Lines Matching defs:gpuDynInst
323 GPUDynInstPtr gpuDynInst =
327 gpuDynInst->useContinuation = false;
328 injectGlobalMemFence(gpuDynInst, true);
628 GPUDynInstPtr gpuDynInst = sender_state->_gpuDynInst;
633 computeUnit->wfList[gpuDynInst->simdId][gpuDynInst->wfSlotId];
648 computeUnit->cu_id, gpuDynInst->simdId,
649 gpuDynInst->wfSlotId, w->barrierCnt);
651 if (gpuDynInst->useContinuation) {
652 assert(!gpuDynInst->isNoScope());
653 gpuDynInst->execContinuation(gpuDynInst->staticInstruction(),
654 gpuDynInst);
661 if (gpuDynInst->useContinuation) {
662 assert(!gpuDynInst->isNoScope());
663 gpuDynInst->execContinuation(gpuDynInst->staticInstruction(),
664 gpuDynInst);
676 computeUnit->cu_id, gpuDynInst->simdId, gpuDynInst->wfSlotId,
693 GPUDynInstPtr gpuDynInst M5_VAR_USED = retries.front().second;
695 computeUnit->cu_id, gpuDynInst->simdId, gpuDynInst->wfSlotId,
743 ComputeUnit::sendRequest(GPUDynInstPtr gpuDynInst, int index, PacketPtr pkt)
751 pkt->req->setPC(gpuDynInst->wavefront()->pc());
753 pkt->req->setReqInstSeqNum(gpuDynInst->seqNum());
782 cu_id, gpuDynInst->simdId, gpuDynInst->wfSlotId, vaddr);
790 cu_id, gpuDynInst->simdId, gpuDynInst->wfSlotId,
797 pkt->senderState = new DTLBPort::SenderState(gpuDynInst, index);
839 pkt->senderState = new ComputeUnit::DataPort::SenderState(gpuDynInst,
842 gpuDynInst->memStatusVector[pkt->getAddr()].push_back(index);
843 gpuDynInst->tlbHitLevel[index] = hit_level;
852 "scheduled\n", cu_id, gpuDynInst->simdId,
853 gpuDynInst->wfSlotId, index, pkt->req->getPaddr());
860 "failed!\n", cu_id, gpuDynInst->simdId, gpuDynInst->wfSlotId,
872 "failed!\n", cu_id, gpuDynInst->simdId, gpuDynInst->wfSlotId,
879 cu_id, gpuDynInst->simdId, gpuDynInst->wfSlotId, tmp_vaddr);
883 gpuDynInst->statusBitVector = VectorMask(0);
885 gpuDynInst->statusBitVector &= (~(1ll << index));
908 gpuDynInst->simdId, gpuDynInst->wfSlotId, index,
923 ComputeUnit::sendSyncRequest(GPUDynInstPtr gpuDynInst, int index, PacketPtr pkt)
930 pkt->senderState = new ComputeUnit::DataPort::SenderState(gpuDynInst, index,
934 cu_id, gpuDynInst->simdId, gpuDynInst->wfSlotId, index,
941 ComputeUnit::injectGlobalMemFence(GPUDynInstPtr gpuDynInst, bool kernelLaunch,
944 assert(gpuDynInst->isGlobalSeg());
948 0, 0, 0, 0, masterId(), 0, gpuDynInst->wfDynId);
959 gpuDynInst->setRequestFlags(req, kernelLaunch);
969 new ComputeUnit::DataPort::SenderState(gpuDynInst, 0, nullptr);
972 sendSyncRequest(gpuDynInst, 0, pkt);
981 GPUDynInstPtr gpuDynInst = sender_state->_gpuDynInst;
984 assert(gpuDynInst);
987 compute_unit->cu_id, gpuDynInst->simdId, gpuDynInst->wfSlotId,
993 int index = gpuDynInst->memStatusVector[paddr].back();
998 gpuDynInst->memStatusVector[paddr].pop_back();
999 gpuDynInst->pAddr = pkt->req->getPaddr();
1003 if (gpuDynInst->n_reg <= MAX_REGS_FOR_NON_VEC_MEM_INST) {
1004 gpuDynInst->statusBitVector &= (~(1ULL << index));
