portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-14 13:57:07 +01:00
Code Releases Activity

definetely the right merge upstream/develop

Browse Source
This commit is contained in:
Alessandro Lupo
2023-06-16 14:19:46 +01:00
parent 4895ff260e ffd7301649
commit e09dfbf1c2
446 changed files with 46860 additions and 16333 deletions
Download Patch File Download Diff File Expand all files Collapse all files

147
configure.ac
View File

@ -1,5 +1,5 @@
AC_PREREQ([2.63])
AC_INIT([Grid], [0.7.0], [https://github.com/paboyle/Grid], [Grid])
AC_PREREQ([2.69])
AC_INIT([Grid],[0.7.0],[https://github.com/paboyle/Grid],[Grid])
AC_CANONICAL_BUILD
AC_CANONICAL_HOST
AC_CANONICAL_TARGET
@ -20,7 +20,7 @@ m4_ifdef([AM_SILENT_RULES], [AM_SILENT_RULES([yes])])
################ Set flags
# do not move!
AC_ARG_ENABLE([debug],[AC_HELP_STRING([--enable-debug=yes|no], [enable debug compilation ])],
AC_ARG_ENABLE([debug],[AS_HELP_STRING([--enable-debug=yes|no],[enable debug compilation ])],
[ac_DEBUG=${enable_debug}], [ac_DEBUG=no])
case ${ac_DEBUG} in
yes)
@ -114,7 +114,7 @@ AC_ARG_WITH([openssl],
############### lapack
AC_ARG_ENABLE([lapack],
[AC_HELP_STRING([--enable-lapack=yes|no|prefix], [enable LAPACK])],
[AS_HELP_STRING([--enable-lapack=yes|no|prefix],[enable LAPACK])],
[ac_LAPACK=${enable_lapack}], [ac_LAPACK=no])
case ${ac_LAPACK} in
@ -128,38 +128,42 @@ case ${ac_LAPACK} in
AC_DEFINE([USE_LAPACK],[1],[use LAPACK]);;
esac
############### tracing
AC_ARG_ENABLE([tracing],
[AS_HELP_STRING([--enable-tracing=none|nvtx|roctx|timer],[enable tracing])],
[ac_TRACING=${enable_tracing}], [ac_TRACING=none])
############### Symplectic group
AC_ARG_ENABLE([symplectic],
[AC_HELP_STRING([--enable-symplectic=yes|no], [enable gauge group Sp2n])],
[ac_ENABLE_SP=${enable_symplectic}], [ac_ENABLE_SP=no])
AM_CONDITIONAL(BUILD_SP, [ test "${ac_ENABLE_SP}X" == "yesX" ])
case ${ac_ENABLE_SP} in
yes)
AC_DEFINE([Sp2n_config],[1],[gauge group Sp2n], [have_sp2n=true]);;
no)
AC_DEFINE([Sp2n_config],[0],[gauge group SUn], [have_sp2n=false]);;
case ${ac_TRACING} in
nvtx)
AC_DEFINE([GRID_TRACING_NVTX],[1],[use NVTX])
LIBS="${LIBS} -lnvToolsExt64_1"
;;
roctx)
AC_DEFINE([GRID_TRACING_ROCTX],[1],[use ROCTX])
LIBS="${LIBS} -lroctx64"
;;
timer)
AC_DEFINE([GRID_TRACING_TIMER],[1],[use TIMER]);;
*)
AC_MSG_ERROR(["--enable-symplectic is either yes or no"]);;
AC_DEFINE([GRID_TRACING_NONE],[1],[no tracing]);;
esac
############### fermions
AC_ARG_ENABLE([fermion-reps],
[AC_HELP_STRING([--enable-fermion-reps=yes|no], [enable extra fermion representation support])],
[AS_HELP_STRING([--enable-fermion-reps=yes|no],[enable extra fermion representation support])],
[ac_FERMION_REPS=${enable_fermion_reps}], [ac_FERMION_REPS=yes])
AM_CONDITIONAL(BUILD_FERMION_REPS, [ test "${ac_FERMION_REPS}X" == "yesX" ])
AC_ARG_ENABLE([gparity],
[AC_HELP_STRING([--enable-gparity=yes|no], [enable G-parity support])],
