mirror of
https://github.com/paboyle/Grid.git
synced 2025-06-12 20:27:06 +01:00
NUMA improvements worth preserving from AMD EPYC tests
This commit is contained in:
@ -195,7 +195,8 @@ public:
|
||||
#endif
|
||||
size_type bytes = __n*sizeof(_Tp);
|
||||
uint8_t *cp = (uint8_t *)ptr;
|
||||
#pragma omp parallel for
|
||||
// One touch per 4k page, static OMP loop to catch same loop order
|
||||
#pragma omp parallel for schedule(static)
|
||||
for(size_type n=0;n<bytes;n+=4096){
|
||||
cp[n]=0;
|
||||
}
|
||||
|
@ -37,7 +37,8 @@ Author: Peter Boyle <paboyle@ph.ed.ac.uk>
|
||||
#include <sys/ipc.h>
|
||||
#include <sys/shm.h>
|
||||
#include <sys/mman.h>
|
||||
//#include <zlib.h>
|
||||
#include <zlib.h>
|
||||
#include <numaif.h>
|
||||
#ifndef SHM_HUGETLB
|
||||
#define SHM_HUGETLB 04000
|
||||
#endif
|
||||
@ -214,6 +215,23 @@ void CartesianCommunicator::Init(int *argc, char ***argv) {
|
||||
void * ptr = mmap(NULL,size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
|
||||
if ( ptr == MAP_FAILED ) { perror("failed mmap"); assert(0); }
|
||||
assert(((uint64_t)ptr&0x3F)==0);
|
||||
|
||||
int status;
|
||||
int flags=MPOL_MF_MOVE;
|
||||
#ifdef KNL
|
||||
int nodes=1; // numa domain == MCDRAM
|
||||
// Find out if in SNC2,SNC4 mode ?
|
||||
#else
|
||||
int nodes=r; // numa domain == MPI ID
|
||||
#endif
|
||||
unsigned long count=1;
|
||||
for(uint64_t page=0;page<size;page+=4096){
|
||||
void *pages = (void *) ( page + (uint64_t)ptr );
|
||||
uint64_t *cow_it = (uint64_t *)pages; *cow_it = 1;
|
||||
ierr= move_pages(0,count, &pages,&nodes,&status,flags);
|
||||
if (ierr && (page==0)) perror("numa relocate command failed");
|
||||
}
|
||||
|
||||
ShmCommBufs[r] =ptr;
|
||||
|
||||
}
|
||||
|
Reference in New Issue
Block a user