Commit 0eb4ec7a authored by Martin Perdacher's avatar Martin Perdacher

use intel compiler instead of GCC

parent 0d756600
Pipeline #582 failed with stage
in 3 minutes and 5 seconds
...@@ -13,9 +13,10 @@ set(SOURCE_FILES_JOIN mainJoin.cpp blasJoin/blasJoin.cpp ${UTIL_SOURCES}) ...@@ -13,9 +13,10 @@ set(SOURCE_FILES_JOIN mainJoin.cpp blasJoin/blasJoin.cpp ${UTIL_SOURCES})
set(CMAKE_CXX_FLAGS "-std=c++11") set(CMAKE_CXX_FLAGS "-std=c++11")
if ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "GNU") if ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "GNU")
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=knl -mtune=knl -fpic -ffast-math -DNDEBUG -O3 -DNDDEBUG -fopenmp -lboost_system -Wl,--no-as-needed -lmkl_intel_ilp64 -lmkl_intel_thread -lmkl_core -liomp5 -lpthread -lm -ldl") # some issues with GNU, use Intel Compiler instead
# set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=skylake-avx512 -fpic -ffast-math -DNDEBUG -O3 -DNDDEBUG -fopenmp -lboost_system -Wl,--no-as-needed -lmkl_intel_ilp64 -lmkl_intel_thread -lmkl_core -liomp5 -lpthread -lm -ldl")
elseif ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "Intel") elseif ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "Intel")
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -xmic-avx512 -qopenmp -DNDEBUG -O3 -lmkl_intel_thread -liomp5 -lpthread -lmkl_core -lmkl_intel_lp64 -lboost_system") set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -xCore-AVX512 -qopenmp -DNDEBUG -O3 -lmkl_intel_thread -liomp5 -lpthread -lmkl_core -lmkl_intel_lp64 -lboost_system")
# set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -qopt-report=2") # set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -qopt-report=2")
add_definitions(-DDEBUG) add_definitions(-DDEBUG)
endif() endif()
......
...@@ -47,6 +47,7 @@ void blasSelfJoinCountOnly(const double *x, const size_t N, const size_t D, cons ...@@ -47,6 +47,7 @@ void blasSelfJoinCountOnly(const double *x, const size_t N, const size_t D, cons
// perform regular matrix multiplication // perform regular matrix multiplication
// C := alpha*A*B' + beta*C // C := alpha*A*B' + beta*C
// printf("blockRow: %zu, blockCol: %zu, D:%zu\n", blockRow, blockCol, D);
cblas_dgemm(CblasRowMajor, CblasNoTrans, CblasTrans, blockRow, blockCol, D, 1.0, &x[i*BLOCKSIZE*D], D, &x[j*BLOCKSIZE*D], D, 0.0, iresult, blockCol); cblas_dgemm(CblasRowMajor, CblasNoTrans, CblasTrans, blockRow, blockCol, D, 1.0, &x[i*BLOCKSIZE*D], D, &x[j*BLOCKSIZE*D], D, 0.0, iresult, blockCol);
/* /*
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment