Search in sources :

Example 56 with ParameterServerId

use of com.tencent.angel.ps.ParameterServerId in project angel by Tencent.

the class PSAgentMatrixMetaManager method getPartLocation.

/**
 * Get partition location: includes stored pss and the location of the pss
 * TODO: cache
 *
 * @param matrixId partition information
 * @param partId partition id
 * @return partition location
 * @throws ServiceException
 */
public PartitionLocation getPartLocation(int matrixId, int partId) {
    List<ParameterServerId> psIds = getPss(matrixId, partId);
    if (psIds == null) {
        return new PartitionLocation(new ArrayList<>());
    }
    int size = psIds.size();
    List<PSLocation> psLocs = new ArrayList<>(size);
    for (int i = 0; i < size; i++) {
        psLocs.add(new PSLocation(psIds.get(i), PSAgentContext.get().getLocationManager().getPsLocation(psIds.get(i))));
    }
    return new PartitionLocation(psLocs);
}
Also used : PSLocation(com.tencent.angel.ps.server.data.PSLocation) ParameterServerId(com.tencent.angel.ps.ParameterServerId) PartitionLocation(com.tencent.angel.ml.matrix.PartitionLocation)

Example 57 with ParameterServerId

use of com.tencent.angel.ps.ParameterServerId in project angel by Tencent.

the class GetNodeFeatsTest2 method setup.

@Before
public void setup() throws Exception {
    // set basic configuration keys
    Configuration conf = new Configuration();
    conf.setBoolean("mapred.mapper.new-api", true);
    conf.setBoolean(AngelConf.ANGEL_JOB_OUTPUT_PATH_DELETEONEXIST, true);
    conf.set(AngelConf.ANGEL_TASK_USER_TASKCLASS, DummyTask.class.getName());
    // use local deploy mode and dummy dataspliter
    conf.set(AngelConf.ANGEL_DEPLOY_MODE, "LOCAL");
    conf.setBoolean(AngelConf.ANGEL_AM_USE_DUMMY_DATASPLITER, true);
    conf.set(AngelConf.ANGEL_INPUTFORMAT_CLASS, CombineTextInputFormat.class.getName());
    conf.set(AngelConf.ANGEL_SAVE_MODEL_PATH, "file:///F:\\test\\model_1");
    conf.set(AngelConf.ANGEL_TRAIN_DATA_PATH, LOCAL_FS + TMP_PATH + "/in");
    conf.set(AngelConf.ANGEL_LOG_PATH, LOCAL_FS + TMP_PATH + "/log");
    conf.setInt(AngelConf.ANGEL_WORKERGROUP_NUMBER, 1);
    conf.setInt(AngelConf.ANGEL_PS_NUMBER, 1);
    conf.setInt(AngelConf.ANGEL_WORKER_TASK_NUMBER, 1);
    // conf.setInt(AngelConf.ANGEL_MODEL_PARTITIONER_PARTITION_SIZE, 1000);
    conf.setInt(AngelConf.ANGEL_PSAGENT_CACHE_SYNC_TIMEINTERVAL_MS, 10);
    conf.setInt(AngelConf.ANGEL_WORKER_HEARTBEAT_INTERVAL_MS, 1000);
    conf.setInt(AngelConf.ANGEL_PS_HEARTBEAT_INTERVAL_MS, 1000);
    conf.setInt(AngelConf.ANGEL_WORKER_MAX_ATTEMPTS, 1);
    conf.setInt(AngelConf.ANGEL_PS_MAX_ATTEMPTS, 1);
    // get a angel client
    angelClient = AngelClientFactory.get(conf);
    // add sparse float matrix
    MatrixContext siMat = new MatrixContext();
    siMat.setName(NODE);
    siMat.setRowType(RowType.T_ANY_LONGKEY_SPARSE);
    siMat.setRowNum(1);
    siMat.setColNum(10);
    siMat.setMaxColNumInBlock(5);
    siMat.setMaxRowNumInBlock(1);
    // siMat.setValidIndexNum(100);
    // siMat.setColNum(10000000000L);
    siMat.setValueType(Node.class);
    // siMat.setPartitionStorageClass(LongElementMapStorage.class);
    // siMat.setPartitionClass(CSRPartition.class);
    angelClient.addMatrix(siMat);
    // Start PS
    angelClient.startPSServer();
    // Start to run application
    angelClient.run();
    Thread.sleep(5000);
    psId = new ParameterServerId(0);
    psAttempt0Id = new PSAttemptId(psId, 0);
    WorkerGroupId workerGroupId = new WorkerGroupId(0);
    workerId = new WorkerId(workerGroupId, 0);
    workerAttempt0Id = new WorkerAttemptId(workerId, 0);
}
Also used : CombineTextInputFormat(org.apache.hadoop.mapreduce.lib.input.CombineTextInputFormat) MatrixContext(com.tencent.angel.ml.matrix.MatrixContext) Configuration(org.apache.hadoop.conf.Configuration) PSAttemptId(com.tencent.angel.ps.PSAttemptId) WorkerAttemptId(com.tencent.angel.worker.WorkerAttemptId) ParameterServerId(com.tencent.angel.ps.ParameterServerId) WorkerId(com.tencent.angel.worker.WorkerId) WorkerGroupId(com.tencent.angel.worker.WorkerGroupId) Before(org.junit.Before)

