From 4a61950f4d154d6518898a46ca7f1c0e2caaacdf Mon Sep 17 00:00:00 2001
From: ihb2032 <40718643+ihb2032@users.noreply.github.com>
Date: Mon, 13 Oct 2025 22:56:01 +0800
Subject: [PATCH] [Hardware][CPU] Disable torch.compile for RISC-V to prevent
 APIError (#26693)

Signed-off-by: lyd1992 <liuyudong@iscas.ac.cn>
Signed-off-by: ihb2032 <1355790728@qq.com>
Signed-off-by: lyd1992 <liuyudong@iscas.ac.cn
---
 vllm/v1/sample/ops/topk_topp_sampler.py | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/vllm/v1/sample/ops/topk_topp_sampler.py b/vllm/v1/sample/ops/topk_topp_sampler.py
index f3322dc8a4ce..ed8bc55a3cf2 100644
--- a/vllm/v1/sample/ops/topk_topp_sampler.py
+++ b/vllm/v1/sample/ops/topk_topp_sampler.py
@@ -9,7 +9,7 @@ from packaging import version
 from vllm import envs
 from vllm.config.model import LogprobsMode
 from vllm.logger import init_logger
-from vllm.platforms import current_platform
+from vllm.platforms import CpuArchEnum, current_platform
 
 logger = init_logger(__name__)
 
@@ -73,7 +73,10 @@ class TopKTopPSampler(nn.Module):
                 )
                 self.forward = self.forward_native
         elif current_platform.is_cpu():
-            self.forward = self.forward_cpu
+            if current_platform.get_cpu_architecture() == CpuArchEnum.RISCV:
+                self.forward = self.forward_native
+            else:
+                self.forward = self.forward_cpu
         else:
             self.forward = self.forward_native