Search in sources :

Example 1 with VectorLimitDesc

use of org.apache.hadoop.hive.ql.plan.VectorLimitDesc in project hive by apache.

the class Vectorizer method validateAndVectorizeOperator.

public Operator<? extends OperatorDesc> validateAndVectorizeOperator(Operator<? extends OperatorDesc> op, VectorizationContext vContext, boolean isReduce, boolean isTezOrSpark, VectorTaskColumnInfo vectorTaskColumnInfo) throws HiveException, VectorizerCannotVectorizeException {
    Operator<? extends OperatorDesc> vectorOp = null;
    // This "global" allows various validation methods to set the "not vectorized" reason.
    currentOperator = op;
    boolean isNative;
    try {
        switch(op.getType()) {
            case MAPJOIN:
                {
                    if (op instanceof MapJoinOperator) {
                        if (!validateMapJoinOperator((MapJoinOperator) op)) {
                            throw new VectorizerCannotVectorizeException();
                        }
                    } else if (op instanceof SMBMapJoinOperator) {
                        if (!validateSMBMapJoinOperator((SMBMapJoinOperator) op)) {
                            throw new VectorizerCannotVectorizeException();
                        }
                    } else {
                        setOperatorNotSupported(op);
                        throw new VectorizerCannotVectorizeException();
                    }
                    if (op instanceof MapJoinOperator) {
                        MapJoinDesc desc = (MapJoinDesc) op.getConf();
                        int joinType = desc.getConds()[0].getType();
                        VectorMapJoinDesc vectorMapJoinDesc = new VectorMapJoinDesc();
                        boolean specialize = canSpecializeMapJoin(op, desc, isTezOrSpark, vContext, vectorMapJoinDesc);
                        if (!specialize) {
                            Class<? extends Operator<?>> opClass = null;
                            // *NON-NATIVE* vector map differences for LEFT OUTER JOIN and Filtered...
                            List<ExprNodeDesc> bigTableFilters = desc.getFilters().get((byte) desc.getPosBigTable());
                            boolean isOuterAndFiltered = (!desc.isNoOuterJoin() && bigTableFilters.size() > 0);
                            if (!isOuterAndFiltered) {
                                opClass = VectorMapJoinOperator.class;
                            } else {
                                if (joinType == JoinDesc.FULL_OUTER_JOIN) {
                                    setOperatorIssue("Vectorized & filtered full-outer joins not supported");
                                    throw new VectorizerCannotVectorizeException();
                                }
                                opClass = VectorMapJoinOuterFilteredOperator.class;
                            }
                            vectorOp = OperatorFactory.getVectorOperator(opClass, op.getCompilationOpContext(), desc, vContext, vectorMapJoinDesc);
                            isNative = false;
                        } else {
                            // TEMPORARY Until Native Vector Map Join with Hybrid passes tests...
                            // HiveConf.setBoolVar(physicalContext.getConf(),
                            // HiveConf.ConfVars.HIVEUSEHYBRIDGRACEHASHJOIN, false);
                            vectorOp = specializeMapJoinOperator(op, vContext, desc, vectorMapJoinDesc);
                            isNative = true;
                            if (vectorTaskColumnInfo != null) {
                                if (usesVectorUDFAdaptor(vectorMapJoinDesc.getAllBigTableKeyExpressions())) {
                                    vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                                }
                                if (usesVectorUDFAdaptor(vectorMapJoinDesc.getAllBigTableValueExpressions())) {
                                    vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                                }
                            }
                        }
                    } else {
                        Preconditions.checkState(op instanceof SMBMapJoinOperator);
                        SMBJoinDesc smbJoinSinkDesc = (SMBJoinDesc) op.getConf();
                        // Check additional constraint.
                        if (smbJoinSinkDesc.getFilterMap() != null) {
                            setOperatorIssue("FilterMaps not supported for Vector Pass-Thru SMB MapJoin");
                            throw new VectorizerCannotVectorizeException();
                        }
                        VectorSMBJoinDesc vectorSMBJoinDesc = new VectorSMBJoinDesc();
                        vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), smbJoinSinkDesc, vContext, vectorSMBJoinDesc);
                        isNative = false;
                    }
                }
                break;
            case REDUCESINK:
                {
                    if (!validateReduceSinkOperator((ReduceSinkOperator) op)) {
                        throw new VectorizerCannotVectorizeException();
                    }
                    ReduceSinkDesc reduceDesc = (ReduceSinkDesc) op.getConf();
                    VectorReduceSinkDesc vectorReduceSinkDesc = new VectorReduceSinkDesc();
                    boolean specialize = canSpecializeReduceSink(reduceDesc, isTezOrSpark, vContext, vectorReduceSinkDesc);
                    if (!specialize) {
                        vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), reduceDesc, vContext, vectorReduceSinkDesc);
                        isNative = false;
                    } else {
                        vectorOp = specializeReduceSinkOperator(op, vContext, reduceDesc, vectorReduceSinkDesc);
                        isNative = true;
                        if (vectorTaskColumnInfo != null) {
                            VectorReduceSinkInfo vectorReduceSinkInfo = vectorReduceSinkDesc.getVectorReduceSinkInfo();
                            if (usesVectorUDFAdaptor(vectorReduceSinkInfo.getReduceSinkKeyExpressions())) {
                                vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                            }
                            if (usesVectorUDFAdaptor(vectorReduceSinkInfo.getReduceSinkValueExpressions())) {
                                vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                            }
                        }
                    }
                }
                break;
            case FILTER:
                {
                    if (!validateFilterOperator((FilterOperator) op)) {
                        throw new VectorizerCannotVectorizeException();
                    }
                    VectorFilterDesc vectorFilterDesc = new VectorFilterDesc();