[AS_HELP_STRING([--enable-gparity=yes|no],[enable G-parity support])],
[ac_GPARITY=${enable_gparity}], [ac_GPARITY=yes])
AM_CONDITIONAL(BUILD_GPARITY, [ test "${ac_GPARITY}X" == "yesX" ])
AC_ARG_ENABLE([zmobius],
[AC_HELP_STRING([--enable-zmobius=yes|no], [enable Zmobius support])],
[AS_HELP_STRING([--enable-zmobius=yes|no],[enable Zmobius support])],
[ac_ZMOBIUS=${enable_zmobius}], [ac_ZMOBIUS=yes])
AM_CONDITIONAL(BUILD_ZMOBIUS, [ test "${ac_ZMOBIUS}X" == "yesX" ])
@ -176,7 +180,7 @@ case ${ac_ZMOBIUS} in
esac
############### Nc
AC_ARG_ENABLE([Nc],
[AC_HELP_STRING([--enable-Nc=2|3|4], [enable number of colours])],
[AS_HELP_STRING([--enable-Nc=2|3|4|5],[enable number of colours])],
[ac_Nc=${enable_Nc}], [ac_Nc=3])
case ${ac_Nc} in
@ -197,9 +201,24 @@ case ${ac_Nc} in
AC_MSG_ERROR(["Unsupport gauge group choice Nc = ${ac_Nc}"]);;
esac
############### Symplectic group
AC_ARG_ENABLE([symplectic],
[AC_HELP_STRING([--enable-symplectic=yes|no], [enable gauge group Sp2n])],
[ac_ENABLE_SP=${enable_symplectic}], [ac_ENABLE_SP=no])
AM_CONDITIONAL(BUILD_SP, [ test "${ac_ENABLE_SP}X" == "yesX" ])
case ${ac_ENABLE_SP} in
yes)
AC_DEFINE([Sp2n_config],[1],[gauge group Sp2n], [have_sp2n=true]);;
no)
AC_DEFINE([Sp2n_config],[0],[gauge group SUn], [have_sp2n=false]);;
*)
AC_MSG_ERROR(["--enable-symplectic is either yes or no"]);;
############### FP16 conversions
AC_ARG_ENABLE([sfw-fp16],
[AC_HELP_STRING([--enable-sfw-fp16=yes|no], [enable software fp16 comms])],
[AS_HELP_STRING([--enable-sfw-fp16=yes|no],[enable software fp16 comms])],
[ac_SFW_FP16=${enable_sfw_fp16}], [ac_SFW_FP16=yes])
case ${ac_SFW_FP16} in
yes)
@ -211,11 +230,11 @@ esac
############### Default to accelerator cshift, but revert to host if UCX is buggy or other reasons
AC_ARG_ENABLE([accelerator-cshift],
[AC_HELP_STRING([--enable-accelerator-cshift=yes|no], [run cshift on the device])],
[AS_HELP_STRING([--enable-accelerator-cshift=yes|no],[run cshift on the device])],
[ac_ACC_CSHIFT=${enable_accelerator_cshift}], [ac_ACC_CSHIFT=yes])
AC_ARG_ENABLE([ucx-buggy],
[AC_HELP_STRING([--enable-ucx-buggy=yes|no], [enable workaround for UCX device buffer bugs])],
[AS_HELP_STRING([--enable-ucx-buggy=yes|no],[enable workaround for UCX device buffer bugs])],
[ac_UCXBUGGY=${enable_ucx_buggy}], [ac_UCXBUGGY=no])
case ${ac_UCXBUGGY} in
@ -233,7 +252,7 @@ esac
############### SYCL/CUDA/HIP/none
AC_ARG_ENABLE([accelerator],
[AC_HELP_STRING([--enable-accelerator=cuda|sycl|hip|none], [enable none,cuda,sycl,hip acceleration])],
[AS_HELP_STRING([--enable-accelerator=cuda|sycl|hip|none],[enable none,cuda,sycl,hip acceleration])],
[ac_ACCELERATOR=${enable_accelerator}], [ac_ACCELERATOR=none])
case ${ac_ACCELERATOR} in
cuda)
@ -256,7 +275,7 @@ esac
############### UNIFIED MEMORY
AC_ARG_ENABLE([unified],
[AC_HELP_STRING([--enable-unified=yes|no], [enable unified address space for accelerator loops])],
[AS_HELP_STRING([--enable-unified=yes|no],[enable unified address space for accelerator loops])],
[ac_UNIFIED=${enable_unified}], [ac_UNIFIED=yes])
case ${ac_UNIFIED} in
yes)
@ -270,10 +289,10 @@ esac
############### Intel libraries
AC_ARG_ENABLE([mkl],
[AC_HELP_STRING([--enable-mkl=yes|no|prefix], [enable Intel MKL for LAPACK & FFTW])],
[AS_HELP_STRING([--enable-mkl=yes|no|prefix],[enable Intel MKL for LAPACK & FFTW])],
[ac_MKL=${enable_mkl}], [ac_MKL=no])
AC_ARG_ENABLE([ipp],
[AC_HELP_STRING([--enable-ipp=yes|no|prefix], [enable Intel IPP for fast CRC32C])],
[AS_HELP_STRING([--enable-ipp=yes|no|prefix],[enable Intel IPP for fast CRC32C])],
[ac_IPP=${enable_ipp}], [ac_IPP=no])
case ${ac_MKL} in
@ -371,8 +390,7 @@ CXXFLAGS=$CXXFLAGS_CPY
LDFLAGS=$LDFLAGS_CPY
############### SIMD instruction selection
AC_ARG_ENABLE([simd],[AC_HELP_STRING([--enable-simd=code],
[select SIMD target (cf. README.md)])], [ac_SIMD=${enable_simd}], [ac_SIMD=GEN])
AC_ARG_ENABLE([simd],[AS_HELP_STRING([--enable-simd=code],[select SIMD target (cf. README.md)])], [ac_SIMD=${enable_simd}], [ac_SIMD=GEN])
AC_ARG_ENABLE([gen-simd-width],
[AS_HELP_STRING([--enable-gen-simd-width=size],
@ -416,11 +434,10 @@ case ${CXXTEST} in
fi
;;
hipcc)
# CXXFLAGS="$CXXFLAGS -Xcompiler -fno-strict-aliasing --expt-extended-lambda --expt-relaxed-constexpr"
CXXFLAGS="$CXXFLAGS -fno-strict-aliasing"
CXXLD=${CXX}
if test $ac_openmp = yes; then
CXXFLAGS="$CXXFLAGS -Xcompiler -fopenmp"
CXXFLAGS="$CXXFLAGS -fopenmp"
fi
;;
dpcpp)
@ -438,7 +455,13 @@ case ${ax_cv_cxx_compiler_vendor} in
clang|gnu)
case ${ac_SIMD} in
GPU)
AC_DEFINE([GPU_VEC],[1],[GPU vectorised 512bit])
AC_DEFINE([GPU_VEC],[1],[GPU vectorised])
AC_DEFINE_UNQUOTED([GEN_SIMD_WIDTH],[$ac_gen_simd_width],
[generic SIMD vector width (in bytes)])
SIMD_GEN_WIDTH_MSG=" (width= $ac_gen_simd_width)"
SIMD_FLAGS='';;
GPU-RRII)
AC_DEFINE([GPU_RRII],[1],[GPU vectorised with RRRR / IIII layout])
AC_DEFINE_UNQUOTED([GEN_SIMD_WIDTH],[$ac_gen_simd_width],
[generic SIMD vector width (in bytes)])
SIMD_GEN_WIDTH_MSG=" (width= $ac_gen_simd_width)"
@ -507,6 +530,12 @@ case ${ax_cv_cxx_compiler_vendor} in
GPU)
AC_DEFINE([GPU_VEC],[1],[GPU vectorised ])
SIMD_FLAGS='';;
GPU-RRII)
AC_DEFINE([GPU_RRII],[1],[GPU vectorised with RRRR / IIII layout])
AC_DEFINE_UNQUOTED([GEN_SIMD_WIDTH],[$ac_gen_simd_width],
[generic SIMD vector width (in bytes)])
SIMD_GEN_WIDTH_MSG=" (width= $ac_gen_simd_width)"
SIMD_FLAGS='';;
SSE4)
AC_DEFINE([SSE4],[1],[SSE4 intrinsics])
SIMD_FLAGS='-msse4.2 -xsse4.2';;
@ -554,8 +583,7 @@ AC_DEFINE([GRID_DEFAULT_PRECISION_DOUBLE],[1],[GRID_DEFAULT_PRECISION is DOUBLE]
#########################################################
###################### GRID ALLOCATOR ALIGNMENT ##
#########################################################
AC_ARG_ENABLE([alloc-align],[AC_HELP_STRING([--enable-alloc-align=2MB|4k],
