summaryrefslogtreecommitdiff
path: root/libs/ode-0.16.1/ode/src/fastltsolve_impl.h
diff options
context:
space:
mode:
Diffstat (limited to 'libs/ode-0.16.1/ode/src/fastltsolve_impl.h')
-rw-r--r--libs/ode-0.16.1/ode/src/fastltsolve_impl.h1440
1 files changed, 1440 insertions, 0 deletions
diff --git a/libs/ode-0.16.1/ode/src/fastltsolve_impl.h b/libs/ode-0.16.1/ode/src/fastltsolve_impl.h
new file mode 100644
index 0000000..ca30d9c
--- /dev/null
+++ b/libs/ode-0.16.1/ode/src/fastltsolve_impl.h
@@ -0,0 +1,1440 @@
+
+
+/*************************************************************************
+ * *
+ * Open Dynamics Engine, Copyright (C) 2001,2002 Russell L. Smith. *
+ * All rights reserved. Email: russ@q12.org Web: www.q12.org *
+ * *
+ * This library is free software; you can redistribute it and/or *
+ * modify it under the terms of EITHER: *
+ * (1) The GNU Lesser General Public License as published by the Free *
+ * Software Foundation; either version 2.1 of the License, or (at *
+ * your option) any later version. The text of the GNU Lesser *
+ * General Public License is included with this library in the *
+ * file LICENSE.TXT. *
+ * (2) The BSD-style license that is included with this library in *
+ * the file LICENSE-BSD.TXT. *
+ * *
+ * This library is distributed in the hope that it will be useful, *
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of *
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the files *
+ * LICENSE.TXT and LICENSE-BSD.TXT for more details. *
+ * *
+ *************************************************************************/
+
+/*
+ * Code style improvements and optimizations by Oleh Derevenko ????-2019
+ * L1Transposed cooperative solving code of ThreadedEquationSolverLDLT copyright (c) 2017-2019 Oleh Derevenko, odar@eleks.com (change all "a" to "e")
+ */
+
+
+#ifndef _ODE_FASTLTSOLVE_IMPL_H_
+#define _ODE_FASTLTSOLVE_IMPL_H_
+
+
+/* solve L^T * x=b, with b containing 1 right hand side.
+ * L is an n*n lower triangular matrix with ones on the diagonal.
+ * L is stored by rows and its leading dimension is rowSkip.
+ * b is an n*1 matrix that contains the right hand side.
+ * b is overwritten with x.
+ * this processes blocks of 4.
+ */
+
+template<unsigned int b_stride>
+void solveL1Transposed(const dReal *L, dReal *B, unsigned rowCount, unsigned rowSkip)
+{
+ dIASSERT(rowCount != 0);
+
+ /* special handling for L and B because we're solving L1 *transpose* */
+ const dReal *lastLElement = L + (sizeint)(rowCount - 1) * (rowSkip + 1);
+ dReal *lastBElement = B + (sizeint)(rowCount - 1) * b_stride;
+
+ /* compute rows at end that are not a multiple of block size */
+ const unsigned loopX1RowCount = rowCount % 4;
+
+ unsigned blockStartRow = loopX1RowCount;
+ bool subsequentPass = false;
+
+ /* compute rightmost bottom X(i) block */
+ if (loopX1RowCount != 0)
+ {
+ subsequentPass = true;
+
+ const dReal *ptrLElement = lastLElement;
+ dReal *ptrBElement = lastBElement;
+
+ dReal Y11 = ptrBElement[0 * b_stride]/* - Z11*/;
+ // ptrBElement[0 * b_stride] = Y11; -- unchanged
+
+ if (loopX1RowCount >= 2)
+ {
+ dReal p2 = ptrLElement[-1];
+ dReal Y21 = ptrBElement[-1 * (int)b_stride]/* - Z21 */- p2 * Y11;
+ ptrBElement[-1 * (int)b_stride] = Y21;
+
+ if (loopX1RowCount > 2)
+ {
+ dReal p3 = ptrLElement[-2];
+ dReal p3_1 = (ptrLElement - rowSkip)[-2];
+ dReal Y31 = ptrBElement[-2 * (int)b_stride]/* - Z31 */- p3 * Y11 - p3_1 * Y21;
+ ptrBElement[-2 * (int)b_stride] = Y31;
+ }
+ }
+ }
+
+ /* compute all 4 x 1 blocks of X */
+ for (; !subsequentPass || blockStartRow < rowCount; subsequentPass = true, blockStartRow += 4)
+ {
+ /* compute all 4 x 1 block of X, from rows i..i+4-1 */