Example 58 with ParameterServerId

use of com.tencent.angel.ps.ParameterServerId in project angel by Tencent.

the class IndexGetFuncTest method setup.

@Before
public void setup() throws Exception {
    // set basic configuration keys
    Configuration conf = new Configuration();
    conf.setBoolean("mapred.mapper.new-api", true);
    conf.setBoolean(AngelConf.ANGEL_JOB_OUTPUT_PATH_DELETEONEXIST, true);
    conf.set(AngelConf.ANGEL_TASK_USER_TASKCLASS, DummyTask.class.getName());
    // use local deploy mode and dummy dataspliter
    conf.set(AngelConf.ANGEL_DEPLOY_MODE, "LOCAL");
    conf.setBoolean(AngelConf.ANGEL_AM_USE_DUMMY_DATASPLITER, true);
    conf.set(AngelConf.ANGEL_INPUTFORMAT_CLASS, CombineTextInputFormat.class.getName());
    conf.set(AngelConf.ANGEL_SAVE_MODEL_PATH, LOCAL_FS + TMP_PATH + "/out");
    conf.set(AngelConf.ANGEL_TRAIN_DATA_PATH, LOCAL_FS + TMP_PATH + "/in");
    conf.set(AngelConf.ANGEL_LOG_PATH, LOCAL_FS + TMP_PATH + "/log");
    conf.setInt(AngelConf.ANGEL_WORKERGROUP_NUMBER, 1);
    conf.setInt(AngelConf.ANGEL_PS_NUMBER, 1);
    conf.setInt(AngelConf.ANGEL_WORKER_TASK_NUMBER, 2);
    // get a angel client
    angelClient = AngelClientFactory.get(conf);
    // add dense double matrix
    MatrixContext dMat = new MatrixContext();
    dMat.setName(DENSE_DOUBLE_MAT);
    dMat.setRowNum(1);
    dMat.setColNum(feaNum);
    dMat.setMaxColNumInBlock(feaNum / 3);
    dMat.setRowType(RowType.T_DOUBLE_DENSE);
    angelClient.addMatrix(dMat);
    // add sparse double matrix
    MatrixContext sMat = new MatrixContext();
    sMat.setName(SPARSE_DOUBLE_MAT);
    sMat.setRowNum(1);
    sMat.setColNum(feaNum);
    sMat.setMaxColNumInBlock(feaNum / 3);
    sMat.setRowType(RowType.T_DOUBLE_SPARSE);
    angelClient.addMatrix(sMat);
    // add component sparse double matrix
    MatrixContext sCompMat = new MatrixContext();
    sCompMat.setName(SPARSE_DOUBLE_MAT_COMP);
    sCompMat.setRowNum(1);
    sCompMat.setColNum(feaNum);
    sCompMat.setMaxColNumInBlock(feaNum / 3);
    sCompMat.setRowType(RowType.T_DOUBLE_SPARSE_COMPONENT);
    angelClient.addMatrix(sCompMat);
    // add sparse long-key double matrix
    MatrixContext dLongKeysMatrix = new MatrixContext();
    dLongKeysMatrix.setName(SPARSE_DOUBLE_LONG_MAT);
    dLongKeysMatrix.setRowNum(1);
    dLongKeysMatrix.setColNum(feaNum);
    dLongKeysMatrix.setMaxColNumInBlock(feaNum / 3);
    dLongKeysMatrix.setRowType(RowType.T_DOUBLE_SPARSE_LONGKEY);
    angelClient.addMatrix(dLongKeysMatrix);
    // add component long-key sparse double matrix
    MatrixContext dLongKeysCompMatrix = new MatrixContext();
    dLongKeysCompMatrix.setName(SPARSE_DOUBLE_LONG_MAT_COMP);
    dLongKeysCompMatrix.setRowNum(1);
    dLongKeysCompMatrix.setColNum(feaNum);
    dLongKeysCompMatrix.setMaxColNumInBlock(feaNum / 3);
