Make verifiable a DSO and add NAME_SUFFIX support
Build option DSO=1 generates libverifiable.so which can be used to reduce the combined binary size. Build option NAME_SUFFIX can be used to a add suffix to all generated binaries. e.g. NAME_SUFFIX=_mpi Added new make target: clean_intermediates
Этот коммит содержится в:
David Addison
+25
-68
@@ -1,73 +1,13 @@
|
||||
#
|
||||
# Copyright (c) 2015-2022, NVIDIA CORPORATION. All rights reserved.
|
||||
# Copyright (c) 2015-2025, NVIDIA CORPORATION. All rights reserved.
|
||||
#
|
||||
# See LICENSE.txt for license information
|
||||
#
|
||||
include common.mk
|
||||
|
||||
CUDA_HOME ?= /usr/local/cuda
|
||||
PREFIX ?= /usr/local
|
||||
VERBOSE ?= 0
|
||||
DEBUG ?= 0
|
||||
|
||||
CUDA_LIB ?= $(CUDA_HOME)/lib64
|
||||
CUDA_INC ?= $(CUDA_HOME)/include
|
||||
NVCC ?= $(CUDA_HOME)/bin/nvcc
|
||||
CUDARTLIB ?= cudart
|
||||
|
||||
CUDA_VERSION = $(strip $(shell which $(NVCC) >/dev/null && $(NVCC) --version | grep release | sed 's/.*release //' | sed 's/\,.*//'))
|
||||
CUDA_MAJOR = $(shell echo $(CUDA_VERSION) | cut -d "." -f 1)
|
||||
CUDA_MINOR = $(shell echo $(CUDA_VERSION) | cut -d "." -f 2)
|
||||
|
||||
# Better define NVCC_GENCODE in your environment to the minimal set
|
||||
# of archs to reduce compile time.
|
||||
ifeq ($(shell test "0$(CUDA_MAJOR)" -eq 12 -a "0$(CUDA_MINOR)" -ge 8 -o "0$(CUDA_MAJOR)" -ge 13; echo $$?),0)
|
||||
# Include Blackwell support if we're using CUDA12.8 or above
|
||||
NVCC_GENCODE ?= -gencode=arch=compute_80,code=sm_80 \
|
||||
-gencode=arch=compute_90,code=sm_90 \
|
||||
-gencode=arch=compute_100,code=sm_100 \
|
||||
-gencode=arch=compute_120,code=sm_120 \
|
||||
-gencode=arch=compute_120,code=compute_120
|
||||
else ifeq ($(shell test "0$(CUDA_MAJOR)" -ge 12; echo $$?),0)
|
||||
NVCC_GENCODE ?= -gencode=arch=compute_60,code=sm_60 \
|
||||
-gencode=arch=compute_61,code=sm_61 \
|
||||
-gencode=arch=compute_70,code=sm_70 \
|
||||
-gencode=arch=compute_80,code=sm_80 \
|
||||
-gencode=arch=compute_90,code=sm_90 \
|
||||
-gencode=arch=compute_90,code=compute_90
|
||||
else ifeq ($(shell test "0$(CUDA_MAJOR)" -ge 11; echo $$?),0)
|
||||
NVCC_GENCODE ?= -gencode=arch=compute_60,code=sm_60 \
|
||||
-gencode=arch=compute_61,code=sm_61 \
|
||||
-gencode=arch=compute_70,code=sm_70 \
|
||||
-gencode=arch=compute_80,code=sm_80 \
|
||||
-gencode=arch=compute_80,code=compute_80
|
||||
else
|
||||
NVCC_GENCODE ?= -gencode=arch=compute_35,code=sm_35 \
|
||||
-gencode=arch=compute_50,code=sm_50 \
|
||||
-gencode=arch=compute_60,code=sm_60 \
|
||||
-gencode=arch=compute_61,code=sm_61 \
|
||||
-gencode=arch=compute_70,code=sm_70 \
|
||||
-gencode=arch=compute_70,code=compute_70
|
||||
endif
|
||||
|
||||
NVCUFLAGS := -ccbin $(CXX) $(NVCC_GENCODE) -std=c++11
|
||||
CXXFLAGS := -std=c++11
