[Bugfix/Core] Flashinfer k_scale and v_scale (#9861)

tests/kernels/test_cache.py

@@ -258,19 +258,20 @@ def test_reshape_and_cache_flash(
     del key_caches
     del value_caches
 
+    k_scale = key.amax().item() / 256
+    v_scale = value.amax().item() / 256
+
     # Clone the KV caches.
     if kv_cache_dtype == "fp8":
         cloned_key_cache = torch.empty_like(key_cache, dtype=torch.float16)
-        ops.convert_fp8(cloned_key_cache, key_cache)
+        ops.convert_fp8(cloned_key_cache, key_cache, k_scale, kv_cache_dtype)
         cloned_value_cache = torch.empty_like(value_cache, dtype=torch.float16)
-        ops.convert_fp8(cloned_value_cache, value_cache)
+        ops.convert_fp8(cloned_value_cache, value_cache, v_scale,
+                        kv_cache_dtype)
     else:
         cloned_key_cache = key_cache.clone()
         cloned_value_cache = value_cache.clone()
 
-    # Using default kv_scale
-    k_scale = v_scale = 1.0
-
     # Call the reshape_and_cache kernel.
     opcheck(torch.ops._C_cache_ops.reshape_and_cache_flash,
             (key, value, key_cache, value_cache, slot_mapping, kv_cache_dtype,
@@ -281,9 +282,15 @@ def test_reshape_and_cache_flash(
 
     if kv_cache_dtype == "fp8":
         result_key_cache = torch.empty_like(key_cache, dtype=torch.float16)
-        ops.convert_fp8(result_key_cache, key_cache)
+        ops.convert_fp8(result_key_cache,
+                        key_cache,
+                        k_scale,
+                        kv_dtype=kv_cache_dtype)
         result_value_cache = torch.empty_like(value_cache, dtype=torch.float16)
-        ops.convert_fp8(result_value_cache, value_cache)
+        ops.convert_fp8(result_value_cache,
+                        value_cache,
+                        v_scale,
+                        kv_dtype=kv_cache_dtype)
 
     # Run the reference implementation.
     block_indicies = torch.div(slot_mapping, block_size, rounding_mode="floor")

vllm/attention/backends/flashinfer.py

@@ -759,8 +759,6 @@ class FlashInferImpl(AttentionImpl):
         v_scale: float = 1.0,
         attn_type: AttentionType = AttentionType.DECODER,
     ) -> torch.Tensor:
-        assert k_scale == 1.0 and v_scale == 1.0, (
-            "key/v_scale is not supported in FlashInfer.")
         if attn_type != AttentionType.DECODER:
             raise NotImplementedError("Encoder self-attention and "
                                       "encoder/decoder cross-attention "
@@ -874,7 +872,12 @@ def unified_flash_infer(
             assert prefill_meta is not None
             assert prefill_meta.prefill_wrapper is not None
             prefill_output = prefill_meta.prefill_wrapper.forward(
-                query, kv_cache, logits_soft_cap=logits_soft_cap, causal=True)
+                query,
+                kv_cache,
+                logits_soft_cap=logits_soft_cap,
+                causal=True,
+                k_scale=k_scale,
+                v_scale=v_scale)
     if decode_meta := attn_metadata.decode_metadata:
         assert attn_metadata.decode_metadata is not None
         assert attn_metadata.decode_metadata.decode_wrapper is not None

vllm/model_executor/layers/quantization/modelopt.py

@@ -141,8 +141,11 @@ class ModelOptFp8LinearMethod(LinearMethodBase):
             layer.register_parameter("input_scale", scale)
 
     def process_weights_after_loading(self, layer: Module) -> None:
-        max_w_scale, weight = requantize_with_max_scale(
+        weight = layer.weight
-            layer.weight, layer.weight_scale, layer.logical_widths)
+        max_w_scale = layer.weight_scale.max()
+        if not (layer.weight_scale == layer.weight_scale[0]).all():
+            max_w_scale, weight = requantize_with_max_scale(
+                layer.weight, layer.weight_scale, layer.logical_widths)
         layer.weight = Parameter(weight.t(), requires_grad=False)
         layer.weight_scale = Parameter(max_w_scale, requires_grad=False)
         layer.input_scale = Parameter(layer.input_scale.max(),