+
+ /* declare variables - Z matrix, p and q vectors, etc */
+ const dReal *ptrLElement;
+ dReal *ptrBElement;
+
+ dReal Z41, Z31, Z21, Z11;
+
+ if (subsequentPass)
+ {
+ ptrLElement = lastLElement - blockStartRow;
+ ptrBElement = lastBElement;
+
+ /* set the Z matrix to 0 */
+ Z41 = 0; Z31 = 0; Z21 = 0; Z11 = 0;
+
+ unsigned rowCounter = blockStartRow;
+
+ if (rowCounter % 2 != 0)
+ {
+ dReal q1, p4, p3, p2, p1;
+
+ /* load p and q values */
+ q1 = ptrBElement[0 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ ptrBElement -= 1 * b_stride;
+ rowCounter -= 1;
+ }
+
+ if (rowCounter % 4 != 0)
+ {
+ dReal q1, p4, p3, p2, p1;
+
+ /* load p and q values */
+ q1 = ptrBElement[0 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[-1 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ ptrBElement -= 2 * b_stride;
+ rowCounter -= 2;
+ }
+
+ /* the inner loop that computes outer products and adds them to Z */
+ for (bool exitLoop = rowCounter == 0; !exitLoop; exitLoop = false)
+ {
+ dReal q1, p4, p3, p2, p1;
+
+ /* load p and q values */
+ q1 = ptrBElement[0 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[-1 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[-2 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[-3 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ if (rowCounter > 12)
+ {
+ rowCounter -= 12;
+
+ ptrBElement -= 12 * b_stride;
+
+ /* load p and q values */
+ q1 = ptrBElement[8 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[7 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[6 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[5 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[4 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[3 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[2 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[1 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z41 += p4 * q1;
+ Z31 += p3 * q1;
+ Z21 += p2 * q1;
+ Z11 += p1 * q1;
+ }
+ else
+ {
+ ptrBElement -= 4 * b_stride;
+
+ if ((rowCounter -= 4) == 0)
+ {
+ break;
+ }
+ }
+ /* end of inner loop */
+ }
+ }
+ else
+ {
+ ptrLElement = lastLElement/* - blockStartRow*/; dIASSERT(blockStartRow == 0);
+ ptrBElement = lastBElement;
+
+ /* set the Z matrix to 0 */
+ Z41 = 0; Z31 = 0; Z21 = 0; Z11 = 0;
+ }
+
+ /* finish computing the X(i) block */
+ dReal Y11, Y21, Y31, Y41;
+ {
+ Y11 = ptrBElement[0 * b_stride] - Z11;
+ ptrBElement[0 * b_stride] = Y11;
+ }
+ {
+ dReal p2 = ptrLElement[-1];
+ Y21 = ptrBElement[-1 * (int)b_stride] - Z21 - p2 * Y11;
+ ptrBElement[-1 * (int)b_stride] = Y21;
+ }
+ {
+ dReal p3 = ptrLElement[-2];
+ dReal p3_1 = (ptrLElement - rowSkip)[-2];
+ Y31 = ptrBElement[-2 * (int)b_stride] - Z31 - p3 * Y11 - p3_1 * Y21;
+ ptrBElement[-2 * (int)b_stride] = Y31;
+ }
+ {
+ dReal p4 = ptrLElement[-3];
+ dReal p4_1 = (ptrLElement - rowSkip)[-3];
+ dReal p4_2 = (ptrLElement - rowSkip * 2)[-3];
+ Y41 = ptrBElement[-3 * (int)b_stride] - Z41 - p4 * Y11 - p4_1 * Y21 - p4_2 * Y31;
+ ptrBElement[-3 * (int)b_stride] = Y41;
+ }
+ /* end of outer loop */
+ }
+}
+
+
+
+template<unsigned int block_step>
+/*static */
+sizeint ThreadedEquationSolverLDLT::estimateCooperativelySolvingL1TransposedMemoryRequirement(unsigned rowCount, SolvingL1TransposedMemoryEstimates &ref_solvingMemoryEstimates)
+{
+ unsigned blockCount = deriveSolvingL1TransposedBlockCount(rowCount, block_step);
+ sizeint descriptorSizeRequired = dEFFICIENT_SIZE(sizeof(cellindexint) * blockCount);
+ sizeint contextSizeRequired = dEFFICIENT_SIZE(sizeof(SolveL1TransposedCellContext) * (CCI__MAX + 1) * blockCount);
+ ref_solvingMemoryEstimates.assignData(descriptorSizeRequired, contextSizeRequired);
+
