[2c47b73] | 1 | /** |
---|
| 2 | * @file sz.h |
---|
| 3 | * @author Sheng Di |
---|
| 4 | * @date April, 2015 |
---|
| 5 | * @brief Header file for the whole detector. |
---|
| 6 | * (C) 2015 by Mathematics and Computer Science (MCS), Argonne National Laboratory. |
---|
| 7 | * See COPYRIGHT in top-level directory. |
---|
| 8 | */ |
---|
| 9 | |
---|
| 10 | #ifndef _SZ_H |
---|
| 11 | #define _SZ_H |
---|
| 12 | |
---|
| 13 | #include <stdio.h> |
---|
| 14 | #include <stdint.h> |
---|
| 15 | #include <sys/time.h> /* For gettimeofday(), in microseconds */ |
---|
| 16 | #include <time.h> /* For time(), in seconds */ |
---|
| 17 | #include "iniparser.h" |
---|
| 18 | #include "CompressElement.h" |
---|
| 19 | #include "DynamicByteArray.h" |
---|
| 20 | #include "DynamicIntArray.h" |
---|
| 21 | #include "VarSet.h" |
---|
| 22 | #include "Huffman.h" |
---|
| 23 | #include "TightDataPointStorageD.h" |
---|
| 24 | #include "TightDataPointStorageF.h" |
---|
| 25 | #include "TightDataPointStorageI.h" |
---|
| 26 | #include "conf.h" |
---|
| 27 | #include "dataCompression.h" |
---|
| 28 | #include "ByteToolkit.h" |
---|
| 29 | #include "TypeManager.h" |
---|
| 30 | #include "sz_int8.h" |
---|
| 31 | #include "sz_int16.h" |
---|
| 32 | #include "sz_int32.h" |
---|
| 33 | #include "sz_int64.h" |
---|
| 34 | #include "sz_uint8.h" |
---|
| 35 | #include "sz_uint16.h" |
---|
| 36 | #include "sz_uint32.h" |
---|
| 37 | #include "sz_uint64.h" |
---|
| 38 | #include "sz_float.h" |
---|
| 39 | #include "sz_double.h" |
---|
| 40 | #include "szd_int8.h" |
---|
| 41 | #include "szd_int16.h" |
---|
| 42 | #include "szd_int32.h" |
---|
| 43 | #include "szd_int64.h" |
---|
| 44 | #include "szd_uint8.h" |
---|
| 45 | #include "szd_uint16.h" |
---|
| 46 | #include "szd_uint32.h" |
---|
| 47 | #include "szd_uint64.h" |
---|
| 48 | #include "szd_float.h" |
---|
| 49 | #include "szd_double.h" |
---|
| 50 | #include "sz_float_pwr.h" |
---|
| 51 | #include "sz_double_pwr.h" |
---|
| 52 | #include "callZlib.h" |
---|
| 53 | #include "rw.h" |
---|
| 54 | #include "pastri.h" |
---|
| 55 | #include "sz_float_ts.h" |
---|
| 56 | #include "szd_float_ts.h" |
---|
[9ee2ce3] | 57 | #include "utility.h" |
---|
[2c47b73] | 58 | |
---|
| 59 | #ifdef _WIN32 |
---|
| 60 | #define PATH_SEPARATOR ';' |
---|
| 61 | #else |
---|
| 62 | #define PATH_SEPARATOR ':' |
---|
| 63 | #endif |
---|
| 64 | |
---|
| 65 | #ifdef __cplusplus |
---|
| 66 | extern "C" { |
---|
| 67 | #endif |
---|
| 68 | |
---|
| 69 | //typedef char int8_t; |
---|
| 70 | //typedef unsigned char uint8_t; |
---|
| 71 | //typedef short int16_t; |
---|
| 72 | //typedef unsigned short uint16_t; |
---|
| 73 | //typedef int int32_t; |
---|
| 74 | //typedef unsigned int uint32_t; |
---|
| 75 | //typedef long int64_t; |
---|
| 76 | //typedef unsigned long uint64_t; |
---|
| 77 | |
---|
[9ee2ce3] | 78 | #define SZ_VERNUM 0x0200 |
---|
| 79 | #define SZ_VER_MAJOR 2 |
---|
| 80 | #define SZ_VER_MINOR 0 |
---|
| 81 | #define SZ_VER_BUILD 2 |
---|
| 82 | #define SZ_VER_REVISION 0 |