    dLongKeysCompMatrix.setRowType(RowType.T_DOUBLE_SPARSE_LONGKEY_COMPONENT);
    angelClient.addMatrix(dLongKeysCompMatrix);
    // add dense float matrix
    MatrixContext dfMat = new MatrixContext();
    dfMat.setName(DENSE_FLOAT_MAT);
    dfMat.setRowNum(1);
    dfMat.setColNum(feaNum);
    dfMat.setMaxColNumInBlock(feaNum / 3);
    dfMat.setRowType(RowType.T_FLOAT_DENSE);
    angelClient.addMatrix(dfMat);
    // add sparse float matrix
    MatrixContext sfMat = new MatrixContext();
    sfMat.setName(SPARSE_FLOAT_MAT);
    sfMat.setRowNum(1);
    sfMat.setColNum(feaNum);
    sfMat.setMaxColNumInBlock(feaNum / 3);
    sfMat.setRowType(RowType.T_FLOAT_SPARSE);
    angelClient.addMatrix(sfMat);
    // add component sparse float matrix
    MatrixContext sfCompMat = new MatrixContext();
    sfCompMat.setName(SPARSE_FLOAT_MAT_COMP);
    sfCompMat.setRowNum(1);
    sfCompMat.setColNum(feaNum);
    sfCompMat.setMaxColNumInBlock(feaNum / 3);
    sfCompMat.setRowType(RowType.T_FLOAT_SPARSE_COMPONENT);
    angelClient.addMatrix(sfCompMat);
    // add dense float matrix
    MatrixContext diMat = new MatrixContext();
    diMat.setName(DENSE_INT_MAT);
    diMat.setRowNum(1);
    diMat.setColNum(feaNum);
    diMat.setMaxColNumInBlock(feaNum / 3);
    diMat.setRowType(RowType.T_INT_DENSE);
    angelClient.addMatrix(diMat);
    // add sparse float matrix
    MatrixContext siMat = new MatrixContext();
    siMat.setName(SPARSE_INT_MAT);
    siMat.setRowNum(1);
    siMat.setColNum(feaNum);
    siMat.setMaxColNumInBlock(feaNum / 3);
    siMat.setRowType(RowType.T_INT_SPARSE);
    angelClient.addMatrix(siMat);
    // add component sparse float matrix
    MatrixContext siCompMat = new MatrixContext();
    siCompMat.setName(SPARSE_INT_MAT_COMP);
    siCompMat.setRowNum(1);
    siCompMat.setColNum(feaNum);
    siCompMat.setMaxColNumInBlock(feaNum / 3);
    siCompMat.setRowType(RowType.T_INT_SPARSE_COMPONENT);
    angelClient.addMatrix(siCompMat);
    // Start PS
    angelClient.startPSServer();
    // Start to run application
    angelClient.run();
    Thread.sleep(5000);
    psId = new ParameterServerId(0);
    psAttempt0Id = new PSAttemptId(psId, 0);
    WorkerGroupId workerGroupId = new WorkerGroupId(0);
    workerId = new WorkerId(workerGroupId, 0);
    workerAttempt0Id = new WorkerAttemptId(workerId, 0);
}
Also used : CombineTextInputFormat(org.apache.hadoop.mapreduce.lib.input.CombineTextInputFormat) DummyTask(com.tencent.angel.psagent.DummyTask) MatrixContext(com.tencent.angel.ml.matrix.MatrixContext) Configuration(org.apache.hadoop.conf.Configuration) PSAttemptId(com.tencent.angel.ps.PSAttemptId) WorkerAttemptId(com.tencent.angel.worker.WorkerAttemptId) ParameterServerId(com.tencent.angel.ps.ParameterServerId) WorkerId(com.tencent.angel.worker.WorkerId) WorkerGroupId(com.tencent.angel.worker.WorkerGroupId) Before(org.junit.Before)