+ sizeint totalSizeRequired = descriptorSizeRequired + contextSizeRequired;
+ return totalSizeRequired;
+}
+
+template<unsigned int block_step>
+/*static */
+void ThreadedEquationSolverLDLT::initializeCooperativelySolveL1TransposedMemoryStructures(unsigned rowCount,
+ atomicord32 &out_blockCompletionProgress, cellindexint *blockProgressDescriptors, SolveL1TransposedCellContext *dUNUSED(cellContexts))
+{
+ unsigned blockCount = deriveSolvingL1TransposedBlockCount(rowCount, block_step);
+
+ out_blockCompletionProgress = 0;
+ memset(blockProgressDescriptors, 0, blockCount * sizeof(*blockProgressDescriptors));
+}
+
+template<unsigned int block_step, unsigned int b_stride>
+/*static */
+void ThreadedEquationSolverLDLT::participateSolvingL1Transposed(const dReal *L, dReal *B, unsigned rowCount, unsigned rowSkip,
+ volatile atomicord32 &refBlockCompletionProgress/*=0*/, volatile cellindexint *blockProgressDescriptors/*=[blockCount]*/,
+ SolveL1TransposedCellContext *cellContexts/*=[CCI__MAX x blockCount] + [blockCount]*/, unsigned ownThreadIndex)
+{
+ const unsigned lookaheadRange = 32;
+ const unsigned blockCount = deriveSolvingL1TransposedBlockCount(rowCount, block_step);
+ /* compute rows at end that are not a multiple of block size */
+ const unsigned loopX1RowCount = rowCount % block_step;
+
+ /* special handling for L and B because we're solving L1 *transpose* */
+ const dReal *lastLElement = L + (rowCount - 1) * ((sizeint)rowSkip + 1);
+ dReal *lastBElement = B + (rowCount - 1) * (sizeint)b_stride;
+
+ /* elements adjusted as if the last block was full block_step elements */
+ unsigned x1AdjustmentElements = (block_step - loopX1RowCount) % block_step;
+ const dReal *columnAdjustedLastLElement = lastLElement + x1AdjustmentElements;
+ const dReal *fullyAdjustedLastLElement = columnAdjustedLastLElement + (sizeint)rowSkip * x1AdjustmentElements;
+ dReal *adjustedLastBElement = lastBElement + b_stride * x1AdjustmentElements;
+
+ BlockProcessingState blockProcessingState = BPS_NO_BLOCKS_PROCESSED;
+
+ unsigned completedBlocks = refBlockCompletionProgress;
+ unsigned currentBlock = completedBlocks;
+ dIASSERT(completedBlocks <= blockCount);
+
+ for (bool exitLoop = completedBlocks == blockCount; !exitLoop; exitLoop = false)
+ {
+ bool goForLockedBlockPrimaryCalculation = false, goForLockedBlockDuplicateCalculation = false;
+ bool goAssigningTheResult = false, stayWithinTheBlock = false;
+
+ dReal Z[block_step];
+ dReal Y[block_step];
+
+ dReal *ptrBElement;
+
+ CellContextInstance previousContextInstance;
+ unsigned completedRowBlock;
+ bool partialBlock;
+
+ for (cellindexint testDescriptor = blockProgressDescriptors[currentBlock]; ; )
+ {
+ if (testDescriptor == INVALID_CELLDESCRIPTOR)
+ {
+ // Invalid descriptor is the indication that the row has been fully calculated
+ // Test if this was the last row and break out if so.
+ if (currentBlock + 1 == blockCount)
+ {
+ exitLoop = true;
+ break;
+ }
+
+ // Treat detected row advancement as a row processed
+ // blockProcessingState = BPS_SOME_BLOCKS_PROCESSED; <-- performs better without it
+ break;
+ }
+
+ CooperativeAtomics::AtomicReadReorderBarrier();
+ // It is necessary to read up to date completedBblocks value after the descriptor retrieval
+ // as otherwise the logic below breaks
+ completedBlocks = refBlockCompletionProgress;
+
+ if (!GET_CELLDESCRIPTOR_ISLOCKED(testDescriptor))
+ {
+ completedRowBlock = GET_CELLDESCRIPTOR_COLUMNINDEX(testDescriptor);
+ dIASSERT(completedRowBlock < currentBlock || (completedRowBlock == currentBlock && currentBlock == 0)); // Otherwise, why would the calculation have had stopped if the final column is reachable???