|
||||
|
||||
LDFLAGS := -L${CUDA_LIB} -lcudart -lrt
|
||||
NVLDFLAGS := -L${CUDA_LIB} -l${CUDARTLIB} -lrt
|
||||
|
||||
ifeq ($(DEBUG), 0)
|
||||
NVCUFLAGS += -O3 -g
|
||||
CXXFLAGS += -O3 -g
|
||||
else
|
||||
NVCUFLAGS += -O0 -G -g
|
||||
CXXFLAGS += -O0 -g -ggdb3
|
||||
endif
|
||||
|
||||
ifneq ($(VERBOSE), 0)
|
||||
NVCUFLAGS += -Xcompiler -Wall,-Wextra,-Wno-unused-parameter
|
||||
else
|
||||
.SILENT:
|
||||
endif
|
||||
MPI ?= 0 # Set to 1 to enable MPI support (multi-process/multi-node)
|
||||
NAME_SUFFIX ?= # e.g. _mpi when using MPI=1
|
||||
DSO ?= 0 # Set to 1 to create and use libverifiable.so to reduce binary size
|
||||
|
||||
.PHONY: build clean
|
||||
|
||||
@@ -92,7 +32,7 @@ DST_DIR := $(BUILDDIR)
|
||||
SRC_FILES := $(wildcard *.cu)
|
||||
OBJ_FILES := $(SRC_FILES:%.cu=${DST_DIR}/%.o)
|
||||
BIN_FILES_LIST := all_reduce all_gather broadcast reduce_scatter reduce alltoall scatter gather sendrecv hypercube
|
||||
BIN_FILES := $(BIN_FILES_LIST:%=${DST_DIR}/%_perf)
|
||||
BIN_FILES := $(BIN_FILES_LIST:%=${DST_DIR}/%_perf${NAME_SUFFIX})
|
||||
|
||||
build: ${BIN_FILES}
|
||||
|
||||
@@ -103,18 +43,35 @@ TEST_VERIFIABLE_SRCDIR := ../verifiable
|
||||
TEST_VERIFIABLE_BUILDDIR := $(BUILDDIR)/verifiable
|
||||
include ../verifiable/verifiable.mk
|
||||
|
||||
.PRECIOUS: ${DST_DIR}/%.o
|
||||
|
||||
${DST_DIR}/%.o: %.cu common.h $(TEST_VERIFIABLE_HDRS)
|
||||
@printf "Compiling %-35s > %s\n" $< $@
|
||||
@mkdir -p ${DST_DIR}
|
||||
$(NVCC) -o $@ $(NVCUFLAGS) -c $<
|
||||
|
||||
${DST_DIR}/%$(NAME_SUFFIX).o: %.cu common.h $(TEST_VERIFIABLE_HDRS)
|
||||
@printf "Compiling %-35s > %s\n" $< $@
|
||||
@mkdir -p ${DST_DIR}
|
||||
$(NVCC) -o $@ $(NVCUFLAGS) -c $<
|
||||
|
||||
${DST_DIR}/timer.o: timer.cc timer.h
|
||||
@printf "Compiling %-35s > %s\n" $< $@
|
||||
@mkdir -p ${DST_DIR}
|
||||
$(CXX) $(CXXFLAGS) -o $@ -c timer.cc
|
||||
$(CXX) $(CXXFLAGS) -o $@ -c $<
|
||||
|
||||
${DST_DIR}/%_perf:${DST_DIR}/%.o ${DST_DIR}/common.o ${DST_DIR}/timer.o $(TEST_VERIFIABLE_OBJS)
|
||||
ifeq ($(DSO), 1)
|
||||
${DST_DIR}/%_perf$(NAME_SUFFIX): ${DST_DIR}/%.o ${DST_DIR}/common$(NAME_SUFFIX).o ${DST_DIR}/timer.o $(TEST_VERIFIABLE_LIBS)
|
||||
@printf "Linking %-35s > %s\n" $< $@
|
||||
@mkdir -p ${DST_DIR}
|
||||
$(NVCC) -o $@ $(NVCUFLAGS) $^ -L$(TEST_VERIFIABLE_BUILDDIR) -lverifiable ${NVLDFLAGS} -Xlinker "--enable-new-dtags" -Xlinker "-rpath,\$$ORIGIN:\$$ORIGIN/verifiable"
|
||||
else
|
||||
${DST_DIR}/%_perf$(NAME_SUFFIX):${DST_DIR}/%.o ${DST_DIR}/common$(NAME_SUFFIX).o ${DST_DIR}/timer.o $(TEST_VERIFIABLE_OBJS)
|
||||
@printf "Linking %-35s > %s\n" $< $@
|
||||
@mkdir -p ${DST_DIR}
|
||||
$(NVCC) -o $@ $(NVCUFLAGS) $^ ${NVLDFLAGS}
|
||||
endif
|
||||
|
||||
clean_intermediates:
|
||||
rm -f ${DST_DIR}/*.o $(TEST_VERIFIABLE_OBJS)
