TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCalloc)
{
const size_t size = 1024;
const size_t num = 1;
char *default_str = NULL;
default_str = (char *)memkind_calloc(pmem_kind, num, size);
EXPECT_TRUE(NULL != default_str);
EXPECT_EQ(*default_str, 0);
sprintf(default_str, "memkind_calloc MEMKIND_PMEM\n");
printf("%s", default_str);
memkind_free(pmem_kind, default_str);
// allocate the buffer of the same size (likely at the same address)
default_str = (char *)memkind_calloc(pmem_kind, num, size);
EXPECT_TRUE(NULL != default_str);
EXPECT_EQ(*default_str, 0);
sprintf(default_str, "memkind_calloc MEMKIND_PMEM\n");
printf("%s", default_str);
memkind_free(pmem_kind, default_str);
}
////////////////////////////////////////////////////////////////////////////
// My memkind free function calling memkind_free (with Kind=0).
// Note: memkind_free does not need the exact kind, if kind is 0. Then
// the library can figure out the proper kind itself.
////////////////////////////////////////////////////////////////////////////
void myMemkindFree(void *ptr)
{
DBG(2) printf("In my memkind free, ptr:%p\n", ptr);
memkind_free(0, ptr);
}
void test_allocation(memkind_t kind, size_t size)
{
ASSERT_TRUE(kind != NULL);
void* ptr = memkind_malloc(kind, size);
ASSERT_TRUE(ptr != NULL);
void* memset_ret = memset(ptr, 3, size);
ASSERT_TRUE(memset_ret != NULL);
memkind_free(kind, ptr);
}
void HBWSpace::deallocate( void * const arg_alloc_ptr , const size_t arg_alloc_size ) const
{
if ( arg_alloc_ptr ) {
if ( m_alloc_mech == STD_MALLOC ) {
void * alloc_ptr = *(reinterpret_cast<void **>(arg_alloc_ptr) -1);
memkind_free(MEMKIND_TYPE, alloc_ptr );
}
}
}
TEST_F(FreeingMemorySegfault, test_TC_MEMKIND_freeing_memory_after_thread_finish)
{
void* ptr = nullptr;
std::thread t([&] {
ptr = memkind_malloc(MEMKIND_DEFAULT, 32);
ASSERT_TRUE(ptr != NULL);
});
t.join();
memkind_free(0, ptr);
SUCCEED();
}
TEST_F(MemkindDefaultTests, test_TC_MEMKIND_DefaultCalloc)
{
const size_t size = 1024;
const size_t num = 1;
char *default_str = NULL;
default_str = (char *)memkind_calloc(MEMKIND_DEFAULT, num, size);
EXPECT_TRUE(NULL != default_str);
sprintf(default_str, "memkind_calloc MEMKIND_DEFAULT\n");
printf("%s", default_str);
memkind_free(MEMKIND_DEFAULT, default_str);
}
static void remove_info_entry_from_state(struct distmem *dist_kind, void *ptr) {
int hash_key = get_hashkey(ptr);
struct distmem_memory_information_generic *specific_state = dist_kind->internal_state[hash_key], *last_entry = NULL;
while (specific_state != NULL) {
if (specific_state->ptr == ptr) {
if (last_entry == NULL) {
dist_kind->internal_state[hash_key] = specific_state->next;
} else {
last_entry->next = specific_state->next;
}
if (specific_state->deallocate_specific_information != NULL) {
specific_state->deallocate_specific_information(specific_state->specific_information);
} else {
if (specific_state->specific_information != NULL) {
memkind_free(MEMKIND_DEFAULT, specific_state->specific_information);
}
}
memkind_free(MEMKIND_DEFAULT, specific_state);
break;
}
last_entry = specific_state;
specific_state = specific_state->next;
}
}
////////////////////////////////////////////////////////////////////////////
// This function is executed at library load time.
// Initilize HBW arena by making a dummy allocation/free at library load
// time. Until HBW initialization is complete, we must not call any
// allocation routines with HBW as kind.
////////////////////////////////////////////////////////////////////////////
void __attribute__ ((constructor)) autohbw_load(void)
{
// First set the default memory type this library allocates. This can
// be overridden by env variable
// Note: 'memkind_hbw_preferred' will allow falling back to DDR but
// 'memkind_hbw will not'
// Note: If HBM is not installed on a system, memkind_hbw_preferred call
// woudl fail. Therefore, we need to check for availability first.