+ dIASSERT(completedRowBlock <= completedBlocks); // Since the descriptor is not locked
+
+ if (completedRowBlock == completedBlocks && currentBlock != completedBlocks)
+ {
+ dIASSERT(completedBlocks < currentBlock);
+ break;
+ }
+
+ if (CooperativeAtomics::AtomicCompareExchangeCellindexint(&blockProgressDescriptors[currentBlock], testDescriptor, MARK_CELLDESCRIPTOR_LOCKED(testDescriptor)))
+ {
+ if (completedRowBlock != 0)
+ {
+ CellContextInstance contextInstance = GET_CELLDESCRIPTOR_CONTEXTINSTANCE(testDescriptor);
+ previousContextInstance = contextInstance;
+
+ const SolveL1TransposedCellContext &sourceContext = buildBlockContextRef(cellContexts, currentBlock, contextInstance);
+ sourceContext.loadPrecalculatedZs(Z);
+ }
+ else
+ {
+ previousContextInstance = CCI__MIN;
+ SolveL1TransposedCellContext::initializePrecalculatedZs(Z);
+ }
+
+ goForLockedBlockPrimaryCalculation = true;
+ break;
+ }
+
+ if (blockProcessingState != BPS_COMPETING_FOR_A_BLOCK)
+ {
+ break;
+ }
+
+ testDescriptor = blockProgressDescriptors[currentBlock];
+ }
+ else
+ {
+ if (blockProcessingState != BPS_COMPETING_FOR_A_BLOCK)
+ {
+ break;
+ }
+
+ cellindexint verificativeDescriptor;
+ bool verificationFailure = false;
+
+ completedRowBlock = GET_CELLDESCRIPTOR_COLUMNINDEX(testDescriptor);
+ dIASSERT(completedRowBlock != currentBlock || currentBlock == 0); // There is no reason for computations to stop at the very end other than being the initial value at the very first block
+
+ if (completedRowBlock != 0)
+ {
+ CellContextInstance contextInstance = GET_CELLDESCRIPTOR_CONTEXTINSTANCE(testDescriptor);
+ const SolveL1TransposedCellContext &sourceContext = buildBlockContextRef(cellContexts, currentBlock, contextInstance);
+ sourceContext.loadPrecalculatedZs(Z);
+ }
+ else
+ {
+ SolveL1TransposedCellContext::initializePrecalculatedZs(Z);
+ }
+
+ if (completedRowBlock != 0 && completedRowBlock <= currentBlock)
+ {
+ // Make sure the descriptor is re-read after the precalculates
+ CooperativeAtomics::AtomicReadReorderBarrier();
+ }
+
+ if (completedRowBlock <= currentBlock)
+ {
+ verificativeDescriptor = blockProgressDescriptors[currentBlock];
+ verificationFailure = verificativeDescriptor != testDescriptor;
+ }
+
+ if (!verificationFailure)
+ {
+ dIASSERT(completedRowBlock <= currentBlock + 1);
+
+ goForLockedBlockDuplicateCalculation = true;
+ break;
+ }
+
+ testDescriptor = verificativeDescriptor;
+ }
+ }
+
+ if (exitLoop)
+ {
+ break;
+ }
+
+ if (goForLockedBlockPrimaryCalculation)
+ {
+ blockProcessingState = BPS_SOME_BLOCKS_PROCESSED;
+
+ // Declare and assign the variables at the top to not interfere with any branching -- the compiler is going to eliminate them anyway.
+ bool handleComputationTakenOver = false, columnEndReached = false;
+
+ const dReal *ptrLElement;
+ unsigned finalRowBlock;
+
+ /* check if this is not the partial block of fewer rows */
+ if (currentBlock != 0 || loopX1RowCount == 0)
+ {
+ partialBlock = false;
+
+ ptrLElement = completedRowBlock != 0
+ ? fullyAdjustedLastLElement - currentBlock * block_step - (sizeint)(completedRowBlock * block_step) * rowSkip
+ : columnAdjustedLastLElement - currentBlock * block_step;
+ ptrBElement = completedRowBlock != 0
+ ? adjustedLastBElement - (sizeint)(completedRowBlock * block_step) * b_stride
+ : lastBElement;
+
+ finalRowBlock = dMACRO_MIN(currentBlock, completedBlocks);
+ dIASSERT(finalRowBlock != completedRowBlock || finalRowBlock == 0);
+
+ unsigned rowCounter = finalRowBlock - completedRowBlock;
+ bool exitLoop = rowCounter == 0;
+
+ if (exitLoop)
+ {
+ columnEndReached = true;
+ }
+ else if (completedRowBlock == 0 && currentBlock != 0 && loopX1RowCount != 0)
+ {
+ if ((loopX1RowCount & 1) != 0)
+ {
+ dReal q1, p4, p3, p2, p1;
+
+ /* load p and q values */
+ q1 = ptrBElement[0 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ ptrBElement -= 1 * b_stride;
+ }
+
+ if ((loopX1RowCount & 2) != 0)
+ {
+ dReal q1, p4, p3, p2, p1;
+
+ /* load p and q values */
+ q1 = ptrBElement[0 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[-1 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ ptrBElement -= 2 * b_stride;
+ }
+ dSASSERT(block_step == 4);
+
+ if (--rowCounter == 0)
+ {
+ do
+ {
+ if (finalRowBlock == currentBlock)
+ {
+ columnEndReached = true;
+ exitLoop = true;
+ break;
+ }
+
+ // Take a look if any more columns have been completed...
+ completedBlocks = refBlockCompletionProgress;
+ dIASSERT(completedBlocks >= finalRowBlock);
+
+ if (completedBlocks == finalRowBlock)
+ {
+ exitLoop = true;
+ break;
+ }
+
+ // ...continue if so.
