[Kernel][Performance] Enable smaller Scaling Factor tiling for NVFP4 small-batch decoding (#30885)

Signed-off-by: LopezCastroRoberto <roberto.lopez.castro@udc.es>
Signed-off-by: Roberto L. Castro <38211239+LopezCastroRoberto@users.noreply.github.com>
Signed-off-by: LopezCastroRoberto <rocastro@redhat.com>

vllm/model_executor/layers/quantization/compressed_tensors/schemes/compressed_tensors_w4a4_nvfp4.py

@@ -23,7 +23,10 @@ from vllm.model_executor.parameter import (
     ModelWeightParameter,
     PerTensorScaleParameter,
 )
-from vllm.utils.flashinfer import flashinfer_scaled_fp4_mm, has_flashinfer
+from vllm.utils.flashinfer import (
+    flashinfer_scaled_fp4_mm,
+    has_flashinfer,
+)
 
 logger = init_logger(__name__)
 
@@ -187,7 +190,9 @@ class CompressedTensorsW4A4Fp4(CompressedTensorsScheme):
         output_shape = [*x.shape[:-1], layer.weight_packed.shape[0]]
 
         # quantize BF16 or FP16 to (FP4 and interleaved block scale)
-        x_fp4, x_blockscale = scaled_fp4_quant(x, layer.input_global_scale)
+        x_fp4, x_blockscale = scaled_fp4_quant(
+            x, layer.input_global_scale, self.backend
+        )
 
         mm_args = (
             x_fp4,

vllm/model_executor/layers/quantization/modelopt.py

@@ -1291,7 +1291,7 @@ class ModelOptNvFp4LinearMethod(LinearMethodBase):
         output_shape = [x.shape[0], layer.weight.shape[0]]
 
         # quantize BF16 or FP16 to (FP4 and interleaved block scale)
-        x_fp4, x_blockscale = scaled_fp4_quant(x, layer.input_scale_inv)
+        x_fp4, x_blockscale = scaled_fp4_quant(x, layer.input_scale_inv, self.backend)
 
         # validate dtypes of quantized input, input block scale,
         # weight and weight_blockscale