Add AWQ quantization support for NPU. (#10158)

Co-authored-by: Alisehen <814073252@qq.com>
Co-authored-by: Yaochen Han <48639761+Alisehen@users.noreply.github.com>
Co-authored-by: Zhengda Qin <zhengdqin@gmail.com>

python/sglang/srt/layers/linear.py

@@ -51,6 +51,7 @@ WEIGHT_LOADER_V2_SUPPORTED = [
     "CompressedTensorsLinearMethod",
     "AWQMarlinLinearMethod",
     "AWQLinearMethod",
+    "AWQLinearAscendMethod",
     "GPTQMarlinLinearMethod",
     "Fp8LinearMethod",
     "BlockInt8LinearMethod",

python/sglang/srt/layers/quantization/awq.py

@@ -31,6 +31,7 @@ from sglang.srt.layers.quantization.marlin_utils import (
 )
 from sglang.srt.layers.quantization.unquant import UnquantizedLinearMethod
 from sglang.srt.layers.quantization.utils import get_scalar_types, replace_parameter
+from sglang.srt.layers.quantization.w8a8_int8 import npu_fused_experts
 
 if TYPE_CHECKING:
     from sglang.srt.layers.moe.moe_runner import MoeRunnerConfig
@@ -39,11 +40,16 @@ if TYPE_CHECKING:
         CombineInput,
     )
 
-from sglang.srt.utils import is_cuda, is_hip, is_xpu
+from sglang.srt.utils import is_cuda, is_hip, is_npu, is_xpu
 
 _is_cuda = is_cuda()
 _is_hip = is_hip()
 _is_xpu = is_xpu()
+_is_npu = is_npu()
+
+if _is_npu:
+    import torch_npu
+
 if _is_cuda:
     from sgl_kernel import (
         awq_dequantize,
@@ -117,12 +123,17 @@ class AWQConfig(QuantizationConfig):
         return "awq"
 
     def get_supported_act_dtypes(self) -> List[torch.dtype]:
-        return [torch.half]
+        return [torch.float16] if not _is_npu else [torch.float16, torch.bfloat16]
 
     @classmethod
     def get_min_capability(cls) -> int:
         # The AWQ kernel only supports Turing or newer GPUs.
-        return 75
+        if _is_npu:
+            raise NotImplementedError(
+                'NPU hardware does not support "get_min_capability" feature.'
+            )
+        else:
+            return 75
 
     @staticmethod
     def get_config_filenames() -> List[str]:
@@ -146,6 +157,16 @@ class AWQConfig(QuantizationConfig):
         self, layer: torch.nn.Module, prefix: str
     ) -> Optional[LinearMethodBase]:
         from sglang.srt.layers.linear import LinearBase
+        from sglang.srt.layers.moe.fused_moe_triton import FusedMoE
+
+        if _is_npu:
+            if isinstance(layer, LinearBase):
+                if is_layer_skipped_awq(prefix, self.modules_to_not_convert):
+                    return UnquantizedLinearMethod()
+                return AWQLinearAscendMethod(self)
+            elif isinstance(layer, FusedMoE):
+                return AWQMoEAscendMethod(self)
+            return None
 
         if isinstance(layer, LinearBase):
             if is_layer_skipped_awq(prefix, self.modules_to_not_convert):
@@ -575,6 +596,64 @@ class AWQMarlinLinearMethod(LinearMethodBase):
         )
 