[Alignment in bytes of GRID Allocator ])],[ac_ALLOC_ALIGN=${enable_alloc_align}],[ac_ALLOC_ALIGN=2MB])
AC_ARG_ENABLE([alloc-align],[AS_HELP_STRING([--enable-alloc-align=2MB|4k],[Alignment in bytes of GRID Allocator ])],[ac_ALLOC_ALIGN=${enable_alloc_align}],[ac_ALLOC_ALIGN=2MB])
case ${ac_ALLOC_ALIGN} in
4k)
AC_DEFINE([GRID_ALLOC_ALIGN],[(4096)],[GRID_ALLOC_ALIGN]);;
@ -564,8 +592,7 @@ case ${ac_ALLOC_ALIGN} in
*);;
esac
AC_ARG_ENABLE([alloc-cache],[AC_HELP_STRING([--enable-alloc-cache ],
[Cache a pool of recent "frees" to reuse])],[ac_ALLOC_CACHE=${enable_alloc_cache}],[ac_ALLOC_CACHE=yes])
AC_ARG_ENABLE([alloc-cache],[AS_HELP_STRING([--enable-alloc-cache ],[Cache a pool of recent "frees" to reuse])],[ac_ALLOC_CACHE=${enable_alloc_cache}],[ac_ALLOC_CACHE=yes])
case ${ac_ALLOC_CACHE} in
yes)
AC_DEFINE([ALLOCATION_CACHE],[1],[ALLOCATION_CACHE]);;
@ -576,20 +603,21 @@ esac
#########################################################
###################### set GPU device to rank in node ##
#########################################################
AC_ARG_ENABLE([setdevice],[AC_HELP_STRING([--enable-setdevice | --disable-setdevice],
[Set GPU to rank in node with cudaSetDevice or similar])],[ac_SETDEVICE=${enable_SETDEVICE}],[ac_SETDEVICE=no])
AC_ARG_ENABLE([setdevice],[AS_HELP_STRING([--enable-setdevice | --disable-setdevice],[Set GPU to rank in node with cudaSetDevice or similar])],[ac_SETDEVICE=${enable_SETDEVICE}],[ac_SETDEVICE=no])
case ${ac_SETDEVICE} in
yes);;
no)
yes)
echo ENABLE SET DEVICE
;;
*)
AC_DEFINE([GRID_DEFAULT_GPU],[1],[GRID_DEFAULT_GPU] )
echo DISABLE SET DEVICE
;;
esac
#########################################################
###################### Shared memory intranode #########
#########################################################
AC_ARG_ENABLE([shm],[AC_HELP_STRING([--enable-shm=shmopen|shmget|hugetlbfs|shmnone|nvlink|no],
[Select SHM allocation technique])],[ac_SHM=${enable_shm}],[ac_SHM=no])
AC_ARG_ENABLE([shm],[AS_HELP_STRING([--enable-shm=shmopen|shmget|hugetlbfs|shmnone|nvlink|no|none],[Select SHM allocation technique])],[ac_SHM=${enable_shm}],[ac_SHM=no])
case ${ac_SHM} in
@ -608,7 +636,7 @@ case ${ac_SHM} in
AC_DEFINE([GRID_MPI3_SHMGET],[1],[GRID_MPI3_SHMGET] )
;;
shmnone | no)
shmnone | no | none)
AC_DEFINE([GRID_MPI3_SHM_NONE],[1],[GRID_MPI3_SHM_NONE] )
;;
@ -626,25 +654,30 @@ case ${ac_SHM} in
esac
###################### Shared base path for SHMMMAP
AC_ARG_ENABLE([shmpath],[AC_HELP_STRING([--enable-shmpath=path],
[Select SHM mmap base path for hugetlbfs])],
AC_ARG_ENABLE([shmpath],[AS_HELP_STRING([--enable-shmpath=path],[Select SHM mmap base path for hugetlbfs])],
[ac_SHMPATH=${enable_shmpath}],
[ac_SHMPATH=/var/lib/hugetlbfs/global/pagesize-2MB/])
AC_DEFINE_UNQUOTED([GRID_SHM_PATH],["$ac_SHMPATH"],[Path to a hugetlbfs filesystem for MMAPing])
############### force MPI in SMP