---|
[2c47b73] | 83 | |
---|
| 84 | #define PASTRI 103 |
---|
| 85 | #define HZ 102 |
---|
| 86 | #define SZ 101 |
---|
| 87 | |
---|
| 88 | //prediction mode of temporal dimension based compression |
---|
| 89 | #define SZ_PREVIOUS_VALUE_ESTIMATE 0 |
---|
| 90 | |
---|
| 91 | #define MIN_NUM_OF_ELEMENTS 20 //if the # elements <= 20, skip the compression |
---|
| 92 | |
---|
| 93 | #define ABS 0 |
---|
| 94 | #define REL 1 |
---|
| 95 | #define ABS_AND_REL 2 |
---|
| 96 | #define ABS_OR_REL 3 |
---|
| 97 | #define PSNR 4 |
---|
| 98 | |
---|
| 99 | #define PW_REL 10 |
---|
| 100 | #define ABS_AND_PW_REL 11 |
---|
| 101 | #define ABS_OR_PW_REL 12 |
---|
| 102 | #define REL_AND_PW_REL 13 |
---|
| 103 | #define REL_OR_PW_REL 14 |
---|
| 104 | |
---|
| 105 | #define SZ_FLOAT 0 |
---|
| 106 | #define SZ_DOUBLE 1 |
---|
| 107 | #define SZ_UINT8 2 |
---|
| 108 | #define SZ_INT8 3 |
---|
| 109 | #define SZ_UINT16 4 |
---|
| 110 | #define SZ_INT16 5 |
---|
| 111 | #define SZ_UINT32 6 |
---|
| 112 | #define SZ_INT32 7 |
---|
| 113 | #define SZ_UINT64 8 |
---|
| 114 | #define SZ_INT64 9 |
---|
| 115 | |
---|
| 116 | #define LITTLE_ENDIAN_DATA 0 //refers to the endian type of the data read from the disk |
---|
| 117 | #define BIG_ENDIAN_DATA 1 //big_endian (ppc, max, etc.) ; little_endian (x86, x64, etc.) |
---|
| 118 | |
---|
| 119 | #define LITTLE_ENDIAN_SYSTEM 0 //refers to the endian type of the system |
---|
| 120 | #define BIG_ENDIAN_SYSTEM 1 |
---|
| 121 | |
---|
| 122 | #define DynArrayInitLen 1024 |
---|
| 123 | |
---|
| 124 | #define MIN_ZLIB_DEC_ALLOMEM_BYTES 1000000 |
---|
| 125 | |
---|
| 126 | //#define maxRangeRadius 32768 |
---|
| 127 | //#define maxRangeRadius 1048576//131072 |
---|
| 128 | |
---|
| 129 | #define SZ_BEST_SPEED 0 |
---|
| 130 | #define SZ_BEST_COMPRESSION 1 |
---|
| 131 | #define SZ_DEFAULT_COMPRESSION 2 |
---|
| 132 | #define SZ_TEMPORAL_COMPRESSION 3 |
---|
| 133 | |
---|
[9ee2ce3] | 134 | #define SZ_NO_REGRESSION 0 |
---|
| 135 | #define SZ_WITH_LINEAR_REGRESSION 1 |
---|
| 136 | |
---|
[2c47b73] | 137 | #define SZ_PWR_MIN_TYPE 0 |
---|
| 138 | #define SZ_PWR_AVG_TYPE 1 |
---|
| 139 | #define SZ_PWR_MAX_TYPE 2 |
---|
| 140 | |
---|
| 141 | //SUCCESS returning status |
---|
| 142 | #define SZ_SCES 0 //successful |
---|
| 143 | #define SZ_NSCS -1 //Not successful |
---|
| 144 | #define SZ_FERR -2 //Failed to open input file |
---|
| 145 | #define SZ_TERR -3 //wrong data type (should be only float or double) |
---|
| 146 | #define SZ_DERR -4 //dimension error |
---|
| 147 | #define SZ_MERR -5 //sz_mode error |
---|
| 148 | #define SZ_BERR -6 //bound-mode error (should be only ABS, REL, ABS_AND_REL, ABS_OR_REL, or PW_REL) |
---|
| 149 | |
---|
| 150 | #define SZ_MAINTAIN_VAR_DATA 0 |
---|
| 151 | #define SZ_DESTROY_WHOLE_VARSET 1 |
---|
| 152 | |
---|
| 153 | #define GROUP_COUNT 16 //2^{16}=65536 |
---|
| 154 | |
---|
| 155 | #define MetaDataByteLength 20 |
---|
| 156 | |
---|
| 157 | #define numOfBufferedSteps 1 //the number of time steps in the buffer |
---|