                    vectorOp = vectorizeFilterOperator(op, vContext, vectorFilterDesc);
                    isNative = true;
                    if (vectorTaskColumnInfo != null) {
                        VectorExpression vectorPredicateExpr = vectorFilterDesc.getPredicateExpression();
                        if (usesVectorUDFAdaptor(vectorPredicateExpr)) {
                            vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                        }
                    }
                }
                break;
            case TOPNKEY:
                {
                    if (!validateTopNKeyOperator((TopNKeyOperator) op)) {
                        throw new VectorizerCannotVectorizeException();
                    }
                    VectorTopNKeyDesc vectorTopNKeyDesc = new VectorTopNKeyDesc();
                    vectorOp = vectorizeTopNKeyOperator(op, vContext, vectorTopNKeyDesc);
                    isNative = true;
                    if (vectorTaskColumnInfo != null) {
                        VectorExpression[] keyExpressions = vectorTopNKeyDesc.getKeyExpressions();
                        if (usesVectorUDFAdaptor(keyExpressions)) {
                            vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                        }
                    }
                }
                break;
            case SELECT:
                {
                    if (!validateSelectOperator((SelectOperator) op)) {
                        throw new VectorizerCannotVectorizeException();
                    }
                    VectorSelectDesc vectorSelectDesc = new VectorSelectDesc();
                    vectorOp = vectorizeSelectOperator(op, vContext, vectorSelectDesc);
                    isNative = true;
                    if (vectorTaskColumnInfo != null) {
                        VectorExpression[] vectorSelectExprs = vectorSelectDesc.getSelectExpressions();
                        if (usesVectorUDFAdaptor(vectorSelectExprs)) {
                            vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                        }
                    }
                }
                break;
            case GROUPBY:
                {
                    // The validateGroupByOperator method will update vectorGroupByDesc.
                    VectorGroupByDesc vectorGroupByDesc = new VectorGroupByDesc();
                    if (!validateGroupByOperator((GroupByOperator) op, isReduce, isTezOrSpark, vectorGroupByDesc)) {
                        throw new VectorizerCannotVectorizeException();
                    }
                    ImmutablePair<Operator<? extends OperatorDesc>, String> pair = doVectorizeGroupByOperator(op, vContext, vectorGroupByDesc);
                    if (pair.left == null) {
                        setOperatorIssue(pair.right);
                        throw new VectorizerCannotVectorizeException();
                    }
                    vectorOp = pair.left;
                    isNative = false;
                    if (vectorTaskColumnInfo != null) {
                        VectorExpression[] vecKeyExpressions = vectorGroupByDesc.getKeyExpressions();
                        if (usesVectorUDFAdaptor(vecKeyExpressions)) {
                            vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                        }
                        VectorAggregationDesc[] vecAggrDescs = vectorGroupByDesc.getVecAggrDescs();
                        for (VectorAggregationDesc vecAggrDesc : vecAggrDescs) {
                            if (usesVectorUDFAdaptor(vecAggrDesc.getInputExpression())) {
                                vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                            }
                        }
                    }
                }
                break;
            case FILESINK:
                {
                    if (!validateFileSinkOperator((FileSinkOperator) op)) {
                        throw new VectorizerCannotVectorizeException();
                    }
                    FileSinkDesc fileSinkDesc = (FileSinkDesc) op.getConf();
                    VectorFileSinkDesc vectorFileSinkDesc = new VectorFileSinkDesc();
                    boolean isArrowSpecialization = checkForArrowFileSink(fileSinkDesc, isTezOrSpark, vContext, vectorFileSinkDesc);
                    if (isArrowSpecialization) {
                        vectorOp = specializeArrowFileSinkOperator(op, vContext, fileSinkDesc, vectorFileSinkDesc);
                        isNative = true;
                    } else {
                        vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), fileSinkDesc, vContext, vectorFileSinkDesc);
                        isNative = false;
                    }
                }
                break;
            case LIMIT:
                {
                    // No validation.
                    LimitDesc limitDesc = (LimitDesc) op.getConf();
                    VectorLimitDesc vectorLimitDesc = new VectorLimitDesc();
                    vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), limitDesc, vContext, vectorLimitDesc);
                    isNative = true;
                }
                break;
            case EVENT:
                {
                    // No validation.
                    AppMasterEventDesc eventDesc = (AppMasterEventDesc) op.getConf();
                    VectorAppMasterEventDesc vectorEventDesc = new VectorAppMasterEventDesc();
                    vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), eventDesc, vContext, vectorEventDesc);
                    isNative = true;
                }
                break;
            case PTF:
                {
                    // The validatePTFOperator method will update vectorPTFDesc.
                    VectorPTFDesc vectorPTFDesc = new VectorPTFDesc();
                    if (!validatePTFOperator((PTFOperator) op, vContext, vectorPTFDesc)) {
                        throw new VectorizerCannotVectorizeException();
                    }
                    vectorOp = vectorizePTFOperator(op, vContext, vectorPTFDesc);
                    isNative = true;
                }
                break;
            case HASHTABLESINK:
                {
                    // No validation.
                    SparkHashTableSinkDesc sparkHashTableSinkDesc = (SparkHashTableSinkDesc) op.getConf();
                    VectorSparkHashTableSinkDesc vectorSparkHashTableSinkDesc = new VectorSparkHashTableSinkDesc();
                    vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), sparkHashTableSinkDesc, vContext, vectorSparkHashTableSinkDesc);
                    isNative = true;
                }
                break;
            case SPARKPRUNINGSINK:
                {
                    // No validation.