 
+class AWQLinearAscendMethod(AWQLinearMethod):
+    """Linear method for AWQ on Ascend.
+
+    Args:
+        quant_config: The AWQ quantization config.
+    """
+
+    def process_weights_after_loading(self, layer: torch.nn.Module) -> None:
+        layer.scales = torch.nn.Parameter(layer.scales.data, requires_grad=False)
+        qweight_tmp = torch.zeros_like(layer.qweight.data)
+        qzeros_tmp = layer.qzeros.data
+        qzeros_list = []
+        shifts = [0, 4, 1, 5, 2, 6, 3, 7]
+
+        for i in range(0, self.quant_config.pack_factor):
+            shift_num = shifts[i] * 4
+            qzeros_list.append((qzeros_tmp.reshape(-1, 1) >> shift_num) & 0xF)
+            qweight_tmp.bitwise_or_(
+                ((layer.qweight.data >> shift_num) * (2 ** (4 * i))) & (0xF << (4 * i))
+            )
+
+        qweight_tmp.bitwise_xor_(0x88888888)
+
+        qzeros_tmp = torch.cat(qzeros_list, dim=-1).reshape(qzeros_tmp.shape[0], -1)
+        qzeros_tmp = -(qzeros_tmp - 8)
+        qzeros_tmp = qzeros_tmp.to(layer.scales.data.dtype)
+
+        layer.qzeros = torch.nn.Parameter(qzeros_tmp, requires_grad=False)
+        layer.qweight = torch.nn.Parameter(qweight_tmp, requires_grad=False)
+
+    def apply(
+        self,
+        layer: torch.nn.Module,
+        x: torch.Tensor,
+        bias: Optional[torch.Tensor] = None,
+    ) -> torch.Tensor:
+        qweight = layer.qweight
+        scales = layer.scales
+        qzeros = layer.qzeros
+        pack_factor = self.quant_config.pack_factor
+        out_shape = x.shape[:-1] + (qweight.shape[-1] * pack_factor,)
+        reshaped_x = x.reshape(-1, x.shape[-1])
+
+        if bias is not None and bias.dtype == torch.bfloat16:
+            bias = bias.float()
+
+        out = torch_npu.npu_weight_quant_batchmatmul(
+            reshaped_x,
+            qweight,
+            antiquant_scale=scales,
+            antiquant_offset=qzeros,
+            antiquant_group_size=self.quant_config.group_size,
+            bias=bias,
+        )
+
+        return out.reshape(out_shape)
+
+
 class AWQMoEMethod(FusedMoEMethodBase):
 
     def __init__(self, quant_config: AWQMarlinConfig):
@@ -677,7 +756,8 @@ class AWQMoEMethod(FusedMoEMethodBase):
         set_weight_attrs(w2_qzeros, extra_weight_attrs)
 
         device = layer.w13_qweight.device
-        layer.workspace = marlin_make_workspace(device, 4)
+        if not _is_npu:
+            layer.workspace = marlin_make_workspace(device, 4)
 