[9ee2ce3] | 158 | |
---|
| 159 | |
---|
| 160 | #define GZIP_COMPRESSOR 0 //i.e., ZLIB_COMPRSSOR |
---|
| 161 | #define ZSTD_COMPRESSOR 1 |
---|
[2c47b73] | 162 | |
---|
| 163 | //Note: the following setting should be consistent with stateNum in Huffman.h |
---|
| 164 | //#define intvCapacity 65536 |
---|
| 165 | //#define intvRadius 32768 |
---|
| 166 | //#define intvCapacity 131072 |
---|
| 167 | //#define intvRadius 65536 |
---|
| 168 | |
---|
| 169 | #define SZ_COMPUTE_1D_NUMBER_OF_BLOCKS( COUNT, NUM_BLOCKS, BLOCK_SIZE ) \ |
---|
| 170 | if (COUNT <= BLOCK_SIZE){ \ |
---|
| 171 | NUM_BLOCKS = 1; \ |
---|
| 172 | } \ |
---|
| 173 | else{ \ |
---|
| 174 | NUM_BLOCKS = COUNT / BLOCK_SIZE; \ |
---|
| 175 | } \ |
---|
| 176 | |
---|
| 177 | #define SZ_COMPUTE_2D_NUMBER_OF_BLOCKS( COUNT, NUM_BLOCKS, BLOCK_SIZE ) \ |
---|
| 178 | if (COUNT <= BLOCK_SIZE){ \ |
---|
| 179 | NUM_BLOCKS = 1; \ |
---|
| 180 | } \ |
---|
| 181 | else{ \ |
---|
| 182 | NUM_BLOCKS = COUNT / BLOCK_SIZE; \ |
---|
| 183 | } \ |
---|
| 184 | |
---|
| 185 | #define SZ_COMPUTE_3D_NUMBER_OF_BLOCKS( COUNT, NUM_BLOCKS, BLOCK_SIZE ) \ |
---|
| 186 | if (COUNT <= BLOCK_SIZE){ \ |
---|
| 187 | NUM_BLOCKS = 1; \ |
---|
| 188 | } \ |
---|
| 189 | else{ \ |
---|
| 190 | NUM_BLOCKS = COUNT / BLOCK_SIZE; \ |
---|
| 191 | } \ |
---|
| 192 | |
---|
| 193 | #define SZ_COMPUTE_BLOCKCOUNT( COUNT, NUM_BLOCKS, SPLIT_INDEX, \ |
---|
| 194 | EARLY_BLOCK_COUNT, LATE_BLOCK_COUNT ) \ |
---|
| 195 | EARLY_BLOCK_COUNT = LATE_BLOCK_COUNT = COUNT / NUM_BLOCKS; \ |
---|
| 196 | SPLIT_INDEX = COUNT % NUM_BLOCKS; \ |
---|
| 197 | if (0 != SPLIT_INDEX) { \ |
---|
| 198 | EARLY_BLOCK_COUNT = EARLY_BLOCK_COUNT + 1; \ |
---|
| 199 | } \ |
---|
| 200 | |
---|
| 201 | //typedef unsigned long unsigned long; |
---|
| 202 | //typedef unsigned int uint; |
---|
| 203 | |
---|
| 204 | typedef union lint16 |
---|
| 205 | { |
---|
| 206 | unsigned short usvalue; |
---|
| 207 | short svalue; |
---|
| 208 | unsigned char byte[2]; |
---|
| 209 | } lint16; |
---|
| 210 | |
---|
| 211 | typedef union lint32 |
---|
| 212 | { |
---|
| 213 | int ivalue; |
---|
| 214 | unsigned int uivalue; |
---|
| 215 | unsigned char byte[4]; |
---|
| 216 | } lint32; |
---|
| 217 | |
---|
| 218 | typedef union lint64 |
---|
| 219 | { |
---|
| 220 | long lvalue; |
---|
| 221 | unsigned long ulvalue; |
---|
| 222 | unsigned char byte[8]; |
---|
| 223 | } lint64; |
---|
| 224 | |
---|
| 225 | typedef union ldouble |
---|
| 226 | { |
---|
| 227 | double value; |
---|
| 228 | unsigned long lvalue; |
---|
| 229 | unsigned char byte[8]; |
---|
| 230 | } ldouble; |
---|
| 231 | |
---|
| 232 | typedef union lfloat |
---|
| 233 | { |
---|
| 234 | float value; |
---|
| 235 | unsigned int ivalue; |
---|
| 236 | unsigned char byte[4]; |
---|
| 237 | } lfloat; |
---|
| 238 | |
---|
| 239 | /* array meta data and compression parameters for SZ_Init_Params() */ |
---|
| 240 | typedef struct sz_params |
---|
| 241 | { |
---|
| 242 | int dataType; |
---|