                    SparkPartitionPruningSinkDesc sparkPartitionPruningSinkDesc = (SparkPartitionPruningSinkDesc) op.getConf();
                    VectorSparkPartitionPruningSinkDesc vectorSparkPartitionPruningSinkDesc = new VectorSparkPartitionPruningSinkDesc();
                    vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), sparkPartitionPruningSinkDesc, vContext, vectorSparkPartitionPruningSinkDesc);
                    // need to maintain the unique ID so that target map works can
                    // read the output
                    ((SparkPartitionPruningSinkOperator) vectorOp).setUniqueId(((SparkPartitionPruningSinkOperator) op).getUniqueId());
                    isNative = true;
                }
                break;
            default:
                setOperatorNotSupported(op);
                throw new VectorizerCannotVectorizeException();
        }
    } catch (HiveException e) {
        setOperatorIssue(e.getMessage());
        throw new VectorizerCannotVectorizeException();
    }
    Preconditions.checkState(vectorOp != null);
    if (vectorTaskColumnInfo != null && !isNative) {
        vectorTaskColumnInfo.setAllNative(false);
    }
    LOG.debug("vectorizeOperator " + vectorOp.getClass().getName());
    LOG.debug("vectorizeOperator " + vectorOp.getConf().getClass().getName());
    // These operators need to be linked to enable runtime statistics to be gathered/used correctly
    planMapper.link(op, vectorOp);
    return vectorOp;
}
Also used : VectorFileSinkArrowOperator(org.apache.hadoop.hive.ql.exec.vector.filesink.VectorFileSinkArrowOperator) VectorMapJoinAntiJoinMultiKeyOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinAntiJoinMultiKeyOperator) VectorMapJoinInnerBigOnlyMultiKeyOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerBigOnlyMultiKeyOperator) VectorMapJoinLeftSemiMultiKeyOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinLeftSemiMultiKeyOperator) VectorMapJoinOuterFilteredOperator(org.apache.hadoop.hive.ql.exec.vector.VectorMapJoinOuterFilteredOperator) SparkPartitionPruningSinkOperator(org.apache.hadoop.hive.ql.parse.spark.SparkPartitionPruningSinkOperator) VectorizationOperator(org.apache.hadoop.hive.ql.exec.vector.VectorizationOperator) VectorMapJoinInnerMultiKeyOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerMultiKeyOperator) VectorMapJoinFullOuterStringOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinFullOuterStringOperator) VectorMapJoinOperator(org.apache.hadoop.hive.ql.exec.vector.VectorMapJoinOperator) VectorPTFOperator(org.apache.hadoop.hive.ql.exec.vector.ptf.VectorPTFOperator) VectorMapJoinInnerStringOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerStringOperator) VectorMapJoinOuterLongOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinOuterLongOperator) VectorMapJoinLeftSemiStringOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinLeftSemiStringOperator) VectorMapJoinLeftSemiLongOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinLeftSemiLongOperator) VectorMapJoinFullOuterLongOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinFullOuterLongOperator) VectorMapJoinFullOuterMultiKeyOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinFullOuterMultiKeyOperator) VectorMapJoinInnerBigOnlyLongOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerBigOnlyLongOperator) VectorMapJoinAntiJoinStringOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinAntiJoinStringOperator) VectorMapJoinInnerBigOnlyStringOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerBigOnlyStringOperator) VectorMapJoinOuterStringOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinOuterStringOperator) VectorMapJoinInnerLongOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerLongOperator) VectorMapJoinOuterMultiKeyOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinOuterMultiKeyOperator) VectorMapJoinAntiJoinLongOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinAntiJoinLongOperator) VectorAppMasterEventDesc(org.apache.hadoop.hive.ql.plan.VectorAppMasterEventDesc) AppMasterEventDesc(org.apache.hadoop.hive.ql.plan.AppMasterEventDesc) VectorMapJoinDesc(org.apache.hadoop.hive.ql.plan.VectorMapJoinDesc) HiveException(org.apache.hadoop.hive.ql.metadata.HiveException) SMBJoinDesc(org.apache.hadoop.hive.ql.plan.SMBJoinDesc) VectorSMBJoinDesc(org.apache.hadoop.hive.ql.plan.VectorSMBJoinDesc) VectorFileSinkDesc(org.apache.hadoop.hive.ql.plan.VectorFileSinkDesc) FileSinkDesc(org.apache.hadoop.hive.ql.plan.FileSinkDesc) VectorMapJoinOperator(org.apache.hadoop.hive.ql.exec.vector.VectorMapJoinOperator) VectorFileSinkDesc(org.apache.hadoop.hive.ql.plan.VectorFileSinkDesc) VectorReduceSinkInfo(org.apache.hadoop.hive.ql.plan.VectorReduceSinkInfo) VectorSparkPartitionPruningSinkDesc(org.apache.hadoop.hive.ql.plan.VectorSparkPartitionPruningSinkDesc) SparkPartitionPruningSinkDesc(org.apache.hadoop.hive.ql.optimizer.spark.SparkPartitionPruningSinkDesc) VectorAppMasterEventDesc(org.apache.hadoop.hive.ql.plan.VectorAppMasterEventDesc) ArrayList(java.util.ArrayList) List(java.util.List) VectorSelectDesc(org.apache.hadoop.hive.ql.plan.VectorSelectDesc) VectorReduceSinkDesc(org.apache.hadoop.hive.ql.plan.VectorReduceSinkDesc) ReduceSinkDesc(org.apache.hadoop.hive.ql.plan.ReduceSinkDesc) SparkPartitionPruningSinkOperator(org.apache.hadoop.hive.ql.parse.spark.SparkPartitionPruningSinkOperator) VectorMapJoinOperator(org.apache.hadoop.hive.ql.exec.vector.VectorMapJoinOperator) VectorFilterDesc(org.apache.hadoop.hive.ql.plan.VectorFilterDesc) SparkHashTableSinkDesc(org.apache.hadoop.hive.ql.plan.SparkHashTableSinkDesc) VectorSparkHashTableSinkDesc(org.apache.hadoop.hive.ql.plan.VectorSparkHashTableSinkDesc) VectorMapJoinDesc(org.apache.hadoop.hive.ql.plan.VectorMapJoinDesc) MapJoinDesc(org.apache.hadoop.hive.ql.plan.MapJoinDesc) VectorSparkHashTableSinkDesc(org.apache.hadoop.hive.ql.plan.VectorSparkHashTableSinkDesc) VectorSparkPartitionPruningSinkDesc(org.apache.hadoop.hive.ql.plan.VectorSparkPartitionPruningSinkDesc) VectorSMBJoinDesc(org.apache.hadoop.hive.ql.plan.VectorSMBJoinDesc) VectorReduceSinkDesc(org.apache.hadoop.hive.ql.plan.VectorReduceSinkDesc) VectorLimitDesc(org.apache.hadoop.hive.ql.plan.VectorLimitDesc) VectorAggregationDesc(org.apache.hadoop.hive.ql.exec.vector.VectorAggregationDesc) VectorLimitDesc(org.apache.hadoop.hive.ql.plan.VectorLimitDesc) LimitDesc(org.apache.hadoop.hive.ql.plan.LimitDesc) VectorTopNKeyDesc(org.apache.hadoop.hive.ql.plan.VectorTopNKeyDesc) ImmutablePair(org.apache.commons.lang3.tuple.ImmutablePair) VectorMapJoinOuterFilteredOperator(org.apache.hadoop.hive.ql.exec.vector.VectorMapJoinOuterFilteredOperator) VectorGroupByDesc(org.apache.hadoop.hive.ql.plan.VectorGroupByDesc) VectorPTFDesc(org.apache.hadoop.hive.ql.plan.VectorPTFDesc) VectorExpression(org.apache.hadoop.hive.ql.exec.vector.expressions.VectorExpression)