//
int ret = 0;
if (memkind_check_available(MEMKIND_HBW) == 0) {
ret = memkind_get_kind_by_name("memkind_hbw_preferred", &HBW_Type);
}
else {
printf("WARN: *** No HBM found in system. Will use default (DDR) "
"OR user specifid type ***\n");
ret = memkind_get_kind_by_name("memkind_default", &HBW_Type);
}
assert(!ret && "FATAL: Could not find default memory type\n");
// Read any env variables. This has to be done first because DbgLevel
// is set using env variables and debug printing is used below
//
setEnvValues(); // read any env variables
DBG(1) printf("INFO: autohbw.so loaded!\n");
// dummy HBW call to initialize HBW arena
//
void *pp = memkind_malloc(HBW_Type, 16);
//
if (pp) {
// We have successfully initilized HBW arena
//
DBG(2) printf("\t-HBW int call succeeded\n");
memkind_free(0, pp);
MemkindInitDone = TRUE; // enable HBW allocation
}
else {
errPrn("\t-HBW init call FAILED. Is required memory type present on your system?\n");
assert(0 && "HBW/memkind initialization faild");
}
}
TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMalloc)
{
const size_t size = 1024;
char *default_str = NULL;
default_str = (char *)memkind_malloc(pmem_kind, size);
EXPECT_TRUE(NULL != default_str);
sprintf(default_str, "memkind_malloc MEMKIND_PMEM\n");
printf("%s", default_str);
memkind_free(pmem_kind, default_str);
// Out of memory
default_str = (char *)memkind_malloc(pmem_kind, 2 * PMEM_PART_SIZE);
EXPECT_EQ(NULL, default_str);
}
int main(int argc, char *argv[]) {
MPI_Init(&argc, &argv);
int myrank;
MPI_Comm_rank(MPI_COMM_WORLD, &myrank);
distmem_mpi_init();
int * buffer=(int*) distmem_mpi_malloc(MPI_CONTIGUOUS_KIND, 4, 17, MPI_COMM_WORLD);
struct distmem_mpi_memory_information * alloc_info=distmem_mpi_get_info(MPI_CONTIGUOUS_KIND, buffer);
printf("Total number elements=%zu, distributed over %d procs, local number elements=%zu\n", alloc_info->total_number_elements,
alloc_info->procs_distributed_over, alloc_info->number_local_elements);
int i;
for (i=0; i<alloc_info->number_local_elements; i++) {
buffer[i]=i*(myrank+1);
}
memkind_free(MPI_CONTIGUOUS_KIND, buffer);
MPI_Finalize();
return 0;
}
TEST_F(MemkindDefaultTests, test_TC_MEMKIND_DefaultGetSize)
{
const size_t size = 512;
char *default_str = NULL;
int err = 0;
size_t *total, *free;
total = (size_t*) malloc(sizeof(size_t));
free = (size_t*) malloc(sizeof(size_t));
default_str = (char *)memkind_malloc(MEMKIND_DEFAULT, size);
EXPECT_TRUE(NULL != default_str);
err = memkind_get_size(MEMKIND_DEFAULT, total, free);
EXPECT_EQ(0,err);
memkind_free(MEMKIND_DEFAULT, default_str);
}
TEST_F(MemkindPmemTests, PmemGetSize)
{
const size_t size = 512;
char *default_str = NULL;
int err = 0;
size_t total;
size_t free;
default_str = (char *)memkind_malloc(pmem_kind, size);
EXPECT_TRUE(NULL != default_str);
err = memkind_get_size(pmem_kind, &total, &free);
EXPECT_EQ(0, err);
// requested PMEM partition size is internally aligned to 4MB
EXPECT_EQ(total, (size_t)roundup(PMEM_PART_SIZE, CHUNK_SIZE));
memkind_free(pmem_kind, default_str);
}
TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemRealloc)
{
const size_t size1 = 512;
const size_t size2 = 1024;
char *default_str = NULL;