| 243 | unsigned int max_quant_intervals; //max number of quantization intervals for quantization |
---|
| 244 | unsigned int quantization_intervals; |
---|
| 245 | unsigned int maxRangeRadius; |
---|
| 246 | int sol_ID;// it's always SZ, unless the setting is PASTRI compression mode (./configure --enable-pastri) |
---|
[9ee2ce3] | 247 | int losslessCompressor; |
---|
[2c47b73] | 248 | int sampleDistance; //2 bytes |
---|
| 249 | float predThreshold; // 2 bytes |
---|
| 250 | int szMode; //* 0 (best speed) or 1 (better compression with Gzip) or 3 temporal-dimension based compression |
---|
| 251 | int gzipMode; //* four options: Z_NO_COMPRESSION, or Z_BEST_SPEED, Z_BEST_COMPRESSION, Z_DEFAULT_COMPRESSION |
---|
| 252 | int errorBoundMode; //4bits (0.5byte), //ABS, REL, ABS_AND_REL, or ABS_OR_REL, PSNR, or PW_REL, PSNR |
---|
| 253 | double absErrBound; //absolute error bound |
---|
| 254 | double relBoundRatio; //value range based relative error bound ratio |
---|
| 255 | double psnr; //PSNR |
---|
| 256 | double pw_relBoundRatio; //point-wise relative error bound |
---|
| 257 | int segment_size; //only used for 2D/3D data compression with pw_relBoundRatio |
---|
| 258 | int pwr_type; //only used for 2D/3D data compression with pw_relBoundRatio |
---|
| 259 | |
---|
| 260 | int snapshotCmprStep; //perform single-snapshot-based compression if time_step == snapshotCmprStep |
---|
| 261 | int predictionMode; |
---|
| 262 | } sz_params; |
---|
| 263 | |
---|
| 264 | typedef struct sz_metadata |
---|
| 265 | { |
---|
| 266 | int versionNumber[3]; //only used for checking the version by calling SZ_GetMetaData() |
---|
| 267 | int isConstant; //only used for checking if the data are constant values by calling SZ_GetMetaData() |
---|
| 268 | int isLossless; //only used for checking if the data compression was lossless, used only by calling SZ_GetMetaData() |
---|
| 269 | int sizeType; //only used for checking whether the size type is "int" or "long" in the compression, used only by calling SZ_GetMetaData() |
---|
| 270 | size_t dataSeriesLength; //# number of data points in the dataset |
---|
| 271 | int defactoNBBins; //real number of quantization bins |
---|
| 272 | struct sz_params* conf_params; //configuration parameters |
---|
| 273 | } sz_metadata; |
---|
| 274 | |
---|
| 275 | typedef struct sz_exedata |
---|
| 276 | { |
---|
| 277 | char optQuantMode; //opt Quantization (0: fixed ; 1: optimized) |
---|
| 278 | int intvCapacity; // the number of intervals for the linear-scaling quantization |
---|
| 279 | int intvRadius; // the number of intervals for the radius of the quantization range (intvRadius=intvCapacity/2) |
---|
| 280 | int SZ_SIZE_TYPE; //the length (# bytes) of the size_t in the system at runtime //4 or 8: sizeof(size_t) |
---|
| 281 | } sz_exedata; |
---|
| 282 | |
---|
| 283 | /*We use a linked list to maintain time-step meta info for time-step based compression*/ |
---|
| 284 | typedef struct sz_tsc_metainfo |
---|
| 285 | { |
---|
| 286 | int totalNumOfSteps; |
---|
| 287 | int currentStep; |
---|
