[ROCm] [Bugfix] Fix Triton fused shared expert alignment (#51632)

Signed-off-by: Aakif Nawaz <aakif.nawaz@amd.com>
This commit is contained in:
akii96
2026-08-20 00:14:52 +08:00
committed by GitHub
parent 525b7bbb3a
commit 583a00257d
2 changed files with 50 additions and 2 deletions
+44
View File
@@ -459,6 +459,50 @@ def test_fused_moe(
)
def test_fused_shared_expert_alignment(workspace_init):
set_random_seed(7)
m, n, k = 4, 64, 128
routed_experts = 8
physical_experts = routed_experts + 1
dtype = torch.bfloat16
a = torch.randn((m, k), device=DEVICE_TYPE, dtype=dtype) / 10
w1 = torch.randn((physical_experts, 2 * n, k), device=DEVICE_TYPE, dtype=dtype) / 10
w2 = torch.randn((physical_experts, k, n), device=DEVICE_TYPE, dtype=dtype) / 10
topk_ids = torch.tensor(
[[0, 8], [1, 8], [2, 8], [3, 8]], device=DEVICE_TYPE, dtype=torch.int32
)
topk_weights = torch.tensor(
[[0.5, 1.0]] * m, device=DEVICE_TYPE, dtype=torch.float32
)
moe_config = make_dummy_moe_config(
num_experts=physical_experts,
experts_per_token=2,
hidden_dim=k,
intermediate_size=n,
in_dtype=dtype,
max_num_tokens=m,
)
modular_moe = modular_triton_fused_moe(moe_config, FUSED_MOE_UNQUANTIZED_CONFIG)
with set_current_vllm_config(vllm_config):
expected = torch_experts(a, w1, w2, topk_weights, topk_ids)
actual = modular_moe.apply(
hidden_states=a,
w1=w1,
w2=w2,
topk_weights=topk_weights,
topk_ids=topk_ids,
activation=MoEActivation.SILU,
global_num_experts=routed_experts,
expert_map=None,
apply_router_weight_on_input=False,
)
torch.testing.assert_close(actual, expected, atol=2e-2, rtol=0)
def test_fused_moe_int64_overflow(workspace_init):
"""Regression test for int32 overflow in stride*offset products.
@@ -315,13 +315,15 @@ class TritonExperts(LoRAExpertsMixin, mk.FusedMoEExpertsModular):
)
intermediate_cache3 = _resize_cache(workspace2, (num_tokens, top_k_num, K))
# Include fused shared-expert rows while preserving EP remapping.
num_align_experts = w1.shape[0] if expert_map is None else global_num_experts
sorted_token_ids, expert_ids, num_tokens_post_padded = (
_prepare_expert_assignment(
topk_ids,
config,
num_tokens,
top_k_num,
global_num_experts,
num_align_experts,
expert_map,
use_int8_w8a16=self.quant_config.use_int8_w8a16,
use_int4_w4a16=self.quant_config.use_int4_w4a16,
@@ -688,8 +690,10 @@ class TritonWNA16Experts(TritonExperts):
)
intermediate_cache3 = _resize_cache(workspace2, (num_tokens, top_k_num, K))
# Include fused shared-expert rows while preserving EP remapping.
num_align_experts = w1.shape[0] if expert_map is None else global_num_experts
sorted_token_ids, expert_ids, num_tokens_post_padded = moe_align_block_size(
topk_ids, config["BLOCK_SIZE_M"], global_num_experts, expert_map
topk_ids, config["BLOCK_SIZE_M"], num_align_experts, expert_map
)
invoke_fused_moe_wna16_triton_kernel(