A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
Force use C++17 globally to avoid compilation error (#24823)
Signed-off-by: chenfengjin <1871653365@qq.com>
FengjinChen committed
Sep 15, 2025 at 03:30 UTC
79cbcab8717d737945fc458f21fb1f98711ed50e
2 files changed
+4
-1
CMakeLists.txt
+4
@@ -13,6 +13,10 @@ cmake_minimum_required(VERSION 3.26)
13
# cmake --install . --component _C
14
project(vllm_extensions LANGUAGES CXX)
15
16
+set(CMAKE_CXX_STANDARD 17)
17
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
18
+
19
+
20
# CUDA by default, can be overridden by using -DVLLM_TARGET_DEVICE=... (used by setup.py)
21
set(VLLM_TARGET_DEVICE "cuda" CACHE STRING "Target device backend for vLLM")
22
message(STATUS "Build type: ${CMAKE_BUILD_TYPE}")
cmake/utils.cmake
-1
@@ -480,7 +480,6 @@ function (define_gpu_extension_target GPU_MOD_NAME)
480
${GPU_LANGUAGE}_ARCHITECTURES "${GPU_ARCHITECTURES}")
481
endif()
482
483
- set_property(TARGET ${GPU_MOD_NAME} PROPERTY CXX_STANDARD 17)
483
484
target_compile_options(${GPU_MOD_NAME} PRIVATE
485
$<$<COMPILE_LANGUAGE:${GPU_LANGUAGE}>:${GPU_COMPILE_FLAGS}>)