diff options
author | Jordan Henderson <jhenderson@hdfgroup.org> | 2017-08-29 16:41:52 (GMT) |
---|---|---|
committer | Jordan Henderson <jhenderson@hdfgroup.org> | 2017-08-29 16:41:52 (GMT) |
commit | b5b1dd082752c7f55898422f9c60fbf5f33d9dd4 (patch) | |
tree | ec6f74fad88838345ba9b761bd8b357d5f531d87 | |
parent | 7f8a8a68780b8f1e67d20021c27849018b60286d (diff) | |
download | hdf5-b5b1dd082752c7f55898422f9c60fbf5f33d9dd4.zip hdf5-b5b1dd082752c7f55898422f9c60fbf5f33d9dd4.tar.gz hdf5-b5b1dd082752c7f55898422f9c60fbf5f33d9dd4.tar.bz2 |
Finish up Parallel Compression test scaling work
-rw-r--r-- | testpar/t_filters_parallel.c | 118 | ||||
-rw-r--r-- | testpar/t_filters_parallel.h | 24 |
2 files changed, 49 insertions, 93 deletions
diff --git a/testpar/t_filters_parallel.c b/testpar/t_filters_parallel.c index 1641a44..d853043 100644 --- a/testpar/t_filters_parallel.c +++ b/testpar/t_filters_parallel.c @@ -863,18 +863,14 @@ test_filtered_dataset_point_selection(void) filespace = H5Dget_space(dset_id); VRFY((filespace >= 0), "File dataspace retrieval succeeded"); - num_points = (hsize_t) (POINT_SELECTION_FILTERED_CHUNKS_NROWS * POINT_SELECTION_FILTERED_CHUNKS_NCOLS / mpi_size); + num_points = (hsize_t) POINT_SELECTION_FILTERED_CHUNKS_NROWS * (hsize_t) POINT_SELECTION_FILTERED_CHUNKS_NCOLS / (hsize_t) mpi_size; coords = (hsize_t *) calloc(1, 2 * num_points * sizeof(*coords)); VRFY((NULL != coords), "Coords calloc succeeded"); for (i = 0; i < num_points; i++) - for (j = 0; j < POINT_SELECTION_FILTERED_CHUNKS_DATASET_DIMS; j++) { - if (j > 0) - coords[(i * POINT_SELECTION_FILTERED_CHUNKS_DATASET_DIMS) + j] = i % (hsize_t) POINT_SELECTION_FILTERED_CHUNKS_NCOLS; - else - coords[(i * POINT_SELECTION_FILTERED_CHUNKS_DATASET_DIMS) + j] = (hsize_t) mpi_rank - + ((i / (hsize_t) POINT_SELECTION_FILTERED_CHUNKS_NCOLS) * (hsize_t) POINT_SELECTION_FILTERED_CHUNKS_CH_NROWS); - } + for (j = 0; j < POINT_SELECTION_FILTERED_CHUNKS_DATASET_DIMS; j++) + coords[(i * POINT_SELECTION_FILTERED_CHUNKS_DATASET_DIMS) + j] = (j > 0) ? (i % (hsize_t) POINT_SELECTION_FILTERED_CHUNKS_NCOLS) + : ((hsize_t) mpi_rank + ((hsize_t) mpi_size * (i / (hsize_t) POINT_SELECTION_FILTERED_CHUNKS_NCOLS))); VRFY((H5Sselect_elements(filespace, H5S_SELECT_SET, (hsize_t) num_points, (const hsize_t *) coords) >= 0), "Point selection succeeded"); @@ -918,24 +914,6 @@ test_filtered_dataset_point_selection(void) VRFY((H5Dread(dset_id, HDF5_DATATYPE_NAME, H5S_ALL, H5S_ALL, plist_id, read_buf) >= 0), "Dataset read succeeded"); - { - if (MAINPROCESS) { - printf("Read buf: ["); - for (i = 0; i < correct_buf_size / sizeof(*read_buf); i++) - printf("%ld, ", read_buf[i]); - printf("]\n"); - fflush(stdout); - - printf("Correct buf: ["); - for (i = 0; i < correct_buf_size / sizeof(*correct_buf); i++) - printf("%ld, ", correct_buf[i]); - printf("]\n"); - fflush(stdout); - } - - MPI_Barrier(MPI_COMM_WORLD); - } - VRFY((0 == memcmp(read_buf, correct_buf, correct_buf_size)), "Data verification succeeded"); if (coords) free(coords); @@ -999,8 +977,8 @@ test_filtered_dataset_interleaved_write(void) dataset_dims[1] = (hsize_t) INTERLEAVED_WRITE_FILTERED_DATASET_NCOLS; chunk_dims[0] = (hsize_t) INTERLEAVED_WRITE_FILTERED_DATASET_CH_NROWS; chunk_dims[1] = (hsize_t) INTERLEAVED_WRITE_FILTERED_DATASET_CH_NCOLS; - sel_dims[0] = (hsize_t) INTERLEAVED_WRITE_FILTERED_DATASET_NROWS / 2; - sel_dims[1] = (hsize_t) INTERLEAVED_WRITE_FILTERED_DATASET_NCOLS / 2; + sel_dims[0] = (hsize_t) (INTERLEAVED_WRITE_FILTERED_DATASET_NROWS / mpi_size); + sel_dims[1] = (hsize_t) INTERLEAVED_WRITE_FILTERED_DATASET_NCOLS; filespace = H5Screate_simple(INTERLEAVED_WRITE_FILTERED_DATASET_DIMS, dataset_dims, NULL); VRFY((filespace >= 0), "File dataspace creation succeeded"); @@ -1027,14 +1005,14 @@ test_filtered_dataset_interleaved_write(void) /* Each process defines the dataset selection in memory and writes * it to the hyperslab in the file */ - count[0] = (hsize_t) SHARED_FILTERED_CHUNKS_NROWS / 2; - count[1] = (hsize_t) SHARED_FILTERED_CHUNKS_NCOLS / 2; - stride[0] = 2; - stride[1] = (hsize_t) SHARED_FILTERED_CHUNKS_CH_NCOLS; + count[0] = (hsize_t) (INTERLEAVED_WRITE_FILTERED_DATASET_NROWS / INTERLEAVED_WRITE_FILTERED_DATASET_CH_NROWS); + count[1] = (hsize_t) (INTERLEAVED_WRITE_FILTERED_DATASET_NCOLS / INTERLEAVED_WRITE_FILTERED_DATASET_CH_NCOLS); + stride[0] = (hsize_t) INTERLEAVED_WRITE_FILTERED_DATASET_CH_NROWS; + stride[1] = (hsize_t) INTERLEAVED_WRITE_FILTERED_DATASET_CH_NCOLS; block[0] = 1; - block[1] = 1; - offset[0] = (hsize_t) mpi_rank / (hsize_t) SHARED_FILTERED_CHUNKS_CH_NCOLS; - offset[1] = (hsize_t) mpi_rank % (hsize_t) SHARED_FILTERED_CHUNKS_CH_NCOLS; + block[1] = (hsize_t) INTERLEAVED_WRITE_FILTERED_DATASET_CH_NCOLS; + offset[0] = (hsize_t) mpi_rank; + offset[1] = 0; if (VERBOSE_MED) printf("Process %d is writing with count[ %llu, %llu ], stride[ %llu, %llu ], offset[ %llu, %llu ], block size[ %llu, %llu ]\n", @@ -1060,9 +1038,14 @@ test_filtered_dataset_interleaved_write(void) data[i] = (C_DATATYPE) GEN_DATA(i); for (i = 0; i < correct_buf_size / sizeof(*correct_buf); i++) - correct_buf[i] = (C_DATATYPE) (((i % (hsize_t) SHARED_FILTERED_CHUNKS_NCOLS) / (hsize_t) SHARED_FILTERED_CHUNKS_CH_NCOLS) - + ((i % (hsize_t) SHARED_FILTERED_CHUNKS_NCOLS) % (hsize_t) SHARED_FILTERED_CHUNKS_CH_NCOLS) - + ((hsize_t) SHARED_FILTERED_CHUNKS_CH_NCOLS * (i / (hsize_t) SHARED_FILTERED_CHUNKS_NCOLS))); + /* Add Column Index */ + correct_buf[i] = (C_DATATYPE) ( (i % (hsize_t) INTERLEAVED_WRITE_FILTERED_DATASET_NCOLS) + + /* Add the Row Index */ + + ((i % (hsize_t) (mpi_size * INTERLEAVED_WRITE_FILTERED_DATASET_NCOLS)) / (hsize_t) INTERLEAVED_WRITE_FILTERED_DATASET_NCOLS) + + /* Add the amount that gets added when a rank moves down to its next section vertically in the dataset */ + + ((hsize_t) INTERLEAVED_WRITE_FILTERED_DATASET_NCOLS * (i / (hsize_t) (mpi_size * INTERLEAVED_WRITE_FILTERED_DATASET_NCOLS)))); /* Create property list for collective dataset write */ plist_id = H5Pcreate(H5P_DATASET_XFER); @@ -1448,8 +1431,8 @@ test_3d_filtered_dataset_overlap(void) chunk_dims[0] = (hsize_t) SHARED_FILTERED_CHUNKS_3D_CH_NROWS; chunk_dims[1] = (hsize_t) SHARED_FILTERED_CHUNKS_3D_CH_NCOLS; chunk_dims[2] = 1; - sel_dims[0] = (hsize_t) SHARED_FILTERED_CHUNKS_3D_NROWS / 2; - sel_dims[1] = (hsize_t) SHARED_FILTERED_CHUNKS_3D_NCOLS / 2; + sel_dims[0] = (hsize_t) (SHARED_FILTERED_CHUNKS_3D_NROWS / mpi_size); + sel_dims[1] = (hsize_t) SHARED_FILTERED_CHUNKS_3D_NCOLS; sel_dims[2] = (hsize_t) SHARED_FILTERED_CHUNKS_3D_DEPTH; filespace = H5Screate_simple(SHARED_FILTERED_CHUNKS_3D_DATASET_DIMS, dataset_dims, NULL); @@ -1477,17 +1460,17 @@ test_3d_filtered_dataset_overlap(void) /* Each process defines the dataset selection in memory and writes * it to the hyperslab in the file */ - count[0] = (hsize_t) SHARED_FILTERED_CHUNKS_3D_NROWS / 2; - count[1] = (hsize_t) SHARED_FILTERED_CHUNKS_3D_NCOLS / 2; + count[0] = (hsize_t) (SHARED_FILTERED_CHUNKS_3D_NROWS / SHARED_FILTERED_CHUNKS_3D_CH_NROWS); + count[1] = (hsize_t) (SHARED_FILTERED_CHUNKS_3D_NCOLS / SHARED_FILTERED_CHUNKS_3D_CH_NCOLS); count[2] = (hsize_t) SHARED_FILTERED_CHUNKS_3D_DEPTH; - stride[0] = 2; + stride[0] = (hsize_t) SHARED_FILTERED_CHUNKS_3D_CH_NROWS; stride[1] = (hsize_t) SHARED_FILTERED_CHUNKS_3D_CH_NCOLS; stride[2] = 1; block[0] = 1; - block[1] = 1; + block[1] = (hsize_t) SHARED_FILTERED_CHUNKS_3D_CH_NCOLS; block[2] = 1; - offset[0] = (hsize_t) mpi_rank / (hsize_t) SHARED_FILTERED_CHUNKS_3D_CH_NCOLS; - offset[1] = (hsize_t) mpi_rank % (hsize_t) SHARED_FILTERED_CHUNKS_3D_CH_NCOLS; + offset[0] = (hsize_t) mpi_rank; + offset[1] = 0; offset[2] = 0; if (VERBOSE_MED) @@ -1513,40 +1496,15 @@ test_3d_filtered_dataset_overlap(void) for (i = 0; i < data_size / sizeof(*data); i++) data[i] = (C_DATATYPE) GEN_DATA(i); - /* This style of data verification is very ugly for cases like this, - * but avoids the problem of needing to have each rank send its data - * plus its selection to a single rank, then have that rank scatter - * all of the data one at a time to a single buffer and finally - * compare that buffer to the read buffer. - JTH - */ for (i = 0; i < correct_buf_size / sizeof(*correct_buf); i++) - /* Start with a base index */ - correct_buf[i] = (C_DATATYPE) ((i % (hsize_t) mpi_size) - - /* Add the value for the increasing column index */ - + ( (hsize_t) mpi_size * - ( - (i % (hsize_t) (SHARED_FILTERED_CHUNKS_3D_NCOLS * SHARED_FILTERED_CHUNKS_3D_DEPTH)) - / (hsize_t) (mpi_size * SHARED_FILTERED_CHUNKS_3D_CH_NCOLS) - ) - ) - - /* Add the value for the increasing row index */ - + ( - (hsize_t) (SHARED_FILTERED_CHUNKS_3D_NCOLS * (SHARED_FILTERED_CHUNKS_3D_CH_NROWS * SHARED_FILTERED_CHUNKS_3D_CH_NCOLS / mpi_size)) - * (i / (hsize_t) (SHARED_FILTERED_CHUNKS_3D_NCOLS * SHARED_FILTERED_CHUNKS_3D_DEPTH * (SHARED_FILTERED_CHUNKS_3D_CH_NROWS * SHARED_FILTERED_CHUNKS_3D_CH_NCOLS / mpi_size))) - ) - - /* Add the value for the increasing row index in a particular chunk */ - + ( (hsize_t) SHARED_FILTERED_CHUNKS_3D_CH_NCOLS * - ( - (i % (hsize_t) (SHARED_FILTERED_CHUNKS_3D_NCOLS * SHARED_FILTERED_CHUNKS_3D_DEPTH * (SHARED_FILTERED_CHUNKS_3D_CH_NROWS * SHARED_FILTERED_CHUNKS_3D_CH_NCOLS / mpi_size))) - / (hsize_t) (SHARED_FILTERED_CHUNKS_3D_NCOLS * SHARED_FILTERED_CHUNKS_3D_DEPTH) - ) - ) - - /* Add the value for the increasing column index in a particular chunk */ - + ((i % (hsize_t) (SHARED_FILTERED_CHUNKS_3D_CH_NCOLS * SHARED_FILTERED_CHUNKS_3D_DEPTH)) / (hsize_t) mpi_size)); + /* Add the Column Index */ + correct_buf[i] = (C_DATATYPE) ( (i % (hsize_t) (SHARED_FILTERED_CHUNKS_3D_DEPTH * SHARED_FILTERED_CHUNKS_3D_NCOLS)) + + /* Add the Row Index */ + + ((i % (hsize_t) (mpi_size * SHARED_FILTERED_CHUNKS_3D_DEPTH * SHARED_FILTERED_CHUNKS_3D_NCOLS)) / (hsize_t) (SHARED_FILTERED_CHUNKS_3D_DEPTH * SHARED_FILTERED_CHUNKS_3D_NCOLS)) + + /* Add the amount that gets added when a rank moves down to its next section vertically in the dataset */ + + ((hsize_t) (SHARED_FILTERED_CHUNKS_3D_DEPTH * SHARED_FILTERED_CHUNKS_3D_NCOLS) * (i / (hsize_t) (mpi_size * SHARED_FILTERED_CHUNKS_3D_DEPTH * SHARED_FILTERED_CHUNKS_3D_NCOLS)))); /* Create property list for collective dataset write */ plist_id = H5Pcreate(H5P_DATASET_XFER); @@ -2507,7 +2465,7 @@ exit: ALARM_OFF; - /* XXX: h5_clean_files(FILENAME, fapl); */ + h5_clean_files(FILENAME, fapl); H5close(); diff --git a/testpar/t_filters_parallel.h b/testpar/t_filters_parallel.h index 6999c27..d0f245d 100644 --- a/testpar/t_filters_parallel.h +++ b/testpar/t_filters_parallel.h @@ -111,19 +111,17 @@ size_t cd_nelmts = FILTER_NUM_CDVALUES; #define POINT_SELECTION_FILTERED_CHUNKS_CH_NCOLS (DIM1_SCALE_FACTOR) #define POINT_SELECTION_FILTERED_CHUNKS_NROWS (POINT_SELECTION_FILTERED_CHUNKS_CH_NROWS * mpi_size) #define POINT_SELECTION_FILTERED_CHUNKS_NCOLS (POINT_SELECTION_FILTERED_CHUNKS_CH_NCOLS * mpi_size) -#define POINT_SELECTION_FILTERED_CHUNKS_NUM_CHUNKS (mpi_size * mpi_size) /* Defines for the filtered dataset interleaved write test */ #define INTERLEAVED_WRITE_FILTERED_DATASET_NAME "interleaved_write_filtered_dataset" #define INTERLEAVED_WRITE_FILTERED_DATASET_DIMS 2 -#define INTERLEAVED_WRITE_FILTERED_DATASET_NROWS (mpi_size * DIM0_SCALE_FACTOR) /* Must be an even multiple of the number of ranks to avoid issues */ -#define INTERLEAVED_WRITE_FILTERED_DATASET_NCOLS (mpi_size * DIM1_SCALE_FACTOR) /* Must be an even multiple of the number of ranks to avoid issues */ -#define INTERLEAVED_WRITE_FILTERED_DATASET_CH_NROWS (INTERLEAVED_WRITE_FILTERED_DATASET_NROWS / mpi_size) -#define INTERLEAVED_WRITE_FILTERED_DATASET_CH_NCOLS (INTERLEAVED_WRITE_FILTERED_DATASET_NCOLS / mpi_size) -#define