default_str = (char *)memkind_realloc(pmem_kind, default_str, size1);
EXPECT_TRUE(NULL != default_str);
sprintf(default_str, "memkind_realloc MEMKIND_PMEM with size %zu\n", size1);
printf("%s", default_str);
default_str = (char *)memkind_realloc(pmem_kind, default_str, size2);
EXPECT_TRUE(NULL != default_str);
sprintf(default_str, "memkind_realloc MEMKIND_PMEM with size %zu\n", size2);
printf("%s", default_str);
memkind_free(pmem_kind, default_str);
}
TEST_F(MemkindDefaultTests, test_TC_MEMKIND_DefaultRealloc)
{
const size_t size1 = 512;
const size_t size2 = 1024;
char *default_str = NULL;
default_str = (char *)memkind_realloc(MEMKIND_DEFAULT, default_str, size1);
EXPECT_TRUE(NULL != default_str);
sprintf(default_str, "memkind_realloc MEMKIND_DEFAULT with size %zu\n", size1);
printf("%s", default_str);
default_str = (char *)memkind_realloc(MEMKIND_DEFAULT, default_str, size2);
EXPECT_TRUE(NULL != default_str);
sprintf(default_str, "memkind_realloc MEMKIND_DEFAULT with size %zu\n", size2);
printf("%s", default_str);
memkind_free(MEMKIND_DEFAULT, default_str);
}
void *memkind_arena_realloc(struct memkind *kind, void *ptr, size_t size)
{
int err = 0;
unsigned int arena;
if (size == 0 && ptr != NULL) {
memkind_free(kind, ptr);
ptr = NULL;
}
else {
err = kind->ops->get_arena(kind, &arena, size);
if (!err) {
if (ptr == NULL) {
ptr = jemk_mallocx_check(size, MALLOCX_ARENA(arena));
}
else {
ptr = jemk_rallocx_check(ptr, size, MALLOCX_ARENA(arena));
}
}
}
return ptr;
}
int
main(int argc, char **argv)
{
int quantum, checktick();
int BytesPerWord;
int k;
ssize_t j;
STREAM_TYPE scalar;
double t, times[4][NTIMES];
#ifdef ENABLE_DYNAMIC_ALLOC
int err = 0;
memkind_t kind;
char err_msg[ERR_MSG_SIZE];
if (argc > 1 && (strncmp("--help", argv[1], strlen("--help")) == 0 ||
strncmp("-h", argv[1], strlen("-h")) == 0)) {
printf("Usage: %s [memkind_default | memkind_hbw | memkind_hbw_hugetlb | \n"
" memkind_hbw_preferred | memkind_hbw_preferred_hugetlb | \n"
" memkind_hbw_gbtlb | memkind_hbw_preferred_gbtlb | memkind_gbtlb | \n"
" memkind_hbw_interleave | memkind_interleave]\n",
argv[0]);
return 0;
}
#endif
/* --- SETUP --- determine precision and check timing --- */
printf(HLINE);
printf("STREAM version $Revision: 5.10 $\n");
#ifdef ENABLE_DYNAMIC_ALLOC
printf("Variant that uses the memkind library for dynamic memory allocation.\n");
#endif
printf(HLINE);
BytesPerWord = sizeof(STREAM_TYPE);
printf("This system uses %d bytes per array element.\n",
BytesPerWord);
printf(HLINE);
#ifdef N
printf("***** WARNING: ******\n");
printf(" It appears that you set the preprocessor variable N when compiling this code.\n");
printf(" This version of the code uses the preprocesor variable STREAM_ARRAY_SIZE to control the array size\n");
printf(" Reverting to default value of STREAM_ARRAY_SIZE=%llu\n",(unsigned long long) STREAM_ARRAY_SIZE);
printf("***** WARNING: ******\n");
#endif
printf("Array size = %llu (elements), Offset = %d (elements)\n" , (unsigned long long) STREAM_ARRAY_SIZE, OFFSET);
printf("Memory per array = %.1f MiB (= %.1f GiB).\n",
BytesPerWord * ( (double) STREAM_ARRAY_SIZE / 1024.0/1024.0),
BytesPerWord * ( (double) STREAM_ARRAY_SIZE / 1024.0/1024.0/1024.0));
printf("Total memory required = %.1f MiB (= %.1f GiB).\n",
