[MINOR] Fix consistency task partitioning in mm, mmchain, codegen row Project: http://git-wip-us.apache.org/repos/asf/systemml/repo Commit: http://git-wip-us.apache.org/repos/asf/systemml/commit/d75a669a Tree: http://git-wip-us.apache.org/repos/asf/systemml/tree/d75a669a Diff: http://git-wip-us.apache.org/repos/asf/systemml/diff/d75a669a
Branch: refs/heads/master Commit: d75a669a46381a0a5b54109e7b207613e17ab54e Parents: 06d5bb0 Author: Matthias Boehm <mboe...@gmail.com> Authored: Sun Oct 29 16:06:55 2017 -0700 Committer: Matthias Boehm <mboe...@gmail.com> Committed: Sun Oct 29 16:07:05 2017 -0700 ---------------------------------------------------------------------- .../sysml/runtime/codegen/SpoofRowwise.java | 19 +++---- .../runtime/matrix/data/LibMatrixMult.java | 56 ++++++++++++-------- 2 files changed, 42 insertions(+), 33 deletions(-) ---------------------------------------------------------------------- http://git-wip-us.apache.org/repos/asf/systemml/blob/d75a669a/src/main/java/org/apache/sysml/runtime/codegen/SpoofRowwise.java ---------------------------------------------------------------------- diff --git a/src/main/java/org/apache/sysml/runtime/codegen/SpoofRowwise.java b/src/main/java/org/apache/sysml/runtime/codegen/SpoofRowwise.java index 9d5675b..b0afd88 100644 --- a/src/main/java/org/apache/sysml/runtime/codegen/SpoofRowwise.java +++ b/src/main/java/org/apache/sysml/runtime/codegen/SpoofRowwise.java @@ -39,7 +39,6 @@ import org.apache.sysml.runtime.matrix.data.MatrixBlock; import org.apache.sysml.runtime.matrix.data.SparseBlock; import org.apache.sysml.runtime.matrix.data.SparseRow; import org.apache.sysml.runtime.matrix.data.SparseRowVector; -import org.apache.sysml.runtime.util.UtilFunctions; public abstract class SpoofRowwise extends SpoofOperator @@ -198,11 +197,9 @@ public abstract class SpoofRowwise extends SpoofOperator //core parallel execute ExecutorService pool = Executors.newFixedThreadPool( k ); - int nk = (a instanceof CompressedMatrixBlock) ? k : - UtilFunctions.roundToNext(Math.min(8*k,m/32), k); - int blklen = (int)(Math.ceil((double)m/nk)); - if( a instanceof CompressedMatrixBlock ) - blklen = BitmapEncoder.getAlignedBlocksize(blklen); + ArrayList<Integer> blklens = (a instanceof CompressedMatrixBlock) ? + LibMatrixMult.getAlignedBlockSizes(m, k, BitmapEncoder.BITMAP_BLOCK_SZ) : + LibMatrixMult.getBalancedBlockSizesDefault(m, k, false); try { @@ -210,9 +207,9 @@ public abstract class SpoofRowwise extends SpoofOperator //execute tasks ArrayList<ParColAggTask> tasks = new ArrayList<>(); int outLen = out.getNumRows() * out.getNumColumns(); - for( int i=0; i<nk & i*blklen<m; i++ ) - tasks.add(new ParColAggTask(a, b, scalars, n, n2, outLen, i*blklen, Math.min((i+1)*blklen, m))); - List<Future<double[]>> taskret = pool.invokeAll(tasks); + for( int i=0, lb=0; i<blklens.size(); lb+=blklens.get(i), i++ ) + tasks.add(new ParColAggTask(a, b, scalars, n, n2, outLen, lb, lb+blklens.get(i))); + List<Future<double[]>> taskret = pool.invokeAll(tasks); //aggregate partial results int len = _type.isColumnAgg() ? out.getNumRows()*out.getNumColumns() : 1; for( Future<double[]> task : taskret ) @@ -222,8 +219,8 @@ public abstract class SpoofRowwise extends SpoofOperator else { //execute tasks ArrayList<ParExecTask> tasks = new ArrayList<>(); - for( int i=0; i<nk & i*blklen<m; i++ ) - tasks.add(new ParExecTask(a, b, out, scalars, n, n2, i*blklen, Math.min((i+1)*blklen, m))); + for( int i=0, lb=0; i<blklens.size(); lb+=blklens.get(i), i++ ) + tasks.add(new ParExecTask(a, b, out, scalars, n, n2, lb, lb+blklens.get(i))); List<Future<Long>> taskret = pool.invokeAll(tasks); //aggregate nnz, no need to aggregate results long nnz = 0; http://git-wip-us.apache.org/repos/asf/systemml/blob/d75a669a/src/main/java/org/apache/sysml/runtime/matrix/data/LibMatrixMult.java ---------------------------------------------------------------------- diff --git a/src/main/java/org/apache/sysml/runtime/matrix/data/LibMatrixMult.java b/src/main/java/org/apache/sysml/runtime/matrix/data/LibMatrixMult.java index 684f327..a1f648e 100644 --- a/src/main/java/org/apache/sysml/runtime/matrix/data/LibMatrixMult.java +++ b/src/main/java/org/apache/sysml/runtime/matrix/data/LibMatrixMult.java @@ -209,8 +209,7 @@ public class LibMatrixMult try { ExecutorService pool = Executors.newFixedThreadPool( k ); ArrayList<MatrixMultTask> tasks = new ArrayList<>(); - int nk = (pm2r||pm2c) ? k : UtilFunctions.roundToNext(Math.min(8*k,num/32), k); - ArrayList<Integer> blklens = getBalancedBlockSizes(num, nk); + ArrayList<Integer> blklens = getBalancedBlockSizesDefault(num, k, (pm2r||pm2c)); for( int i=0, lb=0; i<blklens.size(); lb+=blklens.get(i), i++ ) tasks.add(new MatrixMultTask(m1, m2, ret, tm2, pm2r, pm2c, lb, lb+blklens.get(i))); //execute tasks @@ -261,7 +260,7 @@ public class LibMatrixMult } //Timing time = new Timing(true); - + //pre-processing: output allocation ret.sparse = false; ret.allocateDenseBlock(); @@ -312,7 +311,7 @@ public class LibMatrixMult } //Timing time = new Timing(true); - + //pre-processing (no need to check isThreadSafe) ret.sparse = false; ret.allocateDenseBlock(); @@ -321,11 +320,10 @@ public class LibMatrixMult //(currently: always parallelization over number of rows) try { ExecutorService pool = Executors.newFixedThreadPool( k ); + ArrayList<Integer> blklens = getBalancedBlockSizesDefault(mX.rlen, k, true); ArrayList<MatrixMultChainTask> tasks = new ArrayList<>(); - int blklen = (int)(Math.ceil((double)mX.rlen/k)); - blklen += (blklen%24 != 0)?24-blklen%24:0; - for( int i=0; i<k & i*blklen<mX.rlen; i++ ) - tasks.add(new MatrixMultChainTask(mX, mV, mW, ct, i*blklen, Math.min((i+1)*blklen, mX.rlen))); + for( int i=0, lb=0; i<blklens.size(); lb+=blklens.get(i), i++ ) + tasks.add(new MatrixMultChainTask(mX, mV, mW, ct, lb, lb+blklens.get(i))); //execute tasks List<Future<double[]>> taskret = pool.invokeAll(tasks); pool.shutdown(); @@ -1606,10 +1604,18 @@ public class LibMatrixMult final int blocksizeI = 24; // constraint: factor of 4 final int blocksizeJ = 1024; double[] tmp = new double[blocksizeI]; + final int bn = (ru-rl) % blocksizeI; + + //compute rest (not aligned to blocksize) + for( int i=rl, aix=rl*cd; i < rl+bn; i++, aix+=cd ) { + double val = dotProduct(a, b, aix, 0, cd); + val *= (weights) ? w[i] : 1; + val -= (weights2) ? w[i] : 0; + vectMultiplyAdd(val, a, c, aix, 0, cd); + } //blockwise mmchain computation - final int bn = ru - ru % blocksizeI; //rl blocksize aligned - for( int bi=rl; bi < bn; bi+=blocksizeI ) + for( int bi=rl+bn; bi < ru; bi+=blocksizeI ) { //compute 1st matrix-vector for row block Arrays.fill(tmp, 0); @@ -1621,10 +1627,10 @@ public class LibMatrixMult //multiply/subtract weights (in-place), if required if( weights ) - vectMultiply(w, tmp, bi, 0, blocksizeI); + vectMultiply(w, tmp, bi, 0, blocksizeI); else if( weights2 ) vectSubtract(w, tmp, bi, 0, blocksizeI); - + //compute 2nd matrix vector for row block and aggregate for( int bj = 0; bj<cd; bj+=blocksizeJ ) { int bjmin = Math.min(cd-bj, blocksizeJ); @@ -1633,14 +1639,6 @@ public class LibMatrixMult a, c, aix, aix+cd, aix+2*cd, aix+3*cd, bj, bjmin); } } - - //compute rest (not aligned to blocksize) - for( int i=bn, aix=bn*cd; i < ru; i++, aix+=cd ) { - double val = dotProduct(a, b, aix, 0, cd); - val *= (weights) ? w[i] : 1; - val -= (weights2) ? w[i] : 0; - vectMultiplyAdd(val, a, c, aix, 0, cd); - } } private static void matrixMultChainSparse(MatrixBlock mX, MatrixBlock mV, MatrixBlock mW, MatrixBlock ret, ChainType ct, int rl, int ru) @@ -3578,9 +3576,9 @@ public class LibMatrixMult public static boolean checkParColumnAgg(MatrixBlock m1, int k, boolean inclFLOPs) { return (8L * m1.clen * k <= MEM_OVERHEAD_THRESHOLD - && (!inclFLOPs || 4L * m1.rlen * m1.clen >= PAR_MINFLOP_THRESHOLD)); + && (!inclFLOPs || 4L * m1.rlen * m1.clen / (m1.sparse?2:1) >= PAR_MINFLOP_THRESHOLD)); } - + private static boolean checkParMatrixMultRightInputRows( MatrixBlock m1, MatrixBlock m2, int k ) { //parallelize over rows in rhs matrix if number of rows in lhs/output is very small return (m1.rlen==1 && LOW_LEVEL_OPTIMIZATION && m2.clen>1 && !(m1.isUltraSparse()||m2.isUltraSparse())) @@ -3676,6 +3674,20 @@ public class LibMatrixMult } + public static ArrayList<Integer> getBalancedBlockSizesDefault(int len, int k, boolean constK) { + int nk = constK ? k : UtilFunctions.roundToNext(Math.min(8*k,len/32), k); + return getBalancedBlockSizes(len, nk); + } + + public static ArrayList<Integer> getAlignedBlockSizes(int len, int k, int align) { + int blklen = (int)(Math.ceil((double)len/k)); + blklen += ((blklen%align != 0) ? align-blklen%align : 0); + ArrayList<Integer> ret = new ArrayList<>(); + for(int i=0; i<len; i+=blklen) + ret.add(Math.min(blklen, len-i)); + return ret; + } + private static ArrayList<Integer> getBalancedBlockSizes(int len, int k) { ArrayList<Integer> ret = new ArrayList<>(); int base = len / k;
