Page MenuHomePhabricator
Differential D45528 Diff 144109 openmp/trunk/libomptarget/deviceRTLs/nvptx/src/loop.cu

Changeset View

Standalone View

View Options

openmp/trunk/libomptarget/deviceRTLs/nvptx/src/loop.cu

Show First 20 Lines • Show All 234 Lines • ▼ Show 20 Lines INLINE static void dispatch_init(kmp_sched_t schedule, T lb, T ub, ST st,
* various dynamic cases. (In paritcular, whether or not a stealing scheme * various dynamic cases. (In paritcular, whether or not a stealing scheme
* is legal). * is legal).
*/ */
schedule = SCHEDULE_WITHOUT_MODIFIERS(schedule); schedule = SCHEDULE_WITHOUT_MODIFIERS(schedule);
// Process schedule. // Process schedule.
if (tnum == 1 || tripCount <= 1 || OrderedSchedule(schedule)) { if (tnum == 1 || tripCount <= 1 || OrderedSchedule(schedule)) {
PRINT(LD_LOOP, PRINT(LD_LOOP,
"go sequential as tnum=%d, trip count %lld, ordered sched=%d\n", "go sequential as tnum=%ld, trip count %lld, ordered sched=%d\n",
tnum, P64(tripCount), schedule); (long)tnum, P64(tripCount), schedule);
schedule = kmp_sched_static_chunk; schedule = kmp_sched_static_chunk;
chunk = tripCount; // one thread gets the whole loop chunk = tripCount; // one thread gets the whole loop
} else if (schedule == kmp_sched_runtime) { } else if (schedule == kmp_sched_runtime) {
// process runtime // process runtime
omp_sched_t rtSched = currTaskDescr->GetRuntimeSched(); omp_sched_t rtSched = currTaskDescr->GetRuntimeSched();
chunk = currTaskDescr->RuntimeChunkSize(); chunk = currTaskDescr->RuntimeChunkSize();
switch (rtSched) { switch (rtSched) {
▲ Show 20 Lines • Show All 43 Lines • ▼ Show 20 Lines if (schedule == kmp_sched_static_chunk) {
T threadId = GetOmpThreadId(tid, isSPMDMode(), isRuntimeUninitialized()); T threadId = GetOmpThreadId(tid, isSPMDMode(), isRuntimeUninitialized());
int lastiter = 0; int lastiter = 0;
ForStaticChunk(lastiter, lb, ub, stride, chunk, threadId, tnum); ForStaticChunk(lastiter, lb, ub, stride, chunk, threadId, tnum);
// save computed params // save computed params
omptarget_nvptx_threadPrivateContext->Chunk(tid) = chunk; omptarget_nvptx_threadPrivateContext->Chunk(tid) = chunk;
omptarget_nvptx_threadPrivateContext->NextLowerBound(tid) = lb; omptarget_nvptx_threadPrivateContext->NextLowerBound(tid) = lb;
omptarget_nvptx_threadPrivateContext->Stride(tid) = stride; omptarget_nvptx_threadPrivateContext->Stride(tid) = stride;
PRINT(LD_LOOP, PRINT(LD_LOOP,
"dispatch init (static chunk) : num threads = %d, ub = %lld," "dispatch init (static chunk) : num threads = %d, ub = %" PRId64 ","
"next lower bound = %lld, stride = %lld\n", "next lower bound = %llu, stride = %llu\n",
GetNumberOfOmpThreads(tid, isSPMDMode(), isRuntimeUninitialized()), GetNumberOfOmpThreads(tid, isSPMDMode(), isRuntimeUninitialized()),
omptarget_nvptx_threadPrivateContext->LoopUpperBound(tid), omptarget_nvptx_threadPrivateContext->LoopUpperBound(tid),
omptarget_nvptx_threadPrivateContext->NextLowerBound(tid), omptarget_nvptx_threadPrivateContext->NextLowerBound(tid),
omptarget_nvptx_threadPrivateContext->Stride(tid)); omptarget_nvptx_threadPrivateContext->Stride(tid));
} else if (schedule == kmp_sched_static_nochunk) { } else if (schedule == kmp_sched_static_nochunk) {
ASSERT0(LT_FUSSY, chunk == 0, "bad chunk value"); ASSERT0(LT_FUSSY, chunk == 0, "bad chunk value");