(3.0 * BytesPerWord) * ( (double) STREAM_ARRAY_SIZE / 1024.0/1024.),
(3.0 * BytesPerWord) * ( (double) STREAM_ARRAY_SIZE / 1024.0/1024./1024.));
printf("Each kernel will be executed %d times.\n", NTIMES);
printf(" The *best* time for each kernel (excluding the first iteration)\n");
printf(" will be used to compute the reported bandwidth.\n");
#ifdef _OPENMP
printf(HLINE);
#pragma omp parallel
{
#pragma omp master
{
k = omp_get_num_threads();
printf ("Number of Threads requested = %i\n",k);
}
}
#endif
#ifdef _OPENMP
k = 0;
#pragma omp parallel
#pragma omp atomic
k++;
printf ("Number of Threads counted = %i\n",k);
#endif
#ifdef ENABLE_DYNAMIC_ALLOC
if (argc > 1) {
err = memkind_get_kind_by_name(argv[1], &kind);
}
else {
err = memkind_get_kind_by_name("memkind_default", &kind);
}
if (err) {
memkind_error_message(err, err_msg, ERR_MSG_SIZE);
fprintf(stderr, "ERROR: %s\n", err_msg);
return -1;
}
err = memkind_posix_memalign(kind, (void **)&a, 2097152, BytesPerWord * (STREAM_ARRAY_SIZE + OFFSET));
if (err) {
fprintf(stderr, "ERROR: Unable to allocate stream array a\n");
return -err;
}
err = memkind_posix_memalign(kind, (void **)&b, 2097152, BytesPerWord * (STREAM_ARRAY_SIZE + OFFSET));
if (err) {
fprintf(stderr, "ERROR: Unable to allocate stream array b\n");
return -err;
}
err = memkind_posix_memalign(kind, (void **)&c, 2097152, BytesPerWord * (STREAM_ARRAY_SIZE + OFFSET));
if (err) {
fprintf(stderr, "ERROR: Unable to allocate stream array c\n");
return -err;
}
#endif
/* Get initial value for system clock. */
#pragma omp parallel for
for (j=0; j<STREAM_ARRAY_SIZE; j++) {
a[j] = 1.0;
b[j] = 2.0;
c[j] = 0.0;
}
printf(HLINE);
if ( (quantum = checktick()) >= 1)
printf("Your clock granularity/precision appears to be "
"%d microseconds.\n", quantum);
else {
printf("Your clock granularity appears to be "
"less than one microsecond.\n");
quantum = 1;
}
t = mysecond();
#pragma omp parallel for
for (j = 0; j < STREAM_ARRAY_SIZE; j++)
a[j] = 2.0E0 * a[j];
t = 1.0E6 * (mysecond() - t);
printf("Each test below will take on the order"
" of %d microseconds.\n", (int) t );
printf(" (= %d clock ticks)\n", (int) (t/quantum) );
printf("Increase the size of the arrays if this shows that\n");
printf("you are not getting at least 20 clock ticks per test.\n");
printf(HLINE);
printf("WARNING -- The above is only a rough guideline.\n");
printf("For best results, please be sure you know the\n");
printf("precision of your system timer.\n");
printf(HLINE);
/* --- MAIN LOOP --- repeat test cases NTIMES times --- */
scalar = 3.0;
for (k=0; k<NTIMES; k++)
{
times[0][k] = mysecond();
#ifdef TUNED
tuned_STREAM_Copy();
#else
#pragma omp parallel for
for (j=0; j<STREAM_ARRAY_SIZE; j++)
c[j] = a[j];
#endif
times[0][k] = mysecond() - times[0][k];
times[1][k] = mysecond();
#ifdef TUNED
tuned_STREAM_Scale(scalar);
#else
#pragma omp parallel for
for (j=0; j<STREAM_ARRAY_SIZE; j++)
b[j] = scalar*c[j];
#endif
times[1][k] = mysecond() - times[1][k];
times[2][k] = mysecond();
#ifdef TUNED
tuned_STREAM_Add();
#else
#pragma omp parallel for
for (j=0; j<STREAM_ARRAY_SIZE; j++)
c[j] = a[j]+b[j];
#endif
times[2][k] = mysecond() - times[2][k];
times[3][k] = mysecond();
