[Kernel] Lazy import FlashInfer (#26977)

tests/v1/sample/test_topk_topp_sampler.py

@@ -5,20 +5,13 @@ import torch
 from torch import Generator
 
 from vllm.platforms import current_platform
-from vllm.v1.sample.ops.topk_topp_sampler import (
-    apply_top_k_top_p,
-    is_flashinfer_available,
-)
+from vllm.v1.sample.ops.topk_topp_sampler import apply_top_k_top_p
 
 DEVICE = current_platform.device_type
 
 BATCH_SIZE = 1024
 VOCAB_SIZE = 128 * 1024
 
-FLASHINFER_ENABLED = current_platform.is_cuda() and is_flashinfer_available
-if is_flashinfer_available:
-    from flashinfer.sampling import top_k_renorm_probs, top_p_renorm_probs
-
 
 @pytest.fixture(autouse=True)
 def reset_default_device():
@@ -65,6 +58,14 @@ def test_flashinfer_sampler():
     sampling results due to randomness), so we will compare the probability
     renormed consequently by top-k and then top-p of FlashInfer implementation.
     """
+    try:
+        from flashinfer.sampling import top_k_renorm_probs, top_p_renorm_probs
+
+        is_flashinfer_available = True
+    except ImportError:
+        is_flashinfer_available = False
+
+    FLASHINFER_ENABLED = current_platform.is_cuda() and is_flashinfer_available
 
     if not FLASHINFER_ENABLED:
         pytest.skip("FlashInfer not installed or not available on this platform.")

vllm/v1/sample/ops/topk_topp_sampler.py

@@ -13,13 +13,6 @@ from vllm.platforms import CpuArchEnum, current_platform
 
 logger = init_logger(__name__)
 
-try:
-    import flashinfer.sampling
-
-    is_flashinfer_available = True
-except ImportError:
-    is_flashinfer_available = False
-
 
 class TopKTopPSampler(nn.Module):
     """
@@ -38,15 +31,7 @@ class TopKTopPSampler(nn.Module):
             logprobs_mode not in ("processed_logits", "processed_logprobs")
             and current_platform.is_cuda()
         ):
-            if is_flashinfer_available:
-                flashinfer_version = flashinfer.__version__
-                if version.parse(flashinfer_version) < version.parse("0.2.3"):
-                    logger.warning_once(
-                        "FlashInfer version >= 0.2.3 required. "
-                        "Falling back to default sampling implementation."
-                    )
-                    self.forward = self.forward_native
-                elif envs.VLLM_USE_FLASHINFER_SAMPLER:
+            if envs.VLLM_USE_FLASHINFER_SAMPLER:
                 # Users must opt in explicitly via VLLM_USE_FLASHINFER_SAMPLER=1.
                 logger.info_once("Using FlashInfer for top-p & top-k sampling.")
                 self.forward = self.forward_cuda
@@ -57,13 +42,7 @@ class TopKTopPSampler(nn.Module):
                     "after verifying accuracy for your workloads."
                 )
                 self.forward = self.forward_native
-            else:
-                logger.warning_once(
-                    "FlashInfer is not available. Falling back to the PyTorch-"
-                    "native implementation of top-p & top-k sampling. For the "
-                    "best performance, please install FlashInfer."
-                )
-                self.forward = self.forward_native
+
         elif current_platform.is_cpu():
             arch = current_platform.get_cpu_architecture()
             # Fall back to native implementation for POWERPC and RISCV.
@@ -278,6 +257,13 @@ def flashinfer_sample(
     does not. Call this function at the end of the forward pass to minimize
     the synchronization overhead.
     """
+    import flashinfer
+
+    if version.parse(flashinfer.__version__) < version.parse("0.2.3"):
+        raise ImportError(
+            "FlashInfer version >= 0.2.3 required for top-k and top-p sampling. "
+        )
+
     assert not (k is None and p is None)
     if k is None:
         # Top-p only.