/gem5/src/gpu-compute/ |
H A D | local_memory_pipeline.cc | 41 #include "gpu-compute/shader.hh" 77 computeUnit->locMemToVrfBus.rdy() && (computeUnit->shader->coissue_return 86 computeUnit->shader->ScheduleAdd(&w->outstandingReqs, m->time, -1); 89 computeUnit->shader->ScheduleAdd(&w->outstandingReqsWrLm, 94 computeUnit->shader->ScheduleAdd(&w->outstandingReqsRdLm, 100 if (computeUnit->shader->coissue_return == 0)
|
H A D | dispatcher.cc | 45 #include "gpu-compute/shader.hh" 55 shader(p->shader_pointer), driver(p->cl_driver), 59 shader->handshake(this); 194 shader->ReadMem((uint64_t)(curTask.depends), &hs, 200 shader->WriteMem((uint64_t)(&((_cl_event*)hs.event)->start), 237 schedule(&tickEvent, curTick() + shader->ticks(1)); 279 shader->updateContext(ndRangeMap[execId].curCid); 282 if (!shader->dispatch_workgroups(&ndRangeMap[execId])) { 299 shader->hostWakeUp(cpu); 332 shader [all...] |
H A D | condition_register_state.cc | 40 #include "gpu-compute/shader.hh" 79 w->computeUnit->shader->tick_cnt + 80 w->computeUnit->shader->ticks(pipeLen), 0);
|
H A D | wavefront.cc | 42 #include "gpu-compute/shader.hh" 556 computeUnit->aluPipe[simdId].preset(computeUnit->shader-> 559 computeUnit->wfWait[simdId].preset(computeUnit->shader-> 562 computeUnit->wfWait[simdId].preset(computeUnit->shader-> 570 preset(computeUnit->shader->ticks(4)); 572 preset(computeUnit->shader->ticks(computeUnit->issuePeriod)); 575 preset(computeUnit->shader->ticks(4)); 577 preset(computeUnit->shader->ticks(computeUnit->issuePeriod)); 585 preset(computeUnit->shader->ticks(8)); 587 preset(computeUnit->shader [all...] |
H A D | global_memory_pipeline.cc | 43 #include "gpu-compute/shader.hh" 58 globalMemSize = computeUnit->shader->globalMemSize; 83 (computeUnit->shader->coissue_return || 93 computeUnit->shader->ScheduleAdd(&w->outstandingReqs, m->time, -1); 96 computeUnit->shader->ScheduleAdd(&w->outstandingReqsWrGm, 101 computeUnit->shader->ScheduleAdd(&w->outstandingReqsRdGm, 107 if (!computeUnit->shader->coissue_return)
|
H A D | fetch_unit.cc | 46 #include "gpu-compute/shader.hh" 70 timingSim = computeUnit->shader->timingSim; 72 fetchStatusQueue.resize(computeUnit->shader->n_wf); 74 for (int j = 0; j < computeUnit->shader->n_wf; ++j) { 85 for (int j = 0; j < computeUnit->shader->n_wf; ++j) { 165 computeUnit->shader->gpuTc, 192 computeUnit->shader->gpuTc);
|
H A D | vector_register_file.cc | 44 #include "gpu-compute/shader.hh" 186 computeUnit->shader->tick_cnt + 187 computeUnit->shader->ticks(pipeLen), 207 computeUnit->shader->tick_cnt + timestamp + 208 computeUnit->shader->ticks(delay), 0);
|
H A D | compute_unit.cc | 54 #include "gpu-compute/shader.hh" 166 for (int i = 0; i < shader->n_wf; ++i) { 207 if (timestampVec[i] <= shader->tick_cnt) { 322 if (shader->impl_kern_boundary_sync) { 336 for (int m = 0; m < shader->n_wf * numSIMDs; ++m) { 395 for (int j = 0; j < shader->n_wf; ++j) { 454 for (int i_wf = 0; i_wf < shader->n_wf; ++i_wf) { 553 glbMemToVrfBus.init(&shader->tick_cnt, shader->ticks(1)); 554 locMemToVrfBus.init(&shader [all...] |
