xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-07-11 17:27:18 +08:00
Code Issues Packages Projects Releases Wiki Activity

cleanup

Browse Source 
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
This commit is contained in:
Lucas Wilkinson 2025-05-15 04:48:08 +00:00
parent b478b18f07
commit 4e00778a60
2 changed files with 16 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
CMakeLists.txt
View File

@ -321,8 +321,6 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
"csrc/quantization/gptq_marlin/gptq_marlin.cu" "csrc/quantization/gptq_marlin/gptq_marlin.cu"
"csrc/quantization/gptq_marlin/gptq_marlin_repack.cu" "csrc/quantization/gptq_marlin/gptq_marlin_repack.cu"
"csrc/quantization/gptq_marlin/awq_marlin_repack.cu" "csrc/quantization/gptq_marlin/awq_marlin_repack.cu"
NO_ARCH_MSG
"Not building Marlin kernels as no compatible archs found in CUDA target architectures"
) )
# AllSpark kernels # AllSpark kernels
@ -332,7 +330,6 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
SRCS SRCS
"csrc/quantization/gptq_allspark/allspark_repack.cu" "csrc/quantization/gptq_allspark/allspark_repack.cu"
"csrc/quantization/gptq_allspark/allspark_qgemm_w8a16.cu" "csrc/quantization/gptq_allspark/allspark_qgemm_w8a16.cu"
NO_ARCH_MSG "Not building AllSpark kernels as no compatible archs found in CUDA target architectures"
) )
@ -411,7 +408,6 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
"csrc/quantization/fp4/nvfp4_scaled_mm_kernels.cu" "csrc/quantization/fp4/nvfp4_scaled_mm_kernels.cu"
"csrc/quantization/fp4/nvfp4_blockwise_moe_kernel.cu" "csrc/quantization/fp4/nvfp4_blockwise_moe_kernel.cu"
FLAGS "-DENABLE_NVFP4=1" FLAGS "-DENABLE_NVFP4=1"
NO_ARCH_MSG "Not building NVFP4 as no compatible archs were found."
) )
# CUTLASS MLA Archs and flags # CUTLASS MLA Archs and flags
@ -421,7 +417,6 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
ARCHS "10.0a" ARCHS "10.0a"
SRCS "csrc/attention/mla/cutlass_mla_kernels.cu" SRCS "csrc/attention/mla/cutlass_mla_kernels.cu"
FLAGS "-DENABLE_CUTLASS_MLA=1" FLAGS "-DENABLE_CUTLASS_MLA=1"
NO_ARCH_MSG "Not building CUTLASS MLA as no compatible archs were found."
) )
# Add MLA-specific include directories only to MLA source files # Add MLA-specific include directories only to MLA source files
set_source_files_properties( set_source_files_properties(
@ -456,8 +451,6 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
VERSION_MSG VERSION_MSG
"Not building Machete kernels as CUDA Compiler version is less than 12.0." "Not building Machete kernels as CUDA Compiler version is less than 12.0."
"We recommend upgrading to CUDA 12.0 or later to run w4a16 quantized models on Hopper." "We recommend upgrading to CUDA 12.0 or later to run w4a16 quantized models on Hopper."
NO_ARCH_MSG
"Not building Machete kernels as no compatible archs found in CUDA target architectures"
) )
# if CUDA endif # if CUDA endif
endif() endif()
@ -503,7 +496,6 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
GEN_SCRIPT "${CMAKE_CURRENT_SOURCE_DIR}/csrc/moe/marlin_moe_wna16/generate_kernels.py" GEN_SCRIPT "${CMAKE_CURRENT_SOURCE_DIR}/csrc/moe/marlin_moe_wna16/generate_kernels.py"
GEN_GLOB "csrc/moe/marlin_moe_wna16/*.cu" GEN_GLOB "csrc/moe/marlin_moe_wna16/*.cu"
SRCS "csrc/moe/moe_wna16.cu" SRCS "csrc/moe/moe_wna16.cu"
NO_ARCH_MSG "Not building Marlin MOE kernels as no compatible archs found in CUDA target architectures"
OUT_SRCS_VAR VLLM_MOE_EXT_SRC OUT_SRCS_VAR VLLM_MOE_EXT_SRC
) )
endif() endif()

32
cmake/utils.cmake
View File

@ -114,22 +114,22 @@ function (hipify_sources_target OUT_SRCS NAME ORIG_SRCS)
set(${OUT_SRCS} ${HIP_SRCS} PARENT_SCOPE) set(${OUT_SRCS} ${HIP_SRCS} PARENT_SCOPE)
endfunction() endfunction()
## Macro to conditionally include CUDA sources based on architecture and CUDA # Macro to conditionally include CUDA sources based on architecture and CUDA
## compiler version, optionally generating sources via a Python script. # compiler version, optionally generating sources via a Python script.
## Usage: # Usage:
## optional_cuda_sources( # optional_cuda_sources(
## NAME <name> # NAME <name>
## [MIN_VERSION <version>] # [MIN_VERSION <version>]
## ARCHS <arch1;arch2;...> # ARCHS <arch1;arch2;...>
## SRCS <static_source1> [<static_source2> ...] # SRCS <static_source1> [<static_source2> ...]
## [FLAGS <flag1> ...] # [FLAGS <flag1> ...]
## [VERSION_MSG <line1> [<line2> ...]] # [VERSION_MSG <line1> [<line2> ...]]
## [NO_ARCH_MSG <line1> [<line2> ...]] # [NO_ARCH_MSG <line1> [<line2> ...]]
## [GEN_SCRIPT <path/to/generate_script.py>] # [GEN_SCRIPT <path/to/generate_script.py>]
## [GEN_GLOB <glob_pattern_for_generated_sources>] # [GEN_GLOB <glob_pattern_for_generated_sources>]
## [APPEND_ARCHS <var_to_append_archs>] # [APPEND_ARCHS <var_to_append_archs>]
## This will run GEN_SCRIPT once when version and arch checks pass, globbing # This will run GEN_SCRIPT once when version and arch checks pass, globbing
## sources matching GEN_GLOB and appending them alongside SRCS. # sources matching GEN_GLOB and appending them alongside SRCS.
macro(optional_cuda_sources) macro(optional_cuda_sources)
set(oneValueArgs NAME MIN_VERSION APPEND_ARCHS GEN_SCRIPT GEN_GLOB OUT_SRCS_VAR) set(oneValueArgs NAME MIN_VERSION APPEND_ARCHS GEN_SCRIPT GEN_GLOB OUT_SRCS_VAR)
set(multiValueArgs ARCHS SRCS FLAGS VERSION_MSG NO_ARCH_MSG) set(multiValueArgs ARCHS SRCS FLAGS VERSION_MSG NO_ARCH_MSG)