Example 59 with ParameterServerId

use of com.tencent.angel.ps.ParameterServerId in project angel by Tencent.

the class ServerPartitionTest method testWriteTo.

@Test
public void testWriteTo() throws Exception {
    // set basic configuration keys
    conf = new Configuration();
    conf.setBoolean("mapred.mapper.new-api", true);
    conf.setBoolean(AngelConf.ANGEL_JOB_OUTPUT_PATH_DELETEONEXIST, true);
    conf.set(AngelConf.ANGEL_TASK_USER_TASKCLASS, DummyTask.class.getName());
    // use local deploy mode and dummy dataspliter
    conf.set(AngelConf.ANGEL_DEPLOY_MODE, "LOCAL");
    conf.setBoolean(AngelConf.ANGEL_AM_USE_DUMMY_DATASPLITER, true);
    conf.set(AngelConf.ANGEL_INPUTFORMAT_CLASS, CombineTextInputFormat.class.getName());
    conf.set(AngelConf.ANGEL_SAVE_MODEL_PATH, LOCAL_FS + TMP_PATH + "/out");
    conf.set(AngelConf.ANGEL_TRAIN_DATA_PATH, LOCAL_FS + TMP_PATH + "/in");
    conf.set(AngelConf.ANGEL_LOG_PATH, LOCAL_FS + TMP_PATH + "/log");
    conf.setInt(AngelConf.ANGEL_WORKERGROUP_NUMBER, 1);
    conf.setInt(AngelConf.ANGEL_PS_NUMBER, 1);
    conf.setInt(AngelConf.ANGEL_WORKER_TASK_NUMBER, 2);
    // get a angel client
    angelClient = AngelClientFactory.get(conf);
    // add matrix
    MatrixContext mMatrix = new MatrixContext();
    mMatrix.setName("w1");
    mMatrix.setRowNum(1);
    mMatrix.setColNum(100000);
    mMatrix.setMaxRowNumInBlock(1);
    mMatrix.setMaxColNumInBlock(50000);
    mMatrix.setRowType(RowType.T_INT_DENSE);
    mMatrix.set(MatrixConf.MATRIX_OPLOG_ENABLEFILTER, "false");
    mMatrix.set(MatrixConf.MATRIX_HOGWILD, "true");
    mMatrix.set(MatrixConf.MATRIX_AVERAGE, "false");
    mMatrix.set(MatrixConf.MATRIX_OPLOG_TYPE, "DENSE_INT");
    angelClient.addMatrix(mMatrix);
    angelClient.startPSServer();
    angelClient.runTask(DummyTask.class);
    Thread.sleep(5000);
    group0Id = new WorkerGroupId(0);
    worker0Id = new WorkerId(group0Id, 0);
    worker0Attempt0Id = new WorkerAttemptId(worker0Id, 0);
    task0Id = new TaskId(0);
