mirror of
https://github.com/HDFGroup/hdf5.git
synced 2024-11-27 02:10:55 +08:00
fc88fcde10
* Merge examples repo into library * Change grepTest to be more fault-tolerant * Update examples macro file * Exclude all Fortran examples from doxygen
142 lines
4.4 KiB
C
142 lines
4.4 KiB
C
/* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
|
|
* Copyright by The HDF Group. *
|
|
* All rights reserved. *
|
|
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
|
|
|
|
/* This example shows different data writing patterns to generate data files
|
|
* that will exhibit substantially different data read speed. The files
|
|
* contain 2-D chunk storage datasets.
|
|
* The 2 writing patterns are:
|
|
* 1. Random--chunks are wriitten in the random order.
|
|
* 2. ByRow--chunks are written by row order.
|
|
*/
|
|
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
|
|
#include "../Perf/h5slab.h"
|
|
#include "hdf5.h"
|
|
|
|
/* Write the chunks in the row order. This provides good and bad read
|
|
* performance if the read pattern is by row and by column respectively.
|
|
*
|
|
* Created by Albert Cheng and Christian Chilan 2010/7/13.
|
|
*/
|
|
int
|
|
createfilebyrow(void)
|
|
{
|
|
hid_t file_id, dset_id, filespace, memspace, fapl, dxpl, dcpl;
|
|
hsize_t dimsf[2], count[2], offset[2], chunk_dims[2] = {CX, CY};
|
|
char *data, dataval, table[RC];
|
|
unsigned long i, j, l, cx;
|
|
fapl = H5Pcreate(H5P_FILE_ACCESS);
|
|
dcpl = H5Pcreate(H5P_DATASET_CREATE);
|
|
dxpl = H5Pcreate(H5P_DATASET_XFER);
|
|
H5Pset_chunk(dcpl, 2, chunk_dims);
|
|
fapl = dxpl = H5P_DEFAULT;
|
|
file_id = H5Fcreate("row_alloc.h5", H5F_ACC_TRUNC, H5P_DEFAULT, fapl);
|
|
dimsf[0] = NX;
|
|
dimsf[1] = NY;
|
|
filespace = H5Screate_simple(2, dimsf, NULL);
|
|
dset_id = H5Dcreate(file_id, "dataset1", H5T_NATIVE_CHAR, filespace, H5P_DEFAULT, dcpl, H5P_DEFAULT);
|
|
count[0] = CX;
|
|
count[1] = NY;
|
|
memspace = H5Screate_simple(2, count, NULL);
|
|
|
|
data = (char *)malloc(count[0] * count[1] * sizeof(char));
|
|
|
|
/* writing the whole chunked rows each time. */
|
|
for (l = 0; l < RC; l++) {
|
|
|
|
offset[0] = l * CX;
|
|
offset[1] = 0;
|
|
|
|
/* fill with values according to row number */
|
|
for (i = 0; i < count[0]; i++)
|
|
for (j = 0; j < count[1]; j++)
|
|
data[i * count[1] + j] = l;
|
|
|
|
H5Sselect_hyperslab(filespace, H5S_SELECT_SET, offset, NULL, count, NULL);
|
|
H5Dwrite(dset_id, H5T_NATIVE_CHAR, memspace, filespace, dxpl, data);
|
|
}
|
|
|
|
free(data);
|
|
H5Dclose(dset_id);
|
|
H5Sclose(filespace);
|
|
H5Sclose(memspace);
|
|
H5Pclose(dxpl);
|
|
H5Pclose(dcpl);
|
|
H5Pclose(fapl);
|
|
H5Fclose(file_id);
|
|
return 0;
|
|
}
|
|
|
|
/* Write the chunks in a random pattern. This provides a read performance
|
|
* worse than when the chunks are written and read in the same order, whether
|
|
* it is by row or by column.
|
|
*
|
|
* Created by Albert Cheng and Christian Chilan 2010/7/13.
|
|
*/
|
|
int
|
|
createfilerandom(void)
|
|
{
|
|
hid_t file_id, dset_id, filespace, memspace, fapl, dxpl, dcpl;
|
|
hsize_t dimsf[2], count[2], offset[2], chunk_dims[2] = {CX, CY};
|
|
char *data, table[RC][CC];
|
|
unsigned long i, j, cx, cy;
|
|
fapl = H5Pcreate(H5P_FILE_ACCESS);
|
|
dcpl = H5Pcreate(H5P_DATASET_CREATE);
|
|
dxpl = H5Pcreate(H5P_DATASET_XFER);
|
|
H5Pset_chunk(dcpl, 2, chunk_dims);
|
|
fapl = dxpl = H5P_DEFAULT;
|
|
file_id = H5Fcreate("random_alloc.h5", H5F_ACC_TRUNC, H5P_DEFAULT, fapl);
|
|
dimsf[0] = NX;
|
|
dimsf[1] = NY;
|
|
filespace = H5Screate_simple(2, dimsf, NULL);
|
|
dset_id = H5Dcreate(file_id, "dataset1", H5T_NATIVE_CHAR, filespace, H5P_DEFAULT, dcpl, H5P_DEFAULT);
|
|
count[0] = CX;
|
|
count[1] = CY;
|
|
memspace = H5Screate_simple(2, count, NULL);
|
|
data = (char *)malloc(count[0] * count[1] * sizeof(char));
|
|
|
|
for (i = 0; i < RC; i++)
|
|
for (j = 0; j < CC; j++)
|
|
table[i][j] = 0;
|
|
|
|
for (i = 0; i < RC * CC; i++) {
|
|
do {
|
|
cx = rand() % RC;
|
|
cy = rand() % CC;
|
|
} while (table[cx][cy]);
|
|
|
|
for (j = 0; j < count[0] * count[1]; j++) {
|
|
data[j] = cx + cy;
|
|
}
|
|
|
|
table[cx][cy] = 1;
|
|
|
|
offset[0] = cx * CX;
|
|
offset[1] = cy * CY;
|
|
|
|
H5Sselect_hyperslab(filespace, H5S_SELECT_SET, offset, NULL, count, NULL);
|
|
H5Dwrite(dset_id, H5T_NATIVE_CHAR, memspace, filespace, dxpl, data);
|
|
}
|
|
|
|
free(data);
|
|
H5Dclose(dset_id);
|
|
H5Sclose(filespace);
|
|
H5Sclose(memspace);
|
|
H5Pclose(dxpl);
|
|
H5Pclose(dcpl);
|
|
H5Pclose(fapl);
|
|
H5Fclose(file_id);
|
|
return 0;
|
|
}
|
|
|
|
int
|
|
main(int argc, char **argv)
|
|
{
|
|
createfilebyrow();
|
|
createfilerandom();
|
|
}
|