H A D | dispatcher.hh | 85 Shader *shader; member in class:GpuDispatcher
|
H A D | lds_state.cc | 45 #include "gpu-compute/shader.hh" 216 parent->shader->ticks(bankConflicts * bankConflictPenalty) + 217 parent->shader->ticks(busLength);
|
H A D | scoreboard_check_stage.cc | 42 #include "gpu-compute/shader.hh" 141 for (int wvId = 0; wvId < computeUnit->shader->n_wf; ++wvId) {
|
H A D | shader.cc | 36 #include "gpu-compute/shader.hh" 71 cuList[i]->shader = this; 137 //Make sure both dispatcher and shader are trying to
|
H A D | compute_unit.hh | 186 Shader *shader; member in class:ComputeUnit
|
/gem5/tests/configs/ |
H A D | gpu-ruby.py | 172 # shader is the GPU 173 shader = Shader(n_wf = options.wfs_per_simd, variable 189 shader.impl_kern_boundary_sync = False 191 shader.impl_kern_boundary_sync = True 231 for k in range(int(shader.n_wf)): 246 shader.CUs = compute_units 248 # this is a uniprocessor only test, thus the shader is the second index in the 260 cpu_list = [cpu] + [shader] + [dispatcher] 341 shader.cpu_pointer = host_cpu 343 dispatcher.shader_pointer = shader [all...] |
/gem5/src/arch/hsail/insts/ |
H A D | mem_impl.hh | 239 m->latency.init(&w->computeUnit->shader->tick_cnt); 244 m->latency.set(w->computeUnit->shader->ticks(1)); 274 m->latency.set(w->computeUnit->shader->ticks(1)); 298 m->latency.set(w->computeUnit->shader->ticks(24)); 306 m->latency.set(w->computeUnit->shader->ticks(1)); 322 m->latency.set(w->computeUnit->shader->ticks(1)); 418 m->latency.init(&w->computeUnit->shader->tick_cnt); 423 m->latency.set(w->computeUnit->shader->ticks(1)); 451 m->latency.set(w->computeUnit->shader->ticks(1)); 470 m->latency.set(w->computeUnit->shader [all...] |
H A D | pseudo_inst.cc | 647 m->latency.init(&w->computeUnit->shader->tick_cnt); 650 m->latency.set(w->computeUnit->shader->ticks(64)); 687 m->latency.init(&w->computeUnit->shader->tick_cnt); 690 m->latency.set(w->computeUnit->shader->ticks(64)); 726 m->latency.init(&w->computeUnit->shader->tick_cnt); 729 m->latency.set(w->computeUnit->shader->ticks(1));
|
H A D | main.cc | 195 w->computeUnit->shader->dispatcher->scheduleDispatch();
|
H A D | mem.hh | 473 if (gpuDynInst->computeUnit()->shader-> 587 if (gpuDynInst->computeUnit()->shader->separate_acquire_release 1011 if (gpuDynInst->computeUnit()->shader->separate_acquire_release 1482 if (gpuDynInst->computeUnit()->shader->separate_acquire_release 1635 if (gpuDynInst->computeUnit()->shader-> 1671 if (gpuDynInst->computeUnit()->shader->separate_acquire_release
|
/gem5/configs/example/ |
H A D | apu_se.py | 202 # shader is the GPU 203 shader = Shader(n_wf = options.wfs_per_simd, variable 218 shader.impl_kern_boundary_sync = False 220 shader.impl_kern_boundary_sync = True 260 for k in range(shader.n_wf): 276 shader.CUs = compute_units 281 # The shader core will be whatever is after the CPU cores are accounted for 284 # The command processor will be whatever is after the shader is accounted for 296 shader.timing = True 308 # Leave shader [all...] |