greg/stasis-aries-wal
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

roseTable csv parser. updated rose.cpp for benchmarking

Browse source
This commit is contained in:
Sears Russell 2007-11-13 23:40:21 +00:00
parent 55e96e7420
commit 58e0466339
3 changed files with 290 additions and 57 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
benchmarks/rose.cpp
View file

@ -26,7 +26,7 @@ typedef int32_t val_t; // XXX needed by lsmWorkers..
decompressing the data actually matches the original dataset.
*/
#define CHECK_OUTPUT
//#define CHECK_OUTPUT
using rose::Pstar;
using rose::Multicolumn;

237
benchmarks/roseTable.cpp
View file

@ -59,12 +59,20 @@ namespace rose {
typename PAGELAYOUT::FMT::TUP s;
long INSERTS;
int file_mode = 0;
char * file = 0;
if(argc == 2) {
INSERTS = atoll(argv[1]);
} else if (argc == 3) {
file_mode = 1;
assert(!strcmp("-f", argv[1]));
file = argv[2];
} else {
INSERTS = 10 * 1000 * 1000;
}
static const long COUNT = INSERTS / 100;
int column[] = { 0 , 1, 2, 3, 4, 5, 6, 7, 8, 9 };
static long COUNT = INSERTS / 100;
long int count = COUNT;
struct timeval start_tv, now_tv;
@ -74,8 +82,232 @@ namespace rose {
start = rose::tv_to_double(start_tv);
last_start = start;
printf("tuple 'size'%d ; %ld\n", PAGELAYOUT::FMT::TUP::sizeofBytes(), sizeof(typename PAGELAYOUT::FMT::TUP));
if(file_mode) {
typename PAGELAYOUT::FMT::TUP scratch;
int max_col_number = 0;
for(int col =0; col < PAGELAYOUT::FMT::TUP::NN ; col++) {
max_col_number = max_col_number < column[col]
? column[col] : max_col_number;
}
char ** toks = (char**)malloc(sizeof(char*)*(max_col_number+1));
printf("Reading from file %s\n", file);
int inserts = 0;
size_t line_len = 100;
// getline wants malloced memmory (it probably calls realloc...)
char * line = (char*) malloc(sizeof(char) * line_len);
FILE * input = fopen(file, "r");
if(!input) {
perror("Couldn't open input");
return 1;
}
size_t read_len;
COUNT = 1000000;
count = 1000000;
while(-1 != (read_len = getline(&line, &line_len, input))) {
int line_tok_count;
{
char * saveptr;
int i;
toks[0] = strtok_r(line, ",\n", &saveptr);
for(i = 1; i < (max_col_number+1); i++) {
toks[i] = strtok_r(0, ",\n", &saveptr);
if(!toks[i]) {
break;
}
}
line_tok_count = i;
}
if(line_tok_count < (max_col_number+1)) {
// printf("!");
if(-1 == getline(&line,&line_len,input)) {
// hit eof.
} else {
printf("Not enough tokesn on line %d (found: %d expected: %d\n",
inserts+1, line_tok_count, max_col_number+1);
return 1;
}
} else {
// printf(".");
inserts ++;
if(0 < PAGELAYOUT::FMT::TUP::NN) {
char * endptr;
errno = 0;
typename PAGELAYOUT::FMT::TUP::TYP0 t = strtoll(toks[column[0]], &endptr, 0);
if(strlen(toks[column[0]]) - (size_t)(endptr-toks[column[0]]) > 1) {
printf("couldnt parse token #%d: ->%s<-\n", column[0], toks[column[0]]);
return 1;
}
if(errno) {
printf("Couldn't parse token #%d: ->%s<-", column[0],toks[column[0]]);
perror("strtoll error is");
return 1;
}
scratch.set0(&t);
}
if(1 < PAGELAYOUT::FMT::TUP::NN) {
char * endptr;
errno = 0;
typename PAGELAYOUT::FMT::TUP::TYP1 t = strtoll(toks[column[1]], &endptr, 0);
if(strlen(toks[column[1]]) - (size_t)(endptr-toks[column[1]]) > 1) {
printf("couldnt parse token #%d: ->%s<-\n", column[1], toks[column[1]]);
return 1;
}
if(errno) {
printf("Couldn't parse token #%d: ->%s<-", column[1],toks[column[1]]);
perror("strtoll error is");
return 1;
}
scratch.set1(&t);
}
if(2 < PAGELAYOUT::FMT::TUP::NN) {
char * endptr;
errno = 0;
typename PAGELAYOUT::FMT::TUP::TYP2 t = strtoll(toks[column[2]], &endptr, 0);