     def process_weights_after_loading(self, layer: torch.nn.Module) -> None:
         num_experts = layer.w13_qweight.shape[0]
@@ -785,3 +865,95 @@ class AWQMoEMethod(FusedMoEMethodBase):
             num_bits=self.quant_config.weight_bits,
         ).to(orig_dtype)
         return StandardCombineInput(hidden_states=output)
+
+
+class AWQMoEAscendMethod(AWQMoEMethod):
+    def __init__(self, quant_config: AWQConfig):
+        self.quant_config = quant_config
+
+    def process_weights_after_loading(self, layer: torch.nn.Module) -> None:
+        w13_qweight_tmp = torch.zeros_like(layer.w13_qweight.data)
+        w2_qweight_tmp = torch.zeros_like(layer.w2_qweight.data)
+        w13_qzeros_list = []
+        w2_qzeros_list = []
+        shifts = [0, 4, 1, 5, 2, 6, 3, 7]
+        for i in range(0, self.quant_config.pack_factor):
+            shift_num = shifts[i] * 4
+            w13_qzeros_list.append(
+                (layer.w13_qzeros.data.reshape(-1, 1) >> shift_num) & 0xF
+            )
+            w2_qzeros_list.append(
+                (layer.w2_qzeros.data.reshape(-1, 1) >> shift_num) & 0xF
+            )
+            w13_qweight_tmp.bitwise_or_(
+                ((layer.w13_qweight.data >> shift_num) * (2 ** (4 * i)))
+                & (0xF << (4 * i))
+            )
+            w2_qweight_tmp.bitwise_or_(
+                ((layer.w2_qweight.data >> shift_num) * (2 ** (4 * i)))
+                & (0xF << (4 * i))
+            )
+
+        w13_qweight_tmp.bitwise_xor_(0x88888888)
+        w2_qweight_tmp.bitwise_xor_(0x88888888)
+
+        w13_qzeros_tmp = torch.cat(w13_qzeros_list, dim=-1).reshape(
+            layer.w13_qzeros.shape[0], layer.w13_qzeros.shape[1], -1
+        )
+        w13_qzeros_tmp = -(w13_qzeros_tmp - 8)
+        w13_qzeros_tmp = w13_qzeros_tmp.to(layer.w13_scales.data.dtype)
+        w2_qzeros_tmp = torch.cat(w2_qzeros_list, dim=-1).reshape(
+            layer.w2_qzeros.shape[0], layer.w2_qzeros.shape[1], -1
+        )
+        w2_qzeros_tmp = -(w2_qzeros_tmp - 8)
+        w2_qzeros_tmp = w2_qzeros_tmp.to(layer.w2_scales.data.dtype)
+
+        layer.register_parameter(
+            "w13_qzeros", torch.nn.Parameter(w13_qzeros_tmp, requires_grad=False)
+        )
+        layer.register_parameter(
+            "w13_qweight", torch.nn.Parameter(w13_qweight_tmp, requires_grad=False)
+        )
+        layer.register_parameter(
+            "w2_qzeros", torch.nn.Parameter(w2_qzeros_tmp, requires_grad=False)
+        )
+        layer.register_parameter(
+            "w2_qweight", torch.nn.Parameter(w2_qweight_tmp, requires_grad=False)
+        )
+
+    def create_moe_runner(
+        self, layer: torch.nn.Module, moe_runner_config: MoeRunnerConfig
+    ):
+        self.moe_runner_config = moe_runner_config
+
+    def apply(
+        self,
+        layer: torch.nn.Module,
+        dispatch_output: StandardDispatchOutput,
+    ) -> torch.Tensor:
+        from sglang.srt.layers.moe.token_dispatcher import StandardCombineInput
+
+        assert (
+            self.moe_runner_config.activation == "silu"
+        ), "Only SiLU activation is supported."
+
+        x = dispatch_output.hidden_states
+        topk_output = dispatch_output.topk_output
+
+        topk_weights, topk_ids, _ = topk_output
+        topk_ids = topk_ids.to(torch.int32)
+        topk_weights = topk_weights.to(x.dtype)
+        output = npu_fused_experts(
+            hidden_states=x,
+            w13=layer.w13_qweight,
+            w13_scale=layer.w13_scales,
+            w13_offset=layer.w13_qzeros,
+            w2=layer.w2_qweight,
+            w2_scale=layer.w2_scales,
+            w2_offset=layer.w2_qzeros,
+            topk_weights=topk_weights,
+            topk_ids=topk_ids,
+            top_k=topk_ids.shape[1],
+            use_wna16=True,
+        )
+        return StandardCombineInput(hidden_states=output)

python/sglang/srt/layers/quantization/awq_triton.py

@@ -337,3 +337,32 @@ def awq_gemm_triton(
     result = result.sum(0)
 
     return result
+
+
+def awq_dequantize_decomposition(
+    qweight: torch.Tensor,
+    scales: torch.Tensor,
+    zeros: torch.Tensor,
+) -> torch.Tensor:
+    qweight_tmp = qweight
+    qzeros_tmp = zeros
+    qweight_list = []
+    qzeros_list = []
+    shifts = [0, 4, 1, 5, 2, 6, 3, 7]
+    for i in range(0, 8):
+        shift_num = shifts[i] * 4
+        qzeros_list.append((qzeros_tmp.reshape(-1, 1) >> shift_num) & 0xF)
+        qweight_list.append((qweight_tmp.reshape(-1, 1) >> shift_num) & 0xF)
+    qzeros_tmp = (
+        torch.cat(qzeros_list, dim=-1).reshape(qzeros_tmp.shape[0], -1).to(scales.dtype)
+    )
+    qweight_tmp = (
+        torch.cat(qweight_list, dim=-1)
+        .reshape(qweight_tmp.shape[0], -1)
+        .to(scales.dtype)
+    )
+    res = (
+        qweight_tmp.reshape(qzeros_tmp.shape[0], -1, qzeros_tmp.shape[1])
+        - qzeros_tmp.unsqueeze(1)
+    ) * scales.unsqueeze(1)
+    return res.reshape(qweight_tmp.shape[0], -1)