AC_ARG_ENABLE([shm-force-mpi],[AC_HELP_STRING([--enable-shm-force-mpi],
[Force MPI within shared memory])],[ac_SHM_FORCE_MPI=${enable_shm_force_mpi}],[ac_SHM_FORCE_MPI=no])
AC_ARG_ENABLE([shm-force-mpi],[AS_HELP_STRING([--enable-shm-force-mpi],[Force MPI within shared memory])],[ac_SHM_FORCE_MPI=${enable_shm_force_mpi}],[ac_SHM_FORCE_MPI=no])
case ${ac_SHM_FORCE_MPI} in
yes)
AC_DEFINE([GRID_SHM_FORCE_MPI],[1],[GRID_SHM_FORCE_MPI] )
;;
*) ;;
esac
############### force MPI in SMP
AC_ARG_ENABLE([shm-fast-path],[AS_HELP_STRING([--enable-shm-fast-path],[Allow kernels to remote copy over intranode])],[ac_SHM_FAST_PATH=${enable_shm_fast_path}],[ac_SHM_FAST_PATH=no])
case ${ac_SHM_FAST_PATH} in
yes)
AC_DEFINE([SHM_FAST_PATH],[1],[SHM_FAST_PATH] )
;;
*) ;;
esac
############### communication type selection
AC_ARG_ENABLE([comms-threads],[AC_HELP_STRING([--enable-comms-threads | --disable-comms-threads],
[Use multiple threads in MPI calls])],[ac_COMMS_THREADS=${enable_comms_threads}],[ac_COMMS_THREADS=yes])
AC_ARG_ENABLE([comms-threads],[AS_HELP_STRING([--enable-comms-threads | --disable-comms-threads],[Use multiple threads in MPI calls])],[ac_COMMS_THREADS=${enable_comms_threads}],[ac_COMMS_THREADS=yes])
case ${ac_COMMS_THREADS} in
yes)
@ -654,8 +687,7 @@ case ${ac_COMMS_THREADS} in
esac
############### communication type selection
AC_ARG_ENABLE([comms],[AC_HELP_STRING([--enable-comms=none|mpi|mpi-auto],
[Select communications])],[ac_COMMS=${enable_comms}],[ac_COMMS=none])
AC_ARG_ENABLE([comms],[AS_HELP_STRING([--enable-comms=none|mpi|mpi-auto],[Select communications])],[ac_COMMS=${enable_comms}],[ac_COMMS=none])
case ${ac_COMMS} in
@ -689,8 +721,8 @@ AM_CONDITIONAL(BUILD_COMMS_MPI3, [ test "${comms_type}X" == "mpi3X" ] )
AM_CONDITIONAL(BUILD_COMMS_NONE, [ test "${comms_type}X" == "noneX" ])
############### RNG selection
AC_ARG_ENABLE([rng],[AC_HELP_STRING([--enable-rng=ranlux48|mt19937|sitmo],\
[Select Random Number Generator to be used])],\
AC_ARG_ENABLE([rng],[AS_HELP_STRING([--enable-rng=ranlux48|mt19937|sitmo],[\
Select Random Number Generator to be used])],\
[ac_RNG=${enable_rng}],[ac_RNG=sitmo])
case ${ac_RNG} in
@ -709,8 +741,8 @@ case ${ac_RNG} in
esac
############### Timer option
AC_ARG_ENABLE([timers],[AC_HELP_STRING([--enable-timers],\
[Enable system dependent high res timers])],\
AC_ARG_ENABLE([timers],[AS_HELP_STRING([--enable-timers],[\
Enable system dependent high res timers])],\
[ac_TIMERS=${enable_timers}],[ac_TIMERS=yes])
case ${ac_TIMERS} in
@ -726,8 +758,7 @@ case ${ac_TIMERS} in
esac
############### Chroma regression test
AC_ARG_ENABLE([chroma],[AC_HELP_STRING([--enable-chroma],
[Expect chroma compiled under c++11 ])],ac_CHROMA=yes,ac_CHROMA=no)
AC_ARG_ENABLE([chroma],[AS_HELP_STRING([--enable-chroma],[Expect chroma compiled under c++11 ])],ac_CHROMA=yes,ac_CHROMA=no)
case ${ac_CHROMA} in
yes|no)