1006 assert(gpuDynInst->statusVector[index] > 0);
1007 gpuDynInst->statusVector[index]--;
1009 if (!gpuDynInst->statusVector[index])
1010 gpuDynInst->statusBitVector &= (~(1ULL << index));
1014 gpuDynInst->statusBitVector);
1016 if (gpuDynInst->statusBitVector == VectorMask(0)) {
1017 auto iter = gpuDynInst->memStatusVector.begin();
1018 auto end = gpuDynInst->memStatusVector.end();
1025 gpuDynInst->memStatusVector.clear();
1027 if (gpuDynInst->n_reg > MAX_REGS_FOR_NON_VEC_MEM_INST)
1028 gpuDynInst->statusVector.clear();
1030 compute_unit->globalMemoryPipe.handleResponse(gpuDynInst);
1033 compute_unit->cu_id, gpuDynInst->simdId,
1034 gpuDynInst->wfSlotId);
1040 if (gpuDynInst->useContinuation) {
1041 assert(!gpuDynInst->isNoScope());
1042 gpuDynInst->execContinuation(
1043 gpuDynInst->staticInstruction(),
1044 gpuDynInst);
1049 gpuDynInst->statusBitVector = VectorMask(0);
1051 if (gpuDynInst->useContinuation) {
1052 assert(!gpuDynInst->isNoScope());
1053 gpuDynInst->execContinuation(gpuDynInst->staticInstruction(),
1054 gpuDynInst);
1113 GPUDynInstPtr gpuDynInst = sender_state->_gpuDynInst;
1116 gpuDynInst->memStatusVector[line].push_back(mp_index);
1117 gpuDynInst->tlbHitLevel[mp_index] = hit_level;
1133 int simdId = gpuDynInst->simdId;
1134 int wfSlotId = gpuDynInst->wfSlotId;
1219 new ComputeUnit::DataPort::SenderState(gpuDynInst, mp_index,
1228 computeUnit->cu_id, gpuDynInst->simdId,
1229 gpuDynInst->wfSlotId, mp_index, new_pkt->req->getPaddr());
1257 GPUDynInstPtr gpuDynInst = sender_state->_gpuDynInst;
1261 retries.push_back(std::make_pair(pkt, gpuDynInst));
1265 compute_unit->cu_id, gpuDynInst->simdId,
1266 gpuDynInst->wfSlotId, index,
1271 compute_unit->cu_id, gpuDynInst->simdId,
1272 gpuDynInst->wfSlotId, index,
1669 ComputeUnit::updateInstStats(GPUDynInstPtr gpuDynInst)
1671 if (gpuDynInst->isScalar()) {
1672 if (gpuDynInst->isALU() && !gpuDynInst->isWaitcnt()) {
1675 } else if (gpuDynInst->isLoad()) {
1677 } else if (gpuDynInst->isStore()) {
1681 if (gpuDynInst->isALU()) {
1684 threadCyclesVALU += gpuDynInst->wavefront()->execMask().count();
1685 } else if (gpuDynInst->isFlat()) {
1686 if (gpuDynInst->isLocalMem()) {
1691 } else if (gpuDynInst->isLocalMem()) {
1693 } else if (gpuDynInst->isLoad()) {
1695 } else if (gpuDynInst->isStore()) {
1797 ComputeUnit::sendToLds(GPUDynInstPtr gpuDynInst)
1808 newPacket->senderState = new LDSPort::SenderState(gpuDynInst);
1824 GPUDynInstPtr gpuDynInst = senderState->getMemInst();
1829 computeUnit->localMemoryPipe.getLMRespFIFO().push(gpuDynInst);
1845 GPUDynInstPtr gpuDynInst M5_VAR_USED = sender_state->getMemInst();
1853 computeUnit->cu_id, gpuDynInst->simdId,
1854 gpuDynInst->wfSlotId);
1863 computeUnit->cu_id, gpuDynInst->simdId,
1864 gpuDynInst->wfSlotId, pkt->req->getPaddr());
1868 computeUnit->cu_id, gpuDynInst->simdId,
1869 gpuDynInst->wfSlotId, pkt->req->getPaddr());