[FEAT] [ROCm] [V1]: Add AITER biased group topk for DeepSeekV3 (#17955)

Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
Co-authored-by: tjtanaa <tunjian.tan@embeddedllm.com>

tests/kernels/moe/test_rocm_aiter_topk.py

+# SPDX-License-Identifier: Apache-2.0
+# This is a test for the AITER ops.
+# It tests if the AITER ops are
+# 1. correctly registered as custom ops
+# 2. correctly defined the relationship between
+#    implementation and fake function
+# 3. can be used with torch.compile
+# This file will be skipped if AITER is not installed
+# and the platform is not ROCm.
+
+import importlib.util
+
+import pytest
+import torch
+
+# this import statement is needed to ensure the ops are registered
+import vllm.model_executor.layers.fused_moe.rocm_aiter_fused_moe  # noqa: F401
+from vllm.platforms import current_platform
+
+# need to import once to ensure the ops are registered
+# Check if aiter package is installed
+aiter_available = importlib.util.find_spec("aiter") is not None
+
+pytestmark = pytest.mark.skipif(
+    not (current_platform.is_rocm() and aiter_available),
+    reason="AITER ops are only available on ROCm with aiter package installed")
+
+
+def test_rocm_aiter_biased_grouped_topk_custom_op_registration():
+    """Test that the custom op is correctly registered."""
+    # Check if the op exists in torch.ops.vllm
+    assert hasattr(torch.ops.vllm, 'rocm_aiter_biased_grouped_topk')
+
+    # Check if the op is callable
+    assert callable(torch.ops.vllm.rocm_aiter_biased_grouped_topk)
+
+
+def test_rocm_aiter_biased_grouped_topk_torch_compile_compatibility():
+    """Test that the op can be used with torch.compile."""
+    # Create test tensors
+    token = 64
+    expert = 256
+    num_expert_group = 8
+    topk = 8
+    topk_group = 4
+    renormalize = True
+    scale_factor = 1.0
+
+    gating_output = torch.randn((token, expert),
+                                dtype=torch.bfloat16,
+                                device="cuda")
+    e_score_correction_bias = torch.randn((expert, ),
+                                          dtype=torch.bfloat16,
+                                          device="cuda")
+
+    device = gating_output.device
+    topk_ids = torch.empty((token, topk), dtype=torch.int32, device=device)
+    topk_weights = torch.empty((token, topk),
+                               dtype=torch.float32,
+                               device=device)
+
+    # Define a function that uses the op
+    def biased_grouped_topk_fn(gating_output, e_score_correction_bias,
+                               topk_weights, topk_ids):
+        return torch.ops.vllm.rocm_aiter_biased_grouped_topk(
+            gating_output, e_score_correction_bias, topk_weights, topk_ids,
+            num_expert_group, topk_group, renormalize, scale_factor)
+
+    # Verify the op's fake implementation
+    torch.library.opcheck(
+        torch.ops.vllm.rocm_aiter_biased_grouped_topk,
+        (gating_output, e_score_correction_bias, topk_weights, topk_ids),
+        kwargs={
+            "num_expert_group": num_expert_group,
+            "topk_group": topk_group,
+            "need_renorm": renormalize,
+            "routed_scaling_factor": scale_factor
+        },
+        test_utils=("test_faketensor"))
+
+    # Compile the function with appropriate settings
+    compiled_fn = torch.compile(biased_grouped_topk_fn,
+                                fullgraph=True,
+                                backend="inductor",
+                                mode="reduce-overhead",
+                                dynamic=False)
+
+    topk_weights_original = torch.empty((token, topk),
+                                        dtype=torch.float32,
+                                        device=device)
+    topk_ids_original = torch.empty((token, topk),
+                                    dtype=torch.int32,
+                                    device=device)
+
+    topk_weights_compiled = torch.empty((token, topk),
+                                        dtype=torch.float32,
+                                        device=device)
+    topk_ids_compiled = torch.empty((token, topk),
+                                    dtype=torch.int32,
+                                    device=device)
+
+    # Run both compiled (V1 graph mode) and uncompiled versions (V1 eager mode)
+    biased_grouped_topk_fn(gating_output, e_score_correction_bias,
+                           topk_weights_original, topk_ids_original)
+    compiled_fn(gating_output, e_score_correction_bias, topk_weights_compiled,
+                topk_ids_compiled)
+
+    # Sort the results for comparison since the order might not be deterministic
+    topk_ids_original, indices_original = torch.sort(topk_ids_original)
+    topk_weights_original = torch.gather(topk_weights_original, 1,
+                                         indices_original)
+
+    topk_ids_compiled, indices_compiled = torch.sort(topk_ids_compiled)
+    topk_weights_compiled = torch.gather(topk_weights_compiled, 1,
+                                         indices_compiled)
+
+    # Verify results match
+    assert torch.allclose(topk_weights_original,
+                          topk_weights_compiled,
+                          rtol=1e-2,
+                          atol=1e-2)
+    assert torch.allclose(topk_ids_original, topk_ids_compiled)