    task1Id = new TaskId(1);
    psId = new ParameterServerId(0);
    psAttempt0Id = new PSAttemptId(psId, 0);
    DataOutputStream out = new DataOutputStream(new FileOutputStream("data"));
    ByteBuf buf = Unpooled.buffer(16);
    buf.writeDouble(0.00);
    buf.writeDouble(1.00);
    buf.writeDouble(-1.00);
    buf.writeDouble(-2.00);
    buf.writeDouble(-5.00);
    buf.writeDouble(-6.00);
    buf.writeDouble(-7.00);
    buf.writeDouble(-8.00);
    serverPartition.getRow(6).update(RowType.T_DOUBLE_DENSE, buf, 8);
    serverPartition.save(out);
    out.close();
    DataInputStream in = new DataInputStream(new FileInputStream("data"));
    PartitionKey partitionKeyNew = new PartitionKey(2, 1, 1, 2, 8, 10);
    ServerPartition serverPartitionNew = new ServerPartition(partitionKeyNew, RowType.T_DOUBLE_DENSE);
    serverPartitionNew.init();
    assertNotEquals(((ServerDenseDoubleRow) serverPartition.getRow(6)).getData(), ((ServerDenseDoubleRow) serverPartitionNew.getRow(6)).getData());
    serverPartitionNew.load(in);
    in.close();
    assertEquals(((ServerDenseDoubleRow) serverPartition.getRow(6)).getData(), ((ServerDenseDoubleRow) serverPartitionNew.getRow(6)).getData());
    angelClient.stop();
}
Also used : CombineTextInputFormat(org.apache.hadoop.mapreduce.lib.input.CombineTextInputFormat) DummyTask(com.tencent.angel.master.DummyTask) TaskId(com.tencent.angel.worker.task.TaskId) Configuration(org.apache.hadoop.conf.Configuration) WorkerAttemptId(com.tencent.angel.worker.WorkerAttemptId) WorkerId(com.tencent.angel.worker.WorkerId) ByteBuf(io.netty.buffer.ByteBuf) WorkerGroupId(com.tencent.angel.worker.WorkerGroupId) MatrixContext(com.tencent.angel.ml.matrix.MatrixContext) PSAttemptId(com.tencent.angel.ps.PSAttemptId) PartitionKey(com.tencent.angel.PartitionKey) ParameterServerId(com.tencent.angel.ps.ParameterServerId) Test(org.junit.Test)

Example 60 with ParameterServerId

use of com.tencent.angel.ps.ParameterServerId in project angel by Tencent.

the class MatrixOpLogTest method setup.