Example 2 with VectorLimitDesc

use of org.apache.hadoop.hive.ql.plan.VectorLimitDesc in project hive by apache.

the class TestVectorLimitOperator method validateVectorLimitOperator.

private void validateVectorLimitOperator(int limit, int batchSize, int expectedBatchSize) throws HiveException {
    FakeVectorRowBatchFromObjectIterables frboi = getBatch(batchSize);
    // Get next batch
    VectorizedRowBatch vrb = frboi.produceNextBatch();
    // Create limit desc with limit value
    LimitDesc ld = new LimitDesc(limit);
    VectorLimitDesc vectorDesc = new VectorLimitDesc();
    VectorLimitOperator lo = new VectorLimitOperator(new CompilationOpContext(), ld, null, vectorDesc);
    // make sure an object registry is present for the test
    ObjectCache.setupObjectRegistry(new ObjectRegistryImpl());
    lo.initialize(new Configuration(), null);
    // Process the batch
    lo.process(vrb, 0);
    // Verify batch size
    Assert.assertEquals(vrb.size, expectedBatchSize);
}
Also used : Configuration(org.apache.hadoop.conf.Configuration) CompilationOpContext(org.apache.hadoop.hive.ql.CompilationOpContext) ObjectRegistryImpl(org.apache.tez.runtime.common.objectregistry.ObjectRegistryImpl) FakeVectorRowBatchFromObjectIterables(org.apache.hadoop.hive.ql.exec.vector.util.FakeVectorRowBatchFromObjectIterables) VectorLimitDesc(org.apache.hadoop.hive.ql.plan.VectorLimitDesc) LimitDesc(org.apache.hadoop.hive.ql.plan.LimitDesc) VectorLimitDesc(org.apache.hadoop.hive.ql.plan.VectorLimitDesc)