| 288 | char metadata_filename[256]; |
---|
| 289 | FILE *metadata_file; |
---|
[9ee2ce3] | 290 | unsigned char* bit_array; //sihuan added |
---|
| 291 | size_t intersect_size; //sihuan added |
---|
| 292 | int64_t* hist_index; //sihuan added: prestep index |
---|
| 293 | |
---|
[2c47b73] | 294 | } sz_tsc_metadata; |
---|
| 295 | |
---|
| 296 | extern int versionNumber[4]; |
---|
| 297 | |
---|
| 298 | //-------------------key global variables-------------- |
---|
| 299 | extern int dataEndianType; //*endian type of the data read from disk |
---|
| 300 | extern int sysEndianType; //*sysEndianType is actually set automatically. |
---|
| 301 | |
---|
| 302 | extern sz_params *confparams_cpr; |
---|
| 303 | extern sz_params *confparams_dec; |
---|
| 304 | extern sz_exedata *exe_params; |
---|
[9ee2ce3] | 305 | extern int sz_with_regression; |
---|
| 306 | |
---|
[2c47b73] | 307 | //------------------------------------------------ |
---|
| 308 | extern SZ_VarSet* sz_varset; |
---|
| 309 | extern sz_multisteps *multisteps; //compression based on multiple time steps (time-dimension based compression) |
---|
| 310 | extern sz_tsc_metadata *sz_tsc; |
---|
| 311 | |
---|
| 312 | //for pastri |
---|
| 313 | #ifdef PASTRI |
---|
| 314 | extern pastri_params pastri_par; |
---|
| 315 | #endif |
---|
| 316 | |
---|
| 317 | //sz.h |
---|
| 318 | HuffmanTree* SZ_Reset(); |
---|
| 319 | |
---|
| 320 | int SZ_Init(const char *configFilePath); |
---|
| 321 | |
---|
| 322 | int SZ_Init_Params(sz_params *params); |
---|
| 323 | |
---|
| 324 | size_t computeDataLength(size_t r5, size_t r4, size_t r3, size_t r2, size_t r1); |
---|
| 325 | |
---|
| 326 | int computeDimension(size_t r5, size_t r4, size_t r3, size_t r2, size_t r1); |
---|
| 327 | |
---|
| 328 | int SZ_compress_args_float_subblock(unsigned char* compressedBytes, float *oriData, |
---|
| 329 | size_t r5, size_t r4, size_t r3, size_t r2, size_t r1, |
---|
| 330 | size_t s5, size_t s4, size_t s3, size_t s2, size_t s1, |
---|
| 331 | size_t e5, size_t e4, size_t e3, size_t e2, size_t e1, |
---|
| 332 | size_t *outSize, int errBoundMode, double absErr_Bound, double relBoundRatio); |
---|
| 333 | |
---|
| 334 | int SZ_compress_args_double_subblock(unsigned char* compressedBytes, double *oriData, |
---|
| 335 | size_t r5, size_t r4, size_t r3, size_t r2, size_t r1, |
---|
| 336 | size_t s5, size_t s4, size_t s3, size_t s2, size_t s1, |
---|
| 337 | size_t e5, size_t e4, size_t e3, size_t e2, size_t e1, |
---|
| 338 | size_t *outSize, int errBoundMode, double absErr_Bound, double relBoundRatio); |
---|
| 339 | |
---|
| 340 | unsigned char *SZ_compress(int dataType, void *data, size_t *outSize, size_t r5, size_t r4, size_t r3, size_t r2, size_t r1); |
---|
| 341 | |
---|
| 342 | unsigned char* SZ_compress_args(int dataType, void *data, size_t *outSize, int errBoundMode, double absErrBound, |
---|
| 343 | double relBoundRatio, double pwrBoundRatio, size_t r5, size_t r4, size_t r3, size_t r2, size_t r1); |
---|
| 344 | |
---|
