src/gpu-compute/compute_unit.cc

81     globalSeqNum(0), wavefrontSize(p->wfSize),
93     fatal_if(p->wfSize > std::numeric_limits<unsigned long long>::digits ||
94              p->wfSize <= 0,
101         (uint32_t)ceil((double)(wfSize() * sizeof(uint32_t)) /
104     numCyclesPerLoadTransfer = (wfSize() * sizeof(uint32_t))
114             lastVaddrWF[j][i].resize(wfSize());
119             for (int k = 0; k < wfSize(); ++k) {
128         lastVaddrSimd[i].resize(wfSize(), 0);
131     lastVaddrCU.resize(wfSize());
143     memPort.resize(wfSize());
146     int tlbPort_width = perLaneTLB ? wfSize() : 1;
236     for (int k = 0; k < wfSize(); ++k) {
237         if (k + waveId * wfSize() < w->actualWgSzTotal)
245     for (int k = 0; k < wfSize(); ++k) {
246         w->workItemId[0][k] = (k + waveId * wfSize()) % w->actualWgSz[0];
247         w->workItemId[1][k] = ((k + waveId * wfSize()) / w->actualWgSz[0]) %
249         w->workItemId[2][k] = (k + waveId * wfSize()) /
257     w->barrierSlots = divCeil(w->actualWgSzTotal, wfSize());
259     w->barCnt.resize(wfSize(), 0);
266     ndr->q.privMemStart += ndr->q.privMemPerItem * wfSize();
269     ndr->q.spillMemStart += ndr->q.spillMemPerItem * wfSize();
299     if ((waveId + 1) * wfSize() >= w->actualWgSzTotal) {
300         w->spillWidth = w->actualWgSzTotal - (waveId * wfSize());
302         w->spillWidth = wfSize();
345             if (wave_id * wfSize() >= w->actualWgSzTotal)
386     int numWfs = (trueWgSizeTotal + wfSize() - 1) / wfSize();
1549        .init(0, wfSize(), 2)
1563        .init(1, wfSize(), 4)
1569         .init(1, wfSize(), 4)
1575         .init(1, wfSize(), 4)
1581         .init(1, wfSize(), 4)