Example 3 with VectorLimitDesc

use of org.apache.hadoop.hive.ql.plan.VectorLimitDesc in project hive by apache.

the class Vectorizer method vectorizeOperator.

public Operator<? extends OperatorDesc> vectorizeOperator(Operator<? extends OperatorDesc> op, VectorizationContext vContext, boolean isTezOrSpark, VectorTaskColumnInfo vectorTaskColumnInfo) throws HiveException {
    Operator<? extends OperatorDesc> vectorOp = null;
    boolean isNative;
    switch(op.getType()) {
        case TABLESCAN:
            vectorOp = vectorizeTableScanOperator(op, vContext);
            isNative = true;
            break;
        case MAPJOIN:
            {
                if (op instanceof MapJoinOperator) {
                    VectorMapJoinInfo vectorMapJoinInfo = new VectorMapJoinInfo();
                    MapJoinDesc desc = (MapJoinDesc) op.getConf();
                    boolean specialize = canSpecializeMapJoin(op, desc, isTezOrSpark, vContext, vectorMapJoinInfo);
                    if (!specialize) {
                        Class<? extends Operator<?>> opClass = null;
                        // *NON-NATIVE* vector map differences for LEFT OUTER JOIN and Filtered...
                        List<ExprNodeDesc> bigTableFilters = desc.getFilters().get((byte) desc.getPosBigTable());
                        boolean isOuterAndFiltered = (!desc.isNoOuterJoin() && bigTableFilters.size() > 0);
                        if (!isOuterAndFiltered) {
                            opClass = VectorMapJoinOperator.class;
                        } else {
                            opClass = VectorMapJoinOuterFilteredOperator.class;
                        }
                        vectorOp = OperatorFactory.getVectorOperator(opClass, op.getCompilationOpContext(), op.getConf(), vContext);
                        isNative = false;
                    } else {
                        // TEMPORARY Until Native Vector Map Join with Hybrid passes tests...
                        // HiveConf.setBoolVar(physicalContext.getConf(),
                        //    HiveConf.ConfVars.HIVEUSEHYBRIDGRACEHASHJOIN, false);
                        vectorOp = specializeMapJoinOperator(op, vContext, desc, vectorMapJoinInfo);
                        isNative = true;
                        if (vectorTaskColumnInfo != null) {
                            if (usesVectorUDFAdaptor(vectorMapJoinInfo.getBigTableKeyExpressions())) {
                                vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                            }
                            if (usesVectorUDFAdaptor(vectorMapJoinInfo.getBigTableValueExpressions())) {
                                vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                            }
                        }
                    }
                } else {
                    Preconditions.checkState(op instanceof SMBMapJoinOperator);
                    SMBJoinDesc smbJoinSinkDesc = (SMBJoinDesc) op.getConf();
                    VectorSMBJoinDesc vectorSMBJoinDesc = new VectorSMBJoinDesc();
                    smbJoinSinkDesc.setVectorDesc(vectorSMBJoinDesc);
                    vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), smbJoinSinkDesc, vContext);
                    isNative = false;
                }
            }
            break;
        case REDUCESINK:
            {
                VectorReduceSinkInfo vectorReduceSinkInfo = new VectorReduceSinkInfo();
                ReduceSinkDesc desc = (ReduceSinkDesc) op.getConf();
                boolean specialize = canSpecializeReduceSink(desc, isTezOrSpark, vContext, vectorReduceSinkInfo);
                if (!specialize) {
                    vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), op.getConf(), vContext);
                    isNative = false;
                } else {
                    vectorOp = specializeReduceSinkOperator(op, vContext, desc, vectorReduceSinkInfo);
                    isNative = true;
                    if (vectorTaskColumnInfo != null) {
                        if (usesVectorUDFAdaptor(vectorReduceSinkInfo.getReduceSinkKeyExpressions())) {
                            vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                        }
                        if (usesVectorUDFAdaptor(vectorReduceSinkInfo.getReduceSinkValueExpressions())) {
                            vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                        }
                    }
                }
            }
            break;
        case FILTER:
            {
                vectorOp = vectorizeFilterOperator(op, vContext);
                isNative = true;
                if (vectorTaskColumnInfo != null) {
                    VectorFilterDesc vectorFilterDesc = (VectorFilterDesc) ((AbstractOperatorDesc) vectorOp.getConf()).getVectorDesc();