+ unsigned rowCompletedSoFar = finalRowBlock;
+ finalRowBlock = dMACRO_MIN(currentBlock, completedBlocks);
+ rowCounter = finalRowBlock - rowCompletedSoFar;
+ }
+ while (false);
+ }
+ }
+
+ for (; !exitLoop; exitLoop = false)
+ {
+ dReal q1, p4, p3, p2, p1;
+
+ /* load p and q values */
+ q1 = ptrBElement[0 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[-1 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[-2 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[-3 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+ dSASSERT(block_step == 4);
+
+ if (rowCounter > 3)
+ {
+ rowCounter -= 3;
+
+ ptrBElement -= 3 * block_step * b_stride;
+
+ /* load p and q values */
+ q1 = ptrBElement[8 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[7 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[6 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[5 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[4 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[3 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[2 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[1 * b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+ dSASSERT(block_step == 4);
+ }
+ else
+ {
+ ptrBElement -= block_step * b_stride;
+
+ if (--rowCounter == 0)
+ {
+ if (finalRowBlock == currentBlock)
+ {
+ columnEndReached = true;
+ break;
+ }
+
+ // Take a look if any more columns have been completed...
+ completedBlocks = refBlockCompletionProgress;
+ dIASSERT(completedBlocks >= finalRowBlock);
+
+ if (completedBlocks == finalRowBlock)
+ {
+ break;
+ }
+
+ // ...continue if so.
+ unsigned rowCompletedSoFar = finalRowBlock;
+ finalRowBlock = dMACRO_MIN(currentBlock, completedBlocks);
+ rowCounter = finalRowBlock - rowCompletedSoFar;
+ }
+ }
+ /* end of inner loop */
+ }
+ }
+ else /* compute rightmost bottom X(i) block */
+ {
+ partialBlock = true;
+
+ ptrLElement = lastLElement;
+ ptrBElement = lastBElement;
+ dIASSERT(completedRowBlock == 0);
+
+ columnEndReached = true;
+ }
+
+ if (columnEndReached)
+ {
+ // Check whether there is still a need to proceed or if the computation has been taken over by another thread
+ cellindexint oldDescriptor = MAKE_CELLDESCRIPTOR(completedRowBlock, previousContextInstance, true);
+
+ if (blockProgressDescriptors[currentBlock] == oldDescriptor)
+ {
+ if (partialBlock)
+ {
+ Y[0] = ptrBElement[0 * b_stride]/* - Z[0]*/;
+
+ if (loopX1RowCount >= 2)
+ {
+ dReal p2 = ptrLElement[-1];
+ Y[1] = ptrBElement[-1 * (int)b_stride]/* - Z[1] */- p2 * Y[0];
+
+ if (loopX1RowCount > 2)
+ {
+ dReal p3 = ptrLElement[-2];
+ dReal p3_1 = (ptrLElement - rowSkip)[-2];
+ Y[2] = ptrBElement[-2 * (int)b_stride]/* - Z[2] */- p3 * Y[0] - p3_1 * Y[1];
+ }
+ }
+
+ dSASSERT(block_step == 4);
+ }
+ else
+ {
+ Y[0] = ptrBElement[0 * b_stride] - Z[0];
+
+ dReal p2 = ptrLElement[-1];
+ Y[1] = ptrBElement[-1 * (int)b_stride] - Z[1] - p2 * Y[0];
+
+ dReal p3 = ptrLElement[-2];
+ dReal p3_1 = (ptrLElement - rowSkip)[-2];
+ Y[2] = ptrBElement[-2 * (int)b_stride] - Z[2] - p3 * Y[0] - p3_1 * Y[1];
+
+ dReal p4 = ptrLElement[-3];
+ dReal p4_1 = (ptrLElement - rowSkip)[-3];
+ dReal p4_2 = (ptrLElement - rowSkip * 2)[-3];
+ Y[3] = ptrBElement[-3 * (int)b_stride] - Z[3] - p4 * Y[0] - p4_1 * Y[1] - p4_2 * Y[2];
+
+ dSASSERT(block_step == 4);
+ }
+
+ // Use atomic memory barrier to make sure memory reads of ptrBElement[] and blockProgressDescriptors[] are not swapped
+ CooperativeAtomics::AtomicReadReorderBarrier();
+
+ // The descriptor has not been altered yet - this means the ptrBElement[] values used above were not modified yet
+ // and the computation result is valid.
+ if (blockProgressDescriptors[currentBlock] == oldDescriptor)
+ {
+ // Assign the results to the result context (possibly in parallel with other threads
+ // that could and ought to be assigning exactly the same values)
+ SolveL1TransposedCellContext &resultContext = buildResultContextRef(cellContexts, currentBlock, blockCount);
+ resultContext.storePrecalculatedZs(Y);
+
+ // Assign the result assignment progress descriptor
+ cellindexint newDescriptor = MAKE_CELLDESCRIPTOR(currentBlock + 1, CCI__MIN, true);
+ CooperativeAtomics::AtomicCompareExchangeCellindexint(&blockProgressDescriptors[currentBlock], oldDescriptor, newDescriptor); // the result is to be ignored
+
+ // Whether succeeded or not, the result is valid, so go on trying to assign it to the matrix
+ goAssigningTheResult = true;
+ }
+ else
+ {
+ // Otherwise, go on competing for copying the results
+ handleComputationTakenOver = true;
+ }
+ }
+ else
+ {
+ handleComputationTakenOver = true;
+ }
+ }
+ else
+ {
+ // If the final column has not been reached yet, store current values to the context.