#ifdef TUNED
tuned_STREAM_Triad(scalar);
#else
#pragma omp parallel for
for (j=0; j<STREAM_ARRAY_SIZE; j++)
a[j] = b[j]+scalar*c[j];
#endif
times[3][k] = mysecond() - times[3][k];
}
/* --- SUMMARY --- */
for (k=1; k<NTIMES; k++) /* note -- skip first iteration */
{
for (j=0; j<4; j++)
{
avgtime[j] = avgtime[j] + times[j][k];
mintime[j] = MIN(mintime[j], times[j][k]);
maxtime[j] = MAX(maxtime[j], times[j][k]);
}
}
printf("Function Best Rate MB/s Avg time Min time Max time\n");
for (j=0; j<4; j++) {
avgtime[j] = avgtime[j]/(double)(NTIMES-1);
printf("%s%12.1f %11.6f %11.6f %11.6f\n", label[j],
1.0E-06 * bytes[j]/mintime[j],
avgtime[j],
mintime[j],
maxtime[j]);
}
printf(HLINE);
/* --- Check Results --- */
checkSTREAMresults();
printf(HLINE);
#ifdef ENABLE_DYNAMIC_ALLOC
memkind_free(kind, c);
memkind_free(kind, b);
memkind_free(kind, a);
#endif
return 0;
}
void mk_free(void* ptr)
{
assert(ptr);
memkind_free(MEMKIND_DEFAULT,ptr);
}
void mca_mpool_memkind_free(mca_mpool_base_module_t *mpool, void *addr)
{
mca_mpool_memkind_module_t *memkind_module = (mca_mpool_memkind_module_t *) mpool;
memkind_free(memkind_module->kind, addr);
}
void hbw_free(void *ptr)
{
memkind_free(0, ptr);
}
int main(int argc, char *argv[])
{
struct memkind *pmem_kind_unlimited = NULL;
int err = 0;
if (argc > 2) {
fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
return 1;
} else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
return 1;
}
fprintf(stdout,
"This example shows how to use multithreading with one main pmem kind."
"\nPMEM kind directory: %s\n", path);
// Create PMEM partition with unlimited size
err = memkind_create_pmem(path, 0, &pmem_kind_unlimited);
if (err) {
print_err_message(err);
return 1;
}
// Create a few threads which will access to our main pmem_kind
pthread_t pmem_threads[NUM_THREADS];
int *pmem_tint[NUM_THREADS][NUM_ALLOCS];
int t = 0, i = 0;
struct arg_struct *args[NUM_THREADS];
for (t = 0; t<NUM_THREADS; t++) {
args[t] = malloc(sizeof(struct arg_struct));
args[t]->id = t;
args[t]->ptr = &pmem_tint[t][0];
args[t]->kind = pmem_kind_unlimited;
if (pthread_create(&pmem_threads[t], NULL, thread_onekind,
(void *)args[t])!= 0) {
fprintf(stderr, "Unable to create a thread.\n");
return 1;
}
}
sleep(1);
if (pthread_cond_broadcast(&cond) != 0) {
fprintf(stderr, "Unable to broadcast a condition.\n");
return 1;
}
for (t = 0; t < NUM_THREADS; t++) {
if (pthread_join(pmem_threads[t], NULL) != 0) {
fprintf(stderr, "Thread join failed.\n");
return 1;
}
}
// Check if we can read the values outside of threads and free resources
for (t = 0; t < NUM_THREADS; t++) {
for (i = 0; i < NUM_ALLOCS; i++) {
if(*pmem_tint[t][i] != t) {
fprintf(stderr,
"pmem_tint value has not been saved correctly in the thread.\n");
return 1;
}
memkind_free(args[t]->kind, *(args[t]->ptr+i));
}
free(args[t]);
}
fprintf(stdout, "Threads successfully allocated memory in the PMEM kind.\n");
return 0;
}
MEMKIND_EXPORT void hbw_free(void *ptr)
{
memkind_free(0, ptr);
}
void distmem_free(struct memkind *kind, void *ptr) {
remove_info_entry_from_state(distmem_get_distkind_by_name(kind->name), ptr);
memkind_free(MEMKIND_DEFAULT, ptr);
}
void wrapped_free(void* ptr) {memkind_free(kind, ptr);}