                    VectorExpression vectorPredicateExpr = vectorFilterDesc.getPredicateExpression();
                    if (usesVectorUDFAdaptor(vectorPredicateExpr)) {
                        vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                    }
                }
            }
            break;
        case SELECT:
            {
                vectorOp = vectorizeSelectOperator(op, vContext);
                isNative = true;
                if (vectorTaskColumnInfo != null) {
                    VectorSelectDesc vectorSelectDesc = (VectorSelectDesc) ((AbstractOperatorDesc) vectorOp.getConf()).getVectorDesc();
                    VectorExpression[] vectorSelectExprs = vectorSelectDesc.getSelectExpressions();
                    if (usesVectorUDFAdaptor(vectorSelectExprs)) {
                        vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                    }
                }
            }
            break;
        case GROUPBY:
            {
                vectorOp = vectorizeGroupByOperator(op, vContext);
                isNative = false;
                if (vectorTaskColumnInfo != null) {
                    VectorGroupByDesc vectorGroupByDesc = (VectorGroupByDesc) ((AbstractOperatorDesc) vectorOp.getConf()).getVectorDesc();
                    if (!vectorGroupByDesc.isVectorOutput()) {
                        vectorTaskColumnInfo.setGroupByVectorOutput(false);
                    }
                    VectorExpression[] vecKeyExpressions = vectorGroupByDesc.getKeyExpressions();
                    if (usesVectorUDFAdaptor(vecKeyExpressions)) {
                        vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                    }
                    VectorAggregateExpression[] vecAggregators = vectorGroupByDesc.getAggregators();
                    for (VectorAggregateExpression vecAggr : vecAggregators) {
                        if (usesVectorUDFAdaptor(vecAggr.inputExpression())) {
                            vectorTaskColumnInfo.setUsesVectorUDFAdaptor(true);
                        }
                    }
                }
            }
            break;
        case FILESINK:
            {
                FileSinkDesc fileSinkDesc = (FileSinkDesc) op.getConf();
                VectorFileSinkDesc vectorFileSinkDesc = new VectorFileSinkDesc();
                fileSinkDesc.setVectorDesc(vectorFileSinkDesc);
                vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), fileSinkDesc, vContext);
                isNative = false;
            }
            break;
        case LIMIT:
            {
                LimitDesc limitDesc = (LimitDesc) op.getConf();
                VectorLimitDesc vectorLimitDesc = new VectorLimitDesc();
                limitDesc.setVectorDesc(vectorLimitDesc);
                vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), limitDesc, vContext);
                isNative = true;
            }
            break;
        case EVENT:
            {
                AppMasterEventDesc eventDesc = (AppMasterEventDesc) op.getConf();
                VectorAppMasterEventDesc vectorEventDesc = new VectorAppMasterEventDesc();
                eventDesc.setVectorDesc(vectorEventDesc);
                vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), eventDesc, vContext);
                isNative = true;
            }
            break;
        case HASHTABLESINK:
            {
                SparkHashTableSinkDesc sparkHashTableSinkDesc = (SparkHashTableSinkDesc) op.getConf();
                VectorSparkHashTableSinkDesc vectorSparkHashTableSinkDesc = new VectorSparkHashTableSinkDesc();
                sparkHashTableSinkDesc.setVectorDesc(vectorSparkHashTableSinkDesc);
                vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), sparkHashTableSinkDesc, vContext);
                isNative = true;
            }
            break;
        case SPARKPRUNINGSINK:
            {
                SparkPartitionPruningSinkDesc sparkPartitionPruningSinkDesc = (SparkPartitionPruningSinkDesc) op.getConf();
                VectorSparkPartitionPruningSinkDesc vectorSparkPartitionPruningSinkDesc = new VectorSparkPartitionPruningSinkDesc();
                sparkPartitionPruningSinkDesc.setVectorDesc(vectorSparkPartitionPruningSinkDesc);
                vectorOp = OperatorFactory.getVectorOperator(op.getCompilationOpContext(), sparkPartitionPruningSinkDesc, vContext);
                isNative = true;
            }
            break;
        default:
            // These are children of GROUP BY operators with non-vector outputs.
            isNative = false;
            vectorOp = op;
            break;
    }
    Preconditions.checkState(vectorOp != null);
