[ROCm] Temporarily remove GPTQ ROCm support (#2138)

setup.py

@@ -219,13 +219,13 @@ vllm_extension_sources = [
     "csrc/activation_kernels.cu",
     "csrc/layernorm_kernels.cu",
     "csrc/quantization/squeezellm/quant_cuda_kernel.cu",
-    "csrc/quantization/gptq/q_gemm.cu",
     "csrc/cuda_utils_kernels.cu",
     "csrc/pybind.cpp",
 ]
 
 if _is_cuda():
     vllm_extension_sources.append("csrc/quantization/awq/gemm_kernels.cu")
+    vllm_extension_sources.append("csrc/quantization/gptq/q_gemm.cu")
 
 vllm_extension = CUDAExtension(
     name="vllm._C",

vllm/config.py

@@ -143,7 +143,7 @@ class ModelConfig:
 
     def _verify_quantization(self) -> None:
         supported_quantization = ["awq", "gptq", "squeezellm"]
-        rocm_not_supported_quantization = ["awq"]
+        rocm_not_supported_quantization = ["awq", "gptq"]
         if self.quantization is not None:
             self.quantization = self.quantization.lower()