vllm/model_executor/layers/fused_moe/layer.py

@@ -17,6 +17,8 @@ from vllm.distributed import (get_dp_group, get_ep_group,
 from vllm.forward_context import ForwardContext, get_forward_context
 from vllm.logger import init_logger
 from vllm.model_executor.custom_op import CustomOp
+from vllm.model_executor.layers.fused_moe.rocm_aiter_fused_moe import (
+    is_rocm_aiter_moe_enabled)
 from vllm.model_executor.layers.quantization.base_config import (
     QuantizationConfig, QuantizeMethodBase)
 from vllm.model_executor.utils import set_weight_attrs
@@ -28,6 +30,11 @@ if current_platform.is_cuda_alike():
     from .fused_moe import fused_experts
 else:
     fused_experts = None  # type: ignore
+if is_rocm_aiter_moe_enabled():
+    from vllm.model_executor.layers.fused_moe.rocm_aiter_fused_moe import (  # noqa: E501
+        rocm_aiter_biased_group_topk as grouped_topk)
+else:
+    from vllm.model_executor.layers.fused_moe.fused_moe import grouped_topk
 if current_platform.is_tpu():
     # the iterative moe implementation is used until the moe_pallas is fixed
     from .moe_torch_iterative import fused_moe as fused_moe_pallas
@@ -802,8 +809,7 @@ class FusedMoE(torch.nn.Module):
                        custom_routing_function: Optional[Callable] = None,
                        scoring_func: str = "softmax",
                        e_score_correction_bias: Optional[torch.Tensor] = None):
-        from vllm.model_executor.layers.fused_moe.fused_moe import (
-            fused_topk, grouped_topk)
+        from vllm.model_executor.layers.fused_moe.fused_moe import fused_topk
 
         # DeekSeekv2 uses grouped_top_k
         if use_grouped_topk:

vllm/model_executor/layers/fused_moe/rocm_aiter_fused_moe.py

@@ -216,6 +216,37 @@ def rocm_aiter_topk_softmax_fake(topk_weights: torch.Tensor,
     pass
 
 
+def rocm_aiter_biased_grouped_topk_impl(
+        gating_output: torch.Tensor,
+        correction_bias: torch.Tensor,
+        topk_weights: torch.Tensor,
+        topk_ids: torch.Tensor,
+        num_expert_group: int,
+        topk_group: int,
+        need_renorm: bool,
+        routed_scaling_factor: float = 1.0  # mul to topk_weights
+) -> None:
+
+    from aiter import biased_grouped_topk
+
+    biased_grouped_topk(gating_output, correction_bias, topk_weights, topk_ids,
+                        num_expert_group, topk_group, need_renorm,
+                        routed_scaling_factor)
+
+
+def rocm_aiter_biased_grouped_topk_fake(
+        gating_output: torch.Tensor,
+        correction_bias: torch.Tensor,
+        topk_weights: torch.Tensor,
+        topk_ids: torch.Tensor,
+        num_expert_group: int,
+        topk_group: int,
+        need_renorm: bool,
+        routed_scaling_factor: float = 1.0  # mul to topk_weights
+) -> None:
+    pass
+
+
 if current_platform.is_rocm():
 
     direct_register_custom_op(
@@ -258,6 +289,46 @@ if current_platform.is_rocm():
         dispatch_key=current_platform.dispatch_key,
     )
 
+    direct_register_custom_op(
+        op_name="rocm_aiter_biased_grouped_topk",
+        op_func=rocm_aiter_biased_grouped_topk_impl,
+        mutates_args=["topk_weights", "topk_ids"],
+        fake_impl=rocm_aiter_biased_grouped_topk_fake,
+        dispatch_key=current_platform.dispatch_key,
+    )
+
+
+def rocm_aiter_biased_group_topk(
+    hidden_states: torch.Tensor,
+    gating_output: torch.Tensor,
+    topk: int,
+    renormalize: bool,
+    num_expert_group: int = 0,
+    topk_group: int = 0,
+    scoring_func: str = "sigmoid",
+    e_score_correction_bias: Optional[torch.Tensor] = None
+) -> tuple[torch.Tensor, torch.Tensor]:
+    assert scoring_func == "sigmoid", (
+        "rocm_aiter_biased_group_topk only supports 'sigmoid' scoring_func.")
+    assert e_score_correction_bias is not None, (
+        "'e_score_correction_bias' must not be None.")
+    token = hidden_states.shape[0]
+    device = hidden_states.device
+    topk_ids = torch.empty((token, topk), dtype=torch.int32, device=device)
+    topk_weights = torch.empty((token, topk),
+                               dtype=torch.float32,
+                               device=device)
+    torch.ops.vllm.rocm_aiter_biased_grouped_topk(
+        gating_output,
+        e_score_correction_bias,
+        topk_weights,
+        topk_ids,
+        num_expert_group,
+        topk_group,
+        renormalize,
+    )
+    return topk_weights, topk_ids
+
 
 def rocm_aiter_fused_experts(hidden_states: torch.Tensor,
                              w1: torch.Tensor,