    if (vectorTaskColumnInfo != null && !isNative) {
        vectorTaskColumnInfo.setAllNative(false);
    }
    LOG.debug("vectorizeOperator " + vectorOp.getClass().getName());
    LOG.debug("vectorizeOperator " + vectorOp.getConf().getClass().getName());
    if (vectorOp != op) {
        fixupParentChildOperators(op, vectorOp);
        ((AbstractOperatorDesc) vectorOp.getConf()).setVectorMode(true);
    }
    return vectorOp;
}
Also used : VectorMapJoinInnerStringOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerStringOperator) VectorReduceSinkLongOperator(org.apache.hadoop.hive.ql.exec.vector.reducesink.VectorReduceSinkLongOperator) VectorMapJoinOuterLongOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinOuterLongOperator) VectorReduceSinkStringOperator(org.apache.hadoop.hive.ql.exec.vector.reducesink.VectorReduceSinkStringOperator) VectorMapJoinInnerBigOnlyMultiKeyOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerBigOnlyMultiKeyOperator) VectorMapJoinLeftSemiMultiKeyOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinLeftSemiMultiKeyOperator) VectorMapJoinLeftSemiStringOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinLeftSemiStringOperator) VectorMapJoinLeftSemiLongOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinLeftSemiLongOperator) VectorReduceSinkMultiKeyOperator(org.apache.hadoop.hive.ql.exec.vector.reducesink.VectorReduceSinkMultiKeyOperator) VectorMapJoinOuterFilteredOperator(org.apache.hadoop.hive.ql.exec.vector.VectorMapJoinOuterFilteredOperator) VectorMapJoinInnerBigOnlyLongOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerBigOnlyLongOperator) VectorMapJoinInnerBigOnlyStringOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerBigOnlyStringOperator) VectorMapJoinInnerMultiKeyOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerMultiKeyOperator) VectorMapJoinOuterStringOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinOuterStringOperator) VectorMapJoinOperator(org.apache.hadoop.hive.ql.exec.vector.VectorMapJoinOperator) VectorMapJoinInnerLongOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerLongOperator) VectorMapJoinOuterMultiKeyOperator(org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinOuterMultiKeyOperator) AppMasterEventDesc(org.apache.hadoop.hive.ql.plan.AppMasterEventDesc) VectorAppMasterEventDesc(org.apache.hadoop.hive.ql.plan.VectorAppMasterEventDesc) SMBJoinDesc(org.apache.hadoop.hive.ql.plan.SMBJoinDesc) VectorSMBJoinDesc(org.apache.hadoop.hive.ql.plan.VectorSMBJoinDesc) VectorFileSinkDesc(org.apache.hadoop.hive.ql.plan.VectorFileSinkDesc) FileSinkDesc(org.apache.hadoop.hive.ql.plan.FileSinkDesc) VectorMapJoinOperator(org.apache.hadoop.hive.ql.exec.vector.VectorMapJoinOperator) VectorFileSinkDesc(org.apache.hadoop.hive.ql.plan.VectorFileSinkDesc) VectorReduceSinkInfo(org.apache.hadoop.hive.ql.plan.VectorReduceSinkInfo) VectorSparkPartitionPruningSinkDesc(org.apache.hadoop.hive.ql.plan.VectorSparkPartitionPruningSinkDesc) SparkPartitionPruningSinkDesc(org.apache.hadoop.hive.ql.optimizer.spark.SparkPartitionPruningSinkDesc) VectorAppMasterEventDesc(org.apache.hadoop.hive.ql.plan.VectorAppMasterEventDesc) ArrayList(java.util.ArrayList) List(java.util.List) VectorSelectDesc(org.apache.hadoop.hive.ql.plan.VectorSelectDesc) VectorReduceSinkDesc(org.apache.hadoop.hive.ql.plan.VectorReduceSinkDesc) ReduceSinkDesc(org.apache.hadoop.hive.ql.plan.ReduceSinkDesc) VectorMapJoinOperator(org.apache.hadoop.hive.ql.exec.vector.VectorMapJoinOperator) VectorFilterDesc(org.apache.hadoop.hive.ql.plan.VectorFilterDesc) SparkHashTableSinkDesc(org.apache.hadoop.hive.ql.plan.SparkHashTableSinkDesc) VectorSparkHashTableSinkDesc(org.apache.hadoop.hive.ql.plan.VectorSparkHashTableSinkDesc) AbstractOperatorDesc(org.apache.hadoop.hive.ql.plan.AbstractOperatorDesc) MapJoinDesc(org.apache.hadoop.hive.ql.plan.MapJoinDesc) VectorMapJoinDesc(org.apache.hadoop.hive.ql.plan.VectorMapJoinDesc) VectorSparkHashTableSinkDesc(org.apache.hadoop.hive.ql.plan.VectorSparkHashTableSinkDesc) VectorSparkPartitionPruningSinkDesc(org.apache.hadoop.hive.ql.plan.VectorSparkPartitionPruningSinkDesc) VectorMapJoinInfo(org.apache.hadoop.hive.ql.plan.VectorMapJoinInfo) VectorSMBJoinDesc(org.apache.hadoop.hive.ql.plan.VectorSMBJoinDesc) VectorAggregateExpression(org.apache.hadoop.hive.ql.exec.vector.expressions.aggregates.VectorAggregateExpression) VectorLimitDesc(org.apache.hadoop.hive.ql.plan.VectorLimitDesc) LimitDesc(org.apache.hadoop.hive.ql.plan.LimitDesc) VectorLimitDesc(org.apache.hadoop.hive.ql.plan.VectorLimitDesc) VectorMapJoinOuterFilteredOperator(org.apache.hadoop.hive.ql.exec.vector.VectorMapJoinOuterFilteredOperator) VectorGroupByDesc(org.apache.hadoop.hive.ql.plan.VectorGroupByDesc) VectorExpression(org.apache.hadoop.hive.ql.exec.vector.expressions.VectorExpression)