@Before
public void setup() throws Exception {
    // set basic configuration keys
    Configuration conf = new Configuration();
    conf.setBoolean("mapred.mapper.new-api", true);
    conf.setBoolean(AngelConf.ANGEL_JOB_OUTPUT_PATH_DELETEONEXIST, true);
    conf.set(AngelConf.ANGEL_TASK_USER_TASKCLASS, DummyTask.class.getName());
    // use local deploy mode and dummy dataspliter
    conf.set(AngelConf.ANGEL_DEPLOY_MODE, "LOCAL");
    conf.setBoolean(AngelConf.ANGEL_AM_USE_DUMMY_DATASPLITER, true);
    conf.set(AngelConf.ANGEL_INPUTFORMAT_CLASS, CombineTextInputFormat.class.getName());
    conf.set(AngelConf.ANGEL_SAVE_MODEL_PATH, LOCAL_FS + TMP_PATH + "/out");
    conf.set(AngelConf.ANGEL_TRAIN_DATA_PATH, LOCAL_FS + TMP_PATH + "/in");
    conf.set(AngelConf.ANGEL_LOG_PATH, LOCAL_FS + TMP_PATH + "/log");
    conf.setInt(AngelConf.ANGEL_WORKERGROUP_NUMBER, 1);
    conf.setInt(AngelConf.ANGEL_PS_NUMBER, 1);
    conf.setInt(AngelConf.ANGEL_WORKER_TASK_NUMBER, 2);
    // get a angel client
    angelClient = AngelClientFactory.get(conf);
    // add matrix
    MatrixContext mMatrix = new MatrixContext();
    mMatrix.setName("w1");
    mMatrix.setRowNum(100);
    mMatrix.setColNum(100000);
    mMatrix.setMaxRowNumInBlock(10);
    mMatrix.setMaxColNumInBlock(50000);
    mMatrix.setRowType(RowType.T_INT_DENSE);
    mMatrix.set(MatrixConf.MATRIX_OPLOG_ENABLEFILTER, "false");
    mMatrix.set(MatrixConf.MATRIX_HOGWILD, "true");
    mMatrix.set(MatrixConf.MATRIX_AVERAGE, "false");
    mMatrix.set(MatrixConf.MATRIX_OPLOG_TYPE, "DENSE_INT");
    angelClient.addMatrix(mMatrix);
    angelClient.startPSServer();
    angelClient.run();
    Thread.sleep(5000);
    psId = new ParameterServerId(0);
    psAttempt0Id = new PSAttemptId(psId, 0);
    WorkerGroupId workerGroupId = new WorkerGroupId(0);
    workerId = new WorkerId(workerGroupId, 0);
    workerAttempt0Id = new WorkerAttemptId(workerId, 0);
}
Also used : CombineTextInputFormat(org.apache.hadoop.mapreduce.lib.input.CombineTextInputFormat) MatrixContext(com.tencent.angel.ml.matrix.MatrixContext) Configuration(org.apache.hadoop.conf.Configuration) PSAttemptId(com.tencent.angel.ps.PSAttemptId) WorkerAttemptId(com.tencent.angel.worker.WorkerAttemptId) ParameterServerId(com.tencent.angel.ps.ParameterServerId) WorkerId(com.tencent.angel.worker.WorkerId) WorkerGroupId(com.tencent.angel.worker.WorkerGroupId) Before(org.junit.Before)

Aggregations

ParameterServerId (com.tencent.angel.ps.ParameterServerId)65 PSAttemptId (com.tencent.angel.ps.PSAttemptId)33 WorkerAttemptId (com.tencent.angel.worker.WorkerAttemptId)28 WorkerGroupId (com.tencent.angel.worker.WorkerGroupId)28 WorkerId (com.tencent.angel.worker.WorkerId)28 Configuration (org.apache.hadoop.conf.Configuration)28 MatrixContext (com.tencent.angel.ml.matrix.MatrixContext)27 CombineTextInputFormat (org.apache.hadoop.mapreduce.lib.input.CombineTextInputFormat)27 Before (org.junit.Before)23 TaskId (com.tencent.angel.worker.task.TaskId)9 PSLocation (com.tencent.angel.ps.server.data.PSLocation)6 HashMap (java.util.HashMap)6 Location (com.tencent.angel.common.location.Location)5 MatrixMeta (com.tencent.angel.ml.matrix.MatrixMeta)5 PartitionLocation (com.tencent.angel.ml.matrix.PartitionLocation)5 ArrayList (java.util.ArrayList)5 ConcurrentHashMap (java.util.concurrent.ConcurrentHashMap)5 Path (org.apache.hadoop.fs.Path)5 Test (org.junit.Test)5 AMParameterServer (com.tencent.angel.master.ps.ps.AMParameterServer)4