INTERLEAVED_WRITE_FILTERED_DATASET_SEL_NPOINTS (INTERLEAVED_WRITE_FILTERED_DATASET_CH_NROWS * INTERLEAVED_WRITE_FILTERED_DATASET_CH_NCOLS / INTERLEAVED_WRITE_FILTERED_DATASET_NUM_RANKS) +#define INTERLEAVED_WRITE_FILTERED_DATASET_CH_NROWS (mpi_size) +#define INTERLEAVED_WRITE_FILTERED_DATASET_CH_NCOLS (DIM1_SCALE_FACTOR) +#define INTERLEAVED_WRITE_FILTERED_DATASET_NROWS (INTERLEAVED_WRITE_FILTERED_DATASET_CH_NROWS * DIM0_SCALE_FACTOR) +#define INTERLEAVED_WRITE_FILTERED_DATASET_NCOLS (INTERLEAVED_WRITE_FILTERED_DATASET_CH_NCOLS * DIM1_SCALE_FACTOR) /* Defines for the 3D unshared filtered dataset separate page write test */ -#define UNSHARED_FILTERED_CHUNKS_3D_SEP_PAGE_DATASET_NAME "3d_unshared_filtered_chunks_separate_pages" +#define UNSHARED_FILTERED_CHUNKS_3D_SEP_PAGE_DATASET_NAME "3D_unshared_filtered_chunks_separate_pages" #define UNSHARED_FILTERED_CHUNKS_3D_SEP_PAGE_DATASET_DIMS 3 #define UNSHARED_FILTERED_CHUNKS_3D_SEP_PAGE_NROWS (mpi_size * DIM0_SCALE_FACTOR) #define UNSHARED_FILTERED_CHUNKS_3D_SEP_PAGE_NCOLS (mpi_size * DIM1_SCALE_FACTOR) @@ -132,7 +130,7 @@ size_t cd_nelmts = FILTER_NUM_CDVALUES; #define UNSHARED_FILTERED_CHUNKS_3D_SEP_PAGE_CH_NCOLS (UNSHARED_FILTERED_CHUNKS_3D_SEP_PAGE_NCOLS / mpi_size) /* Defines for the 3D unshared filtered dataset same page write test */ -#define UNSHARED_FILTERED_CHUNKS_3D_SAME_PAGE_DATASET_NAME "3d_unshared_filtered_chunks_same_pages" +#define UNSHARED_FILTERED_CHUNKS_3D_SAME_PAGE_DATASET_NAME "3D_unshared_filtered_chunks_same_pages" #define UNSHARED_FILTERED_CHUNKS_3D_SAME_PAGE_DATASET_DIMS 3 #define UNSHARED_FILTERED_CHUNKS_3D_SAME_PAGE_NROWS (mpi_size * DIM0_SCALE_FACTOR) #define UNSHARED_FILTERED_CHUNKS_3D_SAME_PAGE_NCOLS (mpi_size * DIM1_SCALE_FACTOR) @@ -141,13 +139,13 @@ size_t cd_nelmts = FILTER_NUM_CDVALUES; #define UNSHARED_FILTERED_CHUNKS_3D_SAME_PAGE_CH_NCOLS (UNSHARED_FILTERED_CHUNKS_3D_SAME_PAGE_NCOLS / mpi_size) /* Defines for the 3d shared filtered dataset write test */ -#define SHARED_FILTERED_CHUNKS_3D_DATASET_NAME "3d_shared_filtered_chunks" +#define SHARED_FILTERED_CHUNKS_3D_DATASET_NAME "3D_shared_filtered_chunks" #define SHARED_FILTERED_CHUNKS_3D_DATASET_DIMS 3 -#define SHARED_FILTERED_CHUNKS_3D_NROWS (mpi_size * DIM0_SCALE_FACTOR) -#define SHARED_FILTERED_CHUNKS_3D_NCOLS (mpi_size * DIM1_SCALE_FACTOR) +#define SHARED_FILTERED_CHUNKS_3D_CH_NROWS (mpi_size) +#define SHARED_FILTERED_CHUNKS_3D_CH_NCOLS (DIM1_SCALE_FACTOR) +#define SHARED_FILTERED_CHUNKS_3D_NROWS (SHARED_FILTERED_CHUNKS_3D_CH_NROWS * DIM0_SCALE_FACTOR) +#define SHARED_FILTERED_CHUNKS_3D_NCOLS (SHARED_FILTERED_CHUNKS_3D_CH_NCOLS * DIM1_SCALE_FACTOR) #define SHARED_FILTERED_CHUNKS_3D_DEPTH (mpi_size) -#define SHARED_FILTERED_CHUNKS_3D_CH_NROWS (SHARED_FILTERED_CHUNKS_3D_NROWS / mpi_size) -#define SHARED_FILTERED_CHUNKS_3D_CH_NCOLS (SHARED_FILTERED_CHUNKS_3D_NCOLS / mpi_size) /* Struct type for the compound datatype filtered dataset tests */ typedef struct { |