return current_platform.get_device_capability() == (12, 0)
|
return current_platform.get_device_capability() == (12, 0)
|
||||||
|
|
||||||
|
|
||||||
|
@lru_cache(maxsize=1)
|
||||||
|
def _is_sm90() -> bool:
|
||||||
|
"""True on sm_90 (H100/H200/H20): selects the sm_90 tuning table."""
|
||||||
|
return current_platform.get_device_capability() == (9, 0)
|
||||||
|
|
||||||
|
|
||||||
@triton.jit(do_not_specialize=["num_rows", "num_requests"])
|
@triton.jit(do_not_specialize=["num_rows", "num_requests"])
|
||||||
def _qsa_sparse_paged_gqa_splitk_kernel(
|
def _qsa_sparse_paged_gqa_splitk_kernel(
|
||||||
q_ptr,
|
q_ptr,
|
||||||
@@ -509,6 +515,33 @@ def _select_sm120_config(
|
|||||||
return 32, 4, 1
|
return 32, 4, 1
|
||||||
|
|
||||||
|
|
||||||
|
def _select_sm90_config(
|
||||||
|
base_programs: int, use_prefill_config: bool, is_fp8: bool
|
||||||
|
) -> tuple[int, int, int]:
|
||||||
|
"""(block_n, target_splits, num_warps) tuned on sm_90 (H20)."""
|
||||||
|
if base_programs > 2048:
|
||||||
|
return (32, 1, 1) if use_prefill_config else (64, 1, 2)
|
||||||
|
if is_fp8:
|
||||||
|
if base_programs <= 24:
|
||||||
|
return 64, 64, 2
|
||||||
|
if base_programs <= 32:
|
||||||
|
return 32, 16, 1
|
||||||
|
if base_programs <= 64:
|
||||||
|
return 32, 8, 1
|
||||||
|
if base_programs <= 128:
|
||||||
|
return 32, 4, 1
|
||||||
|
if base_programs <= 256:
|
||||||
|
return 32, 8, 1
|
||||||
|
return 32, 4, 1
|
||||||
|
if base_programs <= 32:
|
||||||
|
return 64, 64, 2
|
||||||
|
if base_programs <= 64:
|
||||||
|
return 32, 16, 1
|
||||||
|
if base_programs <= 256:
|
||||||
|
return 32, 8, 1
|
||||||
|
return 32, 4, 1
|
||||||
|
|
||||||
|
|
||||||
def _select_config(
|
def _select_config(
|
||||||
num_rows: int,
|
num_rows: int,
|
||||||
num_kv_heads: int,
|
num_kv_heads: int,
|
||||||
@@ -521,13 +554,18 @@ def _select_config(
|
|||||||
Keyed on base_programs = num_rows * num_kv_heads. The bp > 2048 region splits
|
Keyed on base_programs = num_rows * num_kv_heads. The bp > 2048 region splits
|
||||||
on use_prefill_config (capture-stable: at FULL-graph capture max_query_len is
|
on use_prefill_config (capture-stable: at FULL-graph capture max_query_len is
|
||||||
the uniform decode/verify length). This default table was tuned on GB300;
|
the uniform decode/verify length). This default table was tuned on GB300;
|
||||||
sm_120 (RTX PRO 6000 Blackwell) dispatches to _select_sm120_config instead.
|
sm_120 (RTX PRO 6000 Blackwell) dispatches to _select_sm120_config instead,
|
||||||
|
and sm_90 (Hopper) to _select_sm90_config.
|
||||||
"""
|
"""
|
||||||
base_programs = num_rows * num_kv_heads
|
base_programs = num_rows * num_kv_heads
|
||||||
if _is_sm120():
|
if _is_sm120():
|
||||||
BLOCK_N, target_splits, num_warps = _select_sm120_config(
|
BLOCK_N, target_splits, num_warps = _select_sm120_config(
|
||||||
base_programs, use_prefill_config, is_fp8
|
base_programs, use_prefill_config, is_fp8
|
||||||
)
|
)
|
||||||
|
elif _is_sm90():
|
||||||
|
BLOCK_N, target_splits, num_warps = _select_sm90_config(
|
||||||
|
base_programs, use_prefill_config, is_fp8
|
||||||
|
)
|
||||||
elif base_programs > 2048:
|
elif base_programs > 2048:
|
||||||
BLOCK_N, target_splits, num_warps = (
|
BLOCK_N, target_splits, num_warps = (
|
||||||
(32, 1, 1) if use_prefill_config else (64, 1, 2)
|
(32, 1, 1) if use_prefill_config else (64, 1, 2)
|
||||||
|
|||||||
Reference in New Issue
Block a user