+ // Select the other context instance as the previous one might be read by other threads.
+ CellContextInstance nextContextInstance = buildNextContextInstance(previousContextInstance);
+ SolveL1TransposedCellContext &destinationContext = buildBlockContextRef(cellContexts, currentBlock, nextContextInstance);
+ destinationContext.storePrecalculatedZs(Z);
+
+ // Unlock the row until more columns can be used
+ cellindexint oldDescriptor = MAKE_CELLDESCRIPTOR(completedRowBlock, previousContextInstance, true);
+ cellindexint newDescriptor = MAKE_CELLDESCRIPTOR(finalRowBlock, nextContextInstance, false);
+ // The descriptor might have been updated by a competing thread
+ if (!CooperativeAtomics::AtomicCompareExchangeCellindexint(&blockProgressDescriptors[currentBlock], oldDescriptor, newDescriptor))
+ {
+ // Adjust the ptrBElement to point to the result area...
+ ptrBElement = adjustedLastBElement - (sizeint)(currentBlock * block_step) * b_stride;
+ dIASSERT(currentBlock != 0 || adjustedLastBElement == lastBElement);
+ // ...and go on handling the case
+ handleComputationTakenOver = true;
+ }
+ }
+
+ if (handleComputationTakenOver)
+ {
+ cellindexint existingDescriptor = blockProgressDescriptors[currentBlock];
+ // This can only happen if the row was (has become) the uppermost not fully completed one
+ // and the competing thread is at final stage of calculation (i.e., it has reached the currentBlock column).
+ if (existingDescriptor != INVALID_CELLDESCRIPTOR)
+ {
+ // If not fully completed this must be the final stage of the result assignment into the matrix
+ dIASSERT(existingDescriptor == MAKE_CELLDESCRIPTOR(currentBlock + 1, CCI__MIN, true));
+
+ // Go on competing copying the result as anyway the block is the topmost not completed one
+ // and since there was competition for it, there is no other work that can be done right now.
+ const SolveL1TransposedCellContext &resultContext = buildResultContextRef(cellContexts, currentBlock, blockCount);
+ resultContext.loadPrecalculatedZs(Y);
+
+ goAssigningTheResult = true;
+ }
+ else
+ {
+ // everything is over -- just go handling next blocks
+ }
+ }
+ }
+ else if (goForLockedBlockDuplicateCalculation)
+ {
+ blockProcessingState = BPS_SOME_BLOCKS_PROCESSED;
+
+ bool skipToHandlingSubsequentRows = false, skiptoCopyingResult = false;
+
+ /* declare variables */
+ const dReal *ptrLElement;
+
+ if (completedRowBlock < currentBlock)
+ {
+ partialBlock = false;
+
+ ptrLElement = completedRowBlock != 0
+ ? fullyAdjustedLastLElement - currentBlock * block_step - (sizeint)(completedRowBlock * block_step) * rowSkip
+ : columnAdjustedLastLElement - currentBlock * block_step;
+ ptrBElement = completedRowBlock != 0
+ ? adjustedLastBElement - (sizeint)(completedRowBlock * block_step) * b_stride
+ : lastBElement;
+
+ unsigned finalRowBlock = currentBlock/*std::min(currentBlock, completedBlocks)*/;
+ dIASSERT(currentBlock == completedBlocks); // Why would we be competing for a row otherwise?
+
+ bool exitInnerLoop = false;
+ unsigned lastCompletedRow = completedRowBlock;
+ unsigned rowCounter = finalRowBlock - completedRowBlock;
+
+ if (completedRowBlock == 0/* && currentBlock != 0 */&& loopX1RowCount != 0)
+ {
+ if ((loopX1RowCount & 1) != 0)
+ {
+ dReal q1, p4, p3, p2, p1;
+
+ /* load p and q values */
+ q1 = ptrBElement[0 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ ptrBElement -= 1 * b_stride;
+ }
+
+ if ((loopX1RowCount & 2) != 0)
+ {
+ dReal q1, p4, p3, p2, p1;
+
+ /* load p and q values */
+ q1 = ptrBElement[0 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[-1 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ ptrBElement -= 2 * b_stride;
+ }
+ dSASSERT(block_step == 4);
+
+ if (--rowCounter == 0)
+ {
+ exitInnerLoop = true;
+ }
+ }
+
+ for (; !exitInnerLoop; exitInnerLoop = --rowCounter == 0)
+ {
+ dReal q1, p4, p3, p2, p1;
+
+ /* load p and q values */
+ q1 = ptrBElement[0 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[-1 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[-2 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+
+ /* load p and q values */
+ q1 = ptrBElement[-3 * (int)b_stride];
+ p4 = ptrLElement[-3];
+ p3 = ptrLElement[-2];
+ p2 = ptrLElement[-1];
+ p1 = ptrLElement[0];
+ ptrLElement -= rowSkip;
+
+ /* compute outer product and add it to the Z matrix */
+ Z[3] += p4 * q1;
+ Z[2] += p3 * q1;
+ Z[1] += p2 * q1;
+ Z[0] += p1 * q1;
+ dSASSERT(block_step == 4);
+
+ // Check if the primary solver thread has not made any progress
+ cellindexint descriptorVerification = blockProgressDescriptors[currentBlock];
+ unsigned newCompletedRow = GET_CELLDESCRIPTOR_COLUMNINDEX(descriptorVerification);
+
+ if (newCompletedRow != lastCompletedRow)
+ {
+ // Check, this is the first change the current thread detects.