python/sglang/srt/layers/quantization/w8a8_int8.py

@@ -102,7 +102,12 @@ def npu_fused_experts(
     topk_weights: torch.Tensor,
     topk_ids: torch.Tensor,
     top_k: int,
+    **kwargs,
 ):
+    w13_offset = kwargs.get("w13_offset", None)
+    w2_offset = kwargs.get("w2_offset", None)
+    use_wna16 = kwargs.get("use_wna16", False)
+
     original_shape = hidden_states.shape
     original_dtype = hidden_states.dtype
     scale_dtype = original_dtype if original_dtype == torch.bfloat16 else torch.float32
@@ -127,12 +132,22 @@ def npu_fused_experts(
     )
     expert_tokens = expert_tokens.to(torch.int64)
     # gmm1: gate_up_proj
-    hidden_states, pertoken_scale = torch_npu.npu_dynamic_quant(hidden_states)
+    if not use_wna16:
+        hidden_states, pertoken_scale = torch_npu.npu_dynamic_quant(hidden_states)
+        scale_args13 = {
+            "scale": [w13_scale.to(scale_dtype)],
+            "per_token_scale": [pertoken_scale],
+        }
+    else:
+        scale_args13 = {
+            "antiquant_scale": [w13_scale],
+            "antiquant_offset": [w13_offset],
+        }
+
     hidden_states = torch_npu.npu_grouped_matmul(
         x=[hidden_states],
         weight=[w13],
-        scale=[w13_scale.to(scale_dtype)],
+        **scale_args13,
-        per_token_scale=[pertoken_scale],
         split_item=2,
         group_list_type=0,
         group_type=0,
@@ -141,13 +156,20 @@ def npu_fused_experts(
     )[0]
     # act_fn: swiglu
     hidden_states = torch_npu.npu_swiglu(hidden_states)
-    hidden_states, pertoken_scale = torch_npu.npu_dynamic_quant(hidden_states)
+    if not use_wna16:
+        hidden_states, pertoken_scale = torch_npu.npu_dynamic_quant(hidden_states)
+
+        scale_args2 = {
+            "scale": [w2_scale.to(scale_dtype)],
+            "per_token_scale": [pertoken_scale],
+        }
+    else:
+        scale_args2 = {"antiquant_scale": [w2_scale], "antiquant_offset": [w2_offset]}
     # gmm2: down_proj
     hidden_states = torch_npu.npu_grouped_matmul(
         x=[hidden_states],
         weight=[w2],
-        scale=[w2_scale.to(scale_dtype)],
+        **scale_args2,
-        per_token_scale=[pertoken_scale],
         split_item=2,
         group_list_type=0,
         group_type=0,

python/sglang/srt/model_loader/loader.py

@@ -612,6 +612,8 @@ class DefaultModelLoader(BaseModelLoader):
                 # parameters onto device for processing and back off after.
                 with device_loading_context(module, target_device):
                     quant_method.process_weights_after_loading(module)
+                if _is_npu:
+                    torch.npu.empty_cache()
 
 
 class LayeredModelLoader(DefaultModelLoader):

python/sglang/srt/models/deepseek_v2.py

@@ -189,6 +189,10 @@ elif _is_npu:
     import custom_ops  # noqa: F401
     import sgl_kernel_npu  # noqa: F401
     import torch_npu  # noqa: F401
+
+    from sglang.srt.layers.quantization.awq_triton import (
+        awq_dequantize_decomposition as awq_dequantize,
+    )
 else:
     pass
 
@@ -2965,7 +2969,7 @@ class DeepseekV2ForCausalLM(nn.Module):
             )
             if hasattr(self_attn.kv_b_proj, "qweight"):
                 # AWQ compatible
-                if _is_cuda or _is_hip:
+                if _is_cuda or _is_hip or _is_npu:
                     w = awq_dequantize(
                         self_attn.kv_b_proj.qweight,
                         self_attn.kv_b_proj.scales,

python/sglang/srt/utils/common.py

@@ -510,6 +510,8 @@ def get_available_gpu_memory(
                 f"WARNING: current device is not {gpu_id}, but {torch.npu.current_device()}, ",
                 "which may cause useless memory allocation for torch NPU context.",
             )
+        if empty_cache:
+            torch.npu.empty_cache()
         free_gpu_memory, total_gpu_memory = torch.npu.mem_get_info()
 
     if distributed: