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 16, 10:45 UTC
[Build/CI] Set FETCHCONTENT_BASE_DIR to one location for better caching (#8930)
Tyler Michael Smith committed
Sep 28, 2024 at 23:13 UTC
2e7fe7e79f41e294eeed2f484eeb791284ec48a2
2 files changed
+10
.gitignore
+1
@@ -33,6 +33,7 @@ share/python-wheels/
33
.installed.cfg
34
*.egg
35
MANIFEST
36
+/.deps/
37
38
# PyInstaller
39
# Usually these files are written by a python script from a template
CMakeLists.txt
+9
@@ -166,7 +166,16 @@ if(NVCC_THREADS AND VLLM_GPU_LANG STREQUAL "CUDA")
166
list(APPEND VLLM_GPU_FLAGS "--threads=${NVCC_THREADS}")
167
endif()
168
169
+
170
+#
171
+# Use FetchContent for C++ dependencies that are compiled as part of vLLM's build process.
172
+# Configure it to place files in vllm/.deps, in order to play nicely with sccache.
173
+#
174
include(FetchContent)
175
+get_filename_component(PROJECT_ROOT_DIR "${CMAKE_CURRENT_SOURCE_DIR}" ABSOLUTE)
176
+file(MAKE_DIRECTORY "${FETCHCONTENT_BASE_DIR}")
177
+set(FETCHCONTENT_BASE_DIR "${PROJECT_ROOT_DIR}/.deps")
178
+message(STATUS "FetchContent base directory: ${FETCHCONTENT_BASE_DIR}")
179
180
#
181
# Define other extension targets