+ // There is absolutely no reason in code for the computation to stop/resume twice
+ // while the current thread is competing.
+ dIASSERT(lastCompletedRow == completedRowBlock);
+
+ if (descriptorVerification == INVALID_CELLDESCRIPTOR)
+ {
+ skipToHandlingSubsequentRows = true;
+ break;
+ }
+
+ if (newCompletedRow == currentBlock + 1)
+ {
+ skiptoCopyingResult = true;
+ break;
+ }
+
+ // Check if the current thread is behind
+ if (newCompletedRow > finalRowBlock - rowCounter)
+ {
+ // If so, go starting over one more time
+ blockProcessingState = BPS_COMPETING_FOR_A_BLOCK;
+ stayWithinTheBlock = true;
+ skipToHandlingSubsequentRows = true;
+ break;
+ }
+
+ // If current thread is ahead, just save new completed column for further comparisons and go on calculating
+ lastCompletedRow = newCompletedRow;
+ }
+
+ /* advance pointers */
+ ptrBElement -= block_step * b_stride;
+ /* end of inner loop */
+ }
+ }
+ else if (completedRowBlock > currentBlock)
+ {
+ dIASSERT(completedRowBlock == currentBlock + 1);
+
+ partialBlock = currentBlock == 0 && loopX1RowCount != 0;
+
+ skiptoCopyingResult = true;
+ }
+ else
+ {
+ dIASSERT(currentBlock == 0); // Execution can get here within the very first block only
+
+ partialBlock = /*currentBlock == 0 && */loopX1RowCount != 0;
+
+ /* just assign the pointers appropriately and go on computing the results */
+ ptrLElement = lastLElement;
+ ptrBElement = lastBElement;
+ }
+
+ if (!skipToHandlingSubsequentRows)
+ {
+ if (!skiptoCopyingResult)
+ {
+ if (partialBlock)
+ {
+ Y[0] = ptrBElement[0 * b_stride]/* - Z[0]*/;
+
+ if (loopX1RowCount >= 2)
+ {
+ dReal p2 = ptrLElement[-1];
+ Y[1] = ptrBElement[-1 * (int)b_stride]/* - Z[1] */- p2 * Y[0];
+
+ if (loopX1RowCount > 2)
+ {
+ dReal p3 = ptrLElement[-2];
+ dReal p3_1 = (ptrLElement - rowSkip)[-2];
+ Y[2] = ptrBElement[-2 * (int)b_stride]/* - Z[2] */- p3 * Y[0] - p3_1 * Y[1];
+ }
+ }
+
+ dSASSERT(block_step == 4);
+ }
+ else
+ {
+ Y[0] = ptrBElement[0 * b_stride] - Z[0];
+
+ dReal p2 = ptrLElement[-1];
+ Y[1] = ptrBElement[-1 * (int)b_stride] - Z[1] - p2 * Y[0];
+
+ dReal p3 = ptrLElement[-2];
+ dReal p3_1 = (ptrLElement - rowSkip)[-2];
+ Y[2] = ptrBElement[-2 * (int)b_stride] - Z[2] - p3 * Y[0] - p3_1 * Y[1];
+
+ dReal p4 = ptrLElement[-3];
+ dReal p4_1 = (ptrLElement - rowSkip)[-3];
+ dReal p4_2 = (ptrLElement - rowSkip * 2)[-3];
+ Y[3] = ptrBElement[-3 * (int)b_stride] - Z[3] - p4 * Y[0] - p4_1 * Y[1] - p4_2 * Y[2];
+
+ dSASSERT(block_step == 4);
+ }
+
+ // Use atomic memory barrier to make sure memory reads of ptrBElement[] and blockProgressDescriptors[] are not swapped
+ CooperativeAtomics::AtomicReadReorderBarrier();
+
+ cellindexint existingDescriptor = blockProgressDescriptors[currentBlock];
+
+ if (existingDescriptor == INVALID_CELLDESCRIPTOR)
+ {
+ // Everything is over -- proceed to subsequent rows
+ skipToHandlingSubsequentRows = true;
+ }
+ else if (existingDescriptor == MAKE_CELLDESCRIPTOR(currentBlock + 1, CCI__MIN, true))
+ {
+ // The values computed above may not be valid. Copy the values already in the result context.
+ skiptoCopyingResult = true;
+ }
+ else
+ {
+ // The descriptor has not been altered yet - this means the ptrBElement[] values used above were not modified yet
+ // and the computation result is valid.
