Add minimal vLLM 0.16.1 build repo for BI-V150

vllm/lora/ops/xpu_ops/__init__.py

@@ -0,0 +1,6 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+
+from vllm.lora.ops.xpu_ops.lora_ops import bgmv_expand, bgmv_expand_slice, bgmv_shrink
+
+__all__ = ["bgmv_expand", "bgmv_expand_slice", "bgmv_shrink"]

vllm/lora/ops/xpu_ops/lora_ops.py

@@ -0,0 +1,54 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+
+import torch
+
+from vllm.logger import init_logger
+
+logger = init_logger(__name__)
+
+
+def bgmv_shrink(
+    inputs: torch.Tensor,
+    lora_a_weights: torch.Tensor,
+    output_tensor: torch.Tensor,
+    lora_indices_tensor: torch.Tensor,
+    scaling: float = 1.0,
+) -> None:
+    torch.ops._xpu_C.bgmv_shrink(
+        output_tensor, inputs, lora_a_weights, lora_indices_tensor, scaling
+    )
+
+
+def bgmv_expand(
+    inputs: torch.Tensor,
+    lora_b_weights: torch.Tensor,
+    output_tensor: torch.Tensor,
+    lora_indices_tensor: torch.Tensor,
+    add_inputs: bool = True,
+) -> None:
+    torch.ops._xpu_C.bgmv_expand(
+        output_tensor, inputs, lora_b_weights, lora_indices_tensor, add_inputs
+    )
+
+
+def bgmv_expand_slice(
+    inputs: torch.Tensor,
+    lora_b_weights: torch.Tensor,
+    output_tensor: torch.Tensor,
+    lora_indices_tensor: torch.Tensor,
+    slice_offset: int,
+    slice_size: int,
+    add_inputs: bool = True,
+) -> None:
+    assert slice_size == lora_b_weights.size(-2)
+    assert slice_offset + slice_size <= output_tensor.size(1)
+    torch.ops._xpu_C.bgmv_expand_slice(
+        output_tensor,
+        inputs,
+        lora_b_weights,
+        lora_indices_tensor,
+        slice_offset,
+        slice_size,
+        add_inputs,
+    )