|
||||
|
||||
|
||||
@@ -0,0 +1,69 @@
|
||||
#
|
||||
# Copyright (c) 2015-2025, NVIDIA CORPORATION. All rights reserved.
|
||||
#
|
||||
# See LICENSE.txt for license information
|
||||
#
|
||||
CUDA_HOME ?= /usr/local/cuda
|
||||
PREFIX ?= /usr/local
|
||||
VERBOSE ?= 0
|
||||
DEBUG ?= 0
|
||||
|
||||
CUDA_LIB ?= $(CUDA_HOME)/lib64
|
||||
CUDA_INC ?= $(CUDA_HOME)/include
|
||||
NVCC ?= $(CUDA_HOME)/bin/nvcc
|
||||
CUDARTLIB ?= cudart
|
||||
|
||||
CUDA_VERSION = $(strip $(shell which $(NVCC) >/dev/null && $(NVCC) --version | grep release | sed 's/.*release //' | sed 's/\,.*//'))
|
||||
CUDA_MAJOR = $(shell echo $(CUDA_VERSION) | cut -d "." -f 1)
|
||||
CUDA_MINOR = $(shell echo $(CUDA_VERSION) | cut -d "." -f 2)
|
||||
|
||||
# Better define NVCC_GENCODE in your environment to the minimal set
|
||||
# of archs to reduce compile time.
|
||||
ifeq ($(shell test "0$(CUDA_MAJOR)" -eq 12 -a "0$(CUDA_MINOR)" -ge 8 -o "0$(CUDA_MAJOR)" -ge 13; echo $$?),0)
|
||||
# Include Blackwell support if we're using CUDA12.8 or above
|
||||
NVCC_GENCODE ?= -gencode=arch=compute_80,code=sm_80 \
|
||||
-gencode=arch=compute_90,code=sm_90 \
|
||||
-gencode=arch=compute_100,code=sm_100 \
|
||||
-gencode=arch=compute_120,code=sm_120 \
|
||||
-gencode=arch=compute_120,code=compute_120
|
||||
else ifeq ($(shell test "0$(CUDA_MAJOR)" -ge 12; echo $$?),0)
|
||||
NVCC_GENCODE ?= -gencode=arch=compute_60,code=sm_60 \
|
||||
-gencode=arch=compute_61,code=sm_61 \
|
||||
-gencode=arch=compute_70,code=sm_70 \
|
||||
-gencode=arch=compute_80,code=sm_80 \
|
||||
-gencode=arch=compute_90,code=sm_90 \
|
||||
-gencode=arch=compute_90,code=compute_90
|
||||
else ifeq ($(shell test "0$(CUDA_MAJOR)" -ge 11; echo $$?),0)
|
||||
NVCC_GENCODE ?= -gencode=arch=compute_60,code=sm_60 \
|
||||
-gencode=arch=compute_61,code=sm_61 \
|
||||
-gencode=arch=compute_70,code=sm_70 \
|
||||
-gencode=arch=compute_80,code=sm_80 \
|
||||
-gencode=arch=compute_80,code=compute_80
|
||||
else
|
||||
NVCC_GENCODE ?= -gencode=arch=compute_35,code=sm_35 \
|
||||
-gencode=arch=compute_50,code=sm_50 \
|
||||
-gencode=arch=compute_60,code=sm_60 \
|
||||
-gencode=arch=compute_61,code=sm_61 \
|
||||
-gencode=arch=compute_70,code=sm_70 \
|
||||
-gencode=arch=compute_70,code=compute_70
|
||||
endif
|
||||
|
||||
NVCUFLAGS := -ccbin $(CXX) $(NVCC_GENCODE) -std=c++11
|
||||
CXXFLAGS := -std=c++11
|
||||
|
||||
LDFLAGS := -L${CUDA_LIB} -lcudart -lrt
|
||||
NVLDFLAGS := -L${CUDA_LIB} -l${CUDARTLIB} -lrt
|
||||
|
||||
ifeq ($(DEBUG), 0)
|
||||
NVCUFLAGS += -O3 -g
|
||||
CXXFLAGS += -O3 -g
|
||||
else
|
||||
NVCUFLAGS += -O0 -G -g
|
||||
CXXFLAGS += -O0 -g -ggdb3
|
||||
endif
|
||||
|
||||
ifneq ($(VERBOSE), 0)
|
||||
NVCUFLAGS += -Xcompiler -Wall,-Wextra,-Wno-unused-parameter
|
||||
else
|
||||
.SILENT:
|
||||
endif
|
||||
Ссылка в новой задаче
Block a user