+ cellindexint newDescriptor = MAKE_CELLDESCRIPTOR(currentBlock + 1, CCI__MIN, true); // put the computation at the top so that the evaluation result from the expression above is reused
+
+ // Assign the results to the result context (possibly in parallel with other threads
+ // that could and ought to be assigning exactly the same values)
+ SolveL1TransposedCellContext &resultContext = buildResultContextRef(cellContexts, currentBlock, blockCount);
+ resultContext.storePrecalculatedZs(Y);
+
+ // Assign the result assignment progress descriptor
+ CooperativeAtomics::AtomicCompareExchangeCellindexint(&blockProgressDescriptors[currentBlock], existingDescriptor, newDescriptor); // the result is to be ignored
+
+ // Whether succeeded or not, the result is valid, so go on trying to assign it to the matrix
+ }
+ }
+
+ if (!skipToHandlingSubsequentRows)
+ {
+ if (skiptoCopyingResult)
+ {
+ // Extract the result values stored in the result context
+ const SolveL1TransposedCellContext &resultContext = buildResultContextRef(cellContexts, currentBlock, blockCount);
+ resultContext.loadPrecalculatedZs(Y);
+
+ ptrBElement = currentBlock != 0 ? adjustedLastBElement - (sizeint)(currentBlock * block_step) * b_stride : lastBElement;
+ }
+
+ goAssigningTheResult = true;
+ }
+ }
+ }
+
+ if (goAssigningTheResult)
+ {
+ cellindexint existingDescriptor = blockProgressDescriptors[currentBlock];
+ // Check if the assignment has not been completed yet
+ if (existingDescriptor != INVALID_CELLDESCRIPTOR)
+ {
+ // Assign the computation results to B vector
+ if (partialBlock)
+ {
+ // ptrBElement[0 * b_stride] = Y[0]; -- unchanged
+
+ if (loopX1RowCount >= 2)
+ {
+ ptrBElement[-1 * (int)b_stride] = Y[1];
+
+ if (loopX1RowCount > 2)
+ {
+ ptrBElement[-2 * (int)b_stride] = Y[2];
+ }
+ }
+ dSASSERT(block_step == 4);
+ }
+ else
+ {
+ ptrBElement[0 * b_stride] = Y[0];
+ ptrBElement[-1 * (int)b_stride] = Y[1];
+ ptrBElement[-2 * (int)b_stride] = Y[2];
+ ptrBElement[-3 * (int)b_stride] = Y[3];
+ dSASSERT(block_step == 4);
+ }
+
+ ThrsafeIncrementIntUpToLimit(&refBlockCompletionProgress, currentBlock + 1);
+ dIASSERT(refBlockCompletionProgress >= currentBlock + 1);
+
+ // And assign the completed status no matter what
+ CooperativeAtomics::AtomicStoreCellindexint(&blockProgressDescriptors[currentBlock], INVALID_CELLDESCRIPTOR);
+ }
+ else
+ {
+ // everything is over -- just go handling next blocks
+ }
+ }
+
+ if (!stayWithinTheBlock)
+ {
+ completedBlocks = refBlockCompletionProgress;
+
+ if (completedBlocks == blockCount)
+ {
+ break;
+ }
+
+ currentBlock += 1;
+
+ bool lookaheadBoundaryReached = false;
+
+ if (currentBlock == blockCount || completedBlocks == 0)
+ {
+ lookaheadBoundaryReached = true;
+ }
+ else if (currentBlock >= completedBlocks + lookaheadRange)
+ {
+ lookaheadBoundaryReached = blockProcessingState > BPS_NO_BLOCKS_PROCESSED;
+ }
+ else if (currentBlock < completedBlocks)
+ {
+ // Treat detected row advancement as a row processed
+ // blockProcessingState = BPS_SOME_BLOCKS_PROCESSED; <-- performs better without it
+
+ currentBlock = completedBlocks;
+ }
+
+ if (lookaheadBoundaryReached)
+ {
+ dIASSERT(blockProcessingState != BPS_COMPETING_FOR_A_BLOCK); // Why did not we compete???
+
+ // If no row has been processed in the previous pass, compete for the next row to avoid cycling uselessly
+ if (blockProcessingState <= BPS_NO_BLOCKS_PROCESSED)
+ {
+ // Abandon job if too few blocks remain
+ if (blockCount - completedBlocks <= ownThreadIndex)
+ {
+ break;
+ }
+
+ blockProcessingState = BPS_COMPETING_FOR_A_BLOCK;
+ }
+ else
+ {
+ // If there was some progress, just continue to the next pass
+ blockProcessingState = BPS_NO_BLOCKS_PROCESSED;
+ }
+
+ currentBlock = completedBlocks;
+ }
+ }
+ }
+}
+
+
+#endif
generated by cgit v1.2.1 (git 2.18.0) at 2024-10-31 23:58:34 +0000