JkSelf commented on code in PR #8931:
URL: https://github.com/apache/incubator-gluten/pull/8931#discussion_r2906123567
##########
backends-velox/src/main/scala/org/apache/spark/sql/execution/unsafe/UnsafeColumnarBuildSideRelation.scala:
##########
@@ -96,37 +109,128 @@ class UnsafeColumnarBuildSideRelation(
case _ => None
}
+ def isOffload: Boolean = offload
+
/** needed for serialization. */
def this() = {
- this(null, null, null)
+ this(null, null, null, Seq.empty, false)
}
private[unsafe] def getBatches(): Seq[UnsafeByteArray] = {
batches
}
+ private var hashTableData: Long = 0L
+
+ def buildHashTable(broadcastContext: BroadcastHashJoinContext): (Long,
BuildSideRelation) =
+ synchronized {
+ if (hashTableData == 0) {
+ val runtime = Runtimes.contextInstance(
+ BackendsApiManager.getBackendName,
+ "UnsafeColumnarBuildSideRelation#buildHashTable")
+ val jniWrapper = ColumnarBatchSerializerJniWrapper.create(runtime)
+ val serializeHandle: Long = {
+ val allocator = ArrowBufferAllocators.contextInstance()
+ val cSchema = ArrowSchema.allocateNew(allocator)
+ val arrowSchema = SparkArrowUtil.toArrowSchema(
+ SparkShimLoader.getSparkShims.structFromAttributes(output),
+ SQLConf.get.sessionLocalTimeZone)
+ ArrowAbiUtil.exportSchema(allocator, arrowSchema, cSchema)
+ val handle = jniWrapper
+ .init(cSchema.memoryAddress())
+ cSchema.close()
+ handle
+ }
+
+ val batchArray = new ArrayBuffer[Long]
+
+ var batchId = 0
+ while (batchId < batches.size) {
+ val (offset, length) = (batches(batchId).address(),
batches(batchId).size())
+ batchArray.append(jniWrapper.deserializeDirect(serializeHandle,
offset, length.toInt))
+ batchId += 1
+ }
+
+ logDebug(
+ s"BHJ value size: " +
+ s"${broadcastContext.buildHashTableId} = ${batches.size}")
+
+ val (keys, newOutput) = if (newBuildKeys.isEmpty) {
+ (
+ broadcastContext.buildSideJoinKeys.asJava,
+ broadcastContext.buildSideStructure.asJava
+ )
+ } else {
+ (
+ newBuildKeys.asJava,
+ output.asJava
+ )
+ }
+
+ val joinKey = keys.asScala
+ .map {
+ key =>
+ val attr = ConverterUtils.getAttrFromExpr(key)
+ ConverterUtils.genColumnNameWithExprId(attr)
+ }
+ .mkString(",")
+
+ // Build the hash table
+ hashTableData = HashJoinBuilder
+ .nativeBuild(
+ broadcastContext.buildHashTableId,
+ batchArray.toArray,
+ joinKey,
+ broadcastContext.substraitJoinType.ordinal(),
+ broadcastContext.hasMixedFiltCondition,
+ broadcastContext.isExistenceJoin,
+ SubstraitUtil.toNameStruct(newOutput).toByteArray,
+ broadcastContext.isNullAwareAntiJoin,
+ broadcastContext.bloomFilterPushdownSize,
+ broadcastContext.broadcastHashTableBuildThreads
+ )
+
+ jniWrapper.close(serializeHandle)
+ (hashTableData, this)
+ } else {
+ (HashJoinBuilder.cloneHashTable(hashTableData), null)
Review Comment:
Once the hash table is built (i.e., the hashTableData variable is non-zero),
gluten will clone the existing table directly instead of rebuilding it.
##########
backends-velox/src/main/scala/org/apache/spark/rpc/GlutenRpcMessages.scala:
##########
@@ -0,0 +1,53 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.spark.rpc
+
+import java.util
+
+trait GlutenRpcMessage extends Serializable
+
+object GlutenRpcMessages {
+ case class GlutenRegisterExecutor(
+ executorId: String,
+ executorRef: RpcEndpointRef
+ ) extends GlutenRpcMessage
+
+ case class GlutenOnExecutionStart(executionId: String) extends
GlutenRpcMessage
+
+ case class GlutenOnExecutionEnd(executionId: String) extends GlutenRpcMessage
+
+ case class GlutenExecutorRemoved(executorId: String) extends GlutenRpcMessage
+
+ case class GlutenCleanExecutionResource(executionId: String,
broadcastHashIds: util.Set[String])
+ extends GlutenRpcMessage
+
+ // for mergetree cache
+ case class GlutenMergeTreeCacheLoad(
+ mergeTreeTable: String,
+ columns: util.Set[String],
+ onlyMetaCache: Boolean)
+ extends GlutenRpcMessage
+
+ case class GlutenCacheLoadStatus(jobId: String)
+
+ case class CacheJobInfo(status: Boolean, jobId: String, reason: String = "")
+ extends GlutenRpcMessage
+
+ case class GlutenFilesCacheLoad(files: Array[Byte]) extends GlutenRpcMessage
+
+ case class GlutenFilesCacheLoadStatus(jobId: String)
Review Comment:
Will removed.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