// save ub // save ub
omptarget_nvptx_threadPrivateContext->LoopUpperBound(tid) = ub; omptarget_nvptx_threadPrivateContext->LoopUpperBound(tid) = ub;
// compute static chunk // compute static chunk
ST stride; ST stride;
T threadId = GetOmpThreadId(tid, isSPMDMode(), isRuntimeUninitialized()); T threadId = GetOmpThreadId(tid, isSPMDMode(), isRuntimeUninitialized());
int lastiter = 0; int lastiter = 0;
ForStaticNoChunk(lastiter, lb, ub, stride, chunk, threadId, tnum); ForStaticNoChunk(lastiter, lb, ub, stride, chunk, threadId, tnum);
// save computed params // save computed params
omptarget_nvptx_threadPrivateContext->Chunk(tid) = chunk; omptarget_nvptx_threadPrivateContext->Chunk(tid) = chunk;
omptarget_nvptx_threadPrivateContext->NextLowerBound(tid) = lb; omptarget_nvptx_threadPrivateContext->NextLowerBound(tid) = lb;
omptarget_nvptx_threadPrivateContext->Stride(tid) = stride; omptarget_nvptx_threadPrivateContext->Stride(tid) = stride;
PRINT(LD_LOOP, PRINT(LD_LOOP,
"dispatch init (static nochunk) : num threads = %d, ub = %lld," "dispatch init (static nochunk) : num threads = %d, ub = %" PRId64 ","
"next lower bound = %lld, stride = %lld\n", "next lower bound = %llu, stride = %llu\n",
GetNumberOfOmpThreads(tid, isSPMDMode(), isRuntimeUninitialized()), GetNumberOfOmpThreads(tid, isSPMDMode(), isRuntimeUninitialized()),
omptarget_nvptx_threadPrivateContext->LoopUpperBound(tid), omptarget_nvptx_threadPrivateContext->LoopUpperBound(tid),
omptarget_nvptx_threadPrivateContext->NextLowerBound(tid), omptarget_nvptx_threadPrivateContext->NextLowerBound(tid),
omptarget_nvptx_threadPrivateContext->Stride(tid)); omptarget_nvptx_threadPrivateContext->Stride(tid));
} else if (schedule == kmp_sched_dynamic || schedule == kmp_sched_guided) { } else if (schedule == kmp_sched_dynamic || schedule == kmp_sched_guided) {
if (chunk < 1) if (chunk < 1)
chunk = 1; chunk = 1;
Counter eventNum = ((tripCount - 1) / chunk) + 1; // number of chunks Counter eventNum = ((tripCount - 1) / chunk) + 1; // number of chunks
// but each thread (but one) must discover that it is last // but each thread (but one) must discover that it is last
eventNum += tnum; eventNum += tnum;
omptarget_nvptx_threadPrivateContext->Chunk(tid) = chunk; omptarget_nvptx_threadPrivateContext->Chunk(tid) = chunk;
omptarget_nvptx_threadPrivateContext->EventsNumber(tid) = eventNum; omptarget_nvptx_threadPrivateContext->EventsNumber(tid) = eventNum;
PRINT(LD_LOOP, PRINT(LD_LOOP,
"dispatch init (dyn) : num threads = %d, ub = %lld, chunk %lld, " "dispatch init (dyn) : num threads = %d, ub = %" PRId64 ", chunk %" PRIu64 ", "
"events number = %lld\n", "events number = %llu\n",
GetNumberOfOmpThreads(tid, isSPMDMode(), isRuntimeUninitialized()), GetNumberOfOmpThreads(tid, isSPMDMode(), isRuntimeUninitialized()),
omptarget_nvptx_threadPrivateContext->LoopUpperBound(tid), omptarget_nvptx_threadPrivateContext->LoopUpperBound(tid),
omptarget_nvptx_threadPrivateContext->Chunk(tid), omptarget_nvptx_threadPrivateContext->Chunk(tid),
omptarget_nvptx_threadPrivateContext->EventsNumber(tid)); omptarget_nvptx_threadPrivateContext->EventsNumber(tid));
} }
} }
//////////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////////
▲ Show 20 Lines • Show All 422 Lines • Show Last 20 Lines