Example 4 with VectorLimitDesc

use of org.apache.hadoop.hive.ql.plan.VectorLimitDesc in project hive by apache.

the class TestVectorLimitOperator method testGlobalLimitReachedInDaemonOrContainer.

private void testGlobalLimitReachedInDaemonOrContainer(boolean isDaemon, int offset, int limit) throws HiveException {
    // from FakeVectorRowBatchFromObjectIterables
    int actualNumberOfElements = 4;
    LlapProxy.setDaemon(isDaemon);
    if (!isDaemon) {
        // init tez object registry
        ObjectCache.setupObjectRegistry(new ObjectRegistryImpl());
    }
    HiveConf conf = new HiveConf();
    HiveConf.setVar(conf, HiveConf.ConfVars.HIVEQUERYID, "query-" + random.nextInt(10000));
    HiveConf.setVar(conf, HiveConf.ConfVars.HIVE_EXECUTION_ENGINE, "tez");
    conf.set(TezProcessor.HIVE_TEZ_VERTEX_NAME, "Map 1");
    VectorLimitOperator lo1 = new VectorLimitOperator(new CompilationOpContext(), new LimitDesc(offset, limit), null, new VectorLimitDesc());
    lo1.initialize(conf, null);
    lo1.initializeOp(conf);
    VectorLimitOperator lo2 = new VectorLimitOperator(new CompilationOpContext(), new LimitDesc(offset, limit), null, new VectorLimitDesc());
    lo2.initialize(conf, null);
    lo2.initializeOp(conf);
    // operator id is important, as it's the base of the limit cache key
    // these operator instances represent the same operator running in different tasks
    Assert.assertEquals("LIM_0", lo1.getOperatorId());
    Assert.assertEquals("LIM_0", lo2.getOperatorId());
    lo1.process(getBatch(500).produceNextBatch(), 0);
    // lo1 is not done, as that's not checked after forwarding, only before next batch
    Assert.assertFalse(lo1.getDone());
    // number of processed rows properly set to global cache and is equal to limit+offset or equal
    // to batch size if limit+offset > batch size (because the operator cannot read through the
    // current batch obviously)
    Assert.assertEquals(Math.min(limit + offset, actualNumberOfElements), lo1.getCurrentCount().get());
    // if lo1 already processed enough rows, lo2 will turn to done without processing any elements
    lo2.process(getBatch(500).produceNextBatch(), 0);
    Assert.assertEquals(limit + offset <= actualNumberOfElements ? true : false, lo2.getDone());
    // lo1 is done now, as limit is check before processing batch
    lo1.process(getBatch(500).produceNextBatch(), 0);
    Assert.assertTrue(lo1.getDone());
}
Also used : CompilationOpContext(org.apache.hadoop.hive.ql.CompilationOpContext) ObjectRegistryImpl(org.apache.tez.runtime.common.objectregistry.ObjectRegistryImpl) HiveConf(org.apache.hadoop.hive.conf.HiveConf) VectorLimitDesc(org.apache.hadoop.hive.ql.plan.VectorLimitDesc) LimitDesc(org.apache.hadoop.hive.ql.plan.LimitDesc) VectorLimitDesc(org.apache.hadoop.hive.ql.plan.VectorLimitDesc)

Aggregations

LimitDesc (org.apache.hadoop.hive.ql.plan.LimitDesc)4 VectorLimitDesc (org.apache.hadoop.hive.ql.plan.VectorLimitDesc)4 ArrayList (java.util.ArrayList)2 List (java.util.List)2 CompilationOpContext (org.apache.hadoop.hive.ql.CompilationOpContext)2 VectorMapJoinOperator (org.apache.hadoop.hive.ql.exec.vector.VectorMapJoinOperator)2 VectorMapJoinOuterFilteredOperator (org.apache.hadoop.hive.ql.exec.vector.VectorMapJoinOuterFilteredOperator)2 VectorExpression (org.apache.hadoop.hive.ql.exec.vector.expressions.VectorExpression)2 VectorMapJoinInnerBigOnlyLongOperator (org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerBigOnlyLongOperator)2 VectorMapJoinInnerBigOnlyMultiKeyOperator (org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerBigOnlyMultiKeyOperator)2 VectorMapJoinInnerBigOnlyStringOperator (org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerBigOnlyStringOperator)2 VectorMapJoinInnerLongOperator (org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerLongOperator)2 VectorMapJoinInnerMultiKeyOperator (org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerMultiKeyOperator)2 VectorMapJoinInnerStringOperator (org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerStringOperator)2 VectorMapJoinLeftSemiLongOperator (org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinLeftSemiLongOperator)2 VectorMapJoinLeftSemiMultiKeyOperator (org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinLeftSemiMultiKeyOperator)2 VectorMapJoinLeftSemiStringOperator (org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinLeftSemiStringOperator)2 VectorMapJoinOuterLongOperator (org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinOuterLongOperator)2 VectorMapJoinOuterMultiKeyOperator (org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinOuterMultiKeyOperator)2 VectorMapJoinOuterStringOperator (org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinOuterStringOperator)2