| 345 | int SZ_compress_args2(int dataType, void *data, unsigned char* compressed_bytes, size_t *outSize, |
---|
| 346 | int errBoundMode, double absErrBound, double relBoundRatio, double pwrBoundRatio, |
---|
| 347 | size_t r5, size_t r4, size_t r3, size_t r2, size_t r1); |
---|
| 348 | |
---|
| 349 | int SZ_compress_args3(int dataType, void *data, unsigned char* compressed_bytes, size_t *outSize, int errBoundMode, double absErrBound, double relBoundRatio, |
---|
| 350 | size_t r5, size_t r4, size_t r3, size_t r2, size_t r1, |
---|
| 351 | size_t s5, size_t s4, size_t s3, size_t s2, size_t s1, |
---|
| 352 | size_t e5, size_t e4, size_t e3, size_t e2, size_t e1); |
---|
| 353 | |
---|
| 354 | unsigned char *SZ_compress_rev_args(int dataType, void *data, void *reservedValue, size_t *outSize, int errBoundMode, double absErrBound, double relBoundRatio, |
---|
| 355 | size_t r5, size_t r4, size_t r3, size_t r2, size_t r1); |
---|
| 356 | |
---|
| 357 | int SZ_compress_rev_args2(int dataType, void *data, void *reservedValue, unsigned char* compressed_bytes, size_t *outSize, int errBoundMode, double absErrBound, double relBoundRatio, |
---|
| 358 | size_t r5, size_t r4, size_t r3, size_t r2, size_t r1); |
---|
| 359 | unsigned char *SZ_compress_rev(int dataType, void *data, void *reservedValue, size_t *outSize, size_t r5, size_t r4, size_t r3, size_t r2, size_t r1); |
---|
| 360 | |
---|
| 361 | void SZ_Create_ParamsExe(sz_params** conf_params, sz_exedata** exe_params); |
---|
| 362 | |
---|
| 363 | void *SZ_decompress(int dataType, unsigned char *bytes, size_t byteLength, size_t r5, size_t r4, size_t r3, size_t r2, size_t r1); |
---|
| 364 | size_t SZ_decompress_args(int dataType, unsigned char *bytes, size_t byteLength, void* decompressed_array, size_t r5, size_t r4, size_t r3, size_t r2, size_t r1); |
---|
| 365 | |
---|
| 366 | sz_metadata* SZ_getMetadata(unsigned char* bytes); |
---|
| 367 | void SZ_printMetadata(sz_metadata* metadata); |
---|
| 368 | |
---|
| 369 | |
---|
| 370 | void filloutDimArray(size_t* dim, size_t r5, size_t r4, size_t r3, size_t r2, size_t r1); |
---|
| 371 | |
---|
| 372 | size_t compute_total_batch_size(); |
---|
| 373 | |
---|
| 374 | void SZ_registerVar(char* varName, int dataType, void* data, |
---|
| 375 | int errBoundMode, double absErrBound, double relBoundRatio, double pwRelBoundRatio, |
---|
| 376 | size_t r5, size_t r4, size_t r3, size_t r2, size_t r1); |
---|
| 377 | int SZ_deregisterVar(char* varName); |
---|
| 378 | int SZ_deregisterAllVars(); |
---|
| 379 | |
---|
| 380 | int SZ_compress_ts(unsigned char** newByteData, size_t *outSize); |
---|
| 381 | void SZ_decompress_ts(unsigned char *bytes, size_t byteLength); |
---|
| 382 | |
---|
| 383 | void SZ_Finalize(); |
---|
| 384 | |
---|
| 385 | void convertSZParamsToBytes(sz_params* params, unsigned char* result); |
---|
| 386 | sz_params* convertBytesToSZParams(unsigned char* bytes); |
---|
| 387 | |
---|
| 388 | #ifdef __cplusplus |
---|
| 389 | } |
---|
| 390 | #endif |
---|
| 391 | |
---|
| 392 | #endif /* ----- #ifndef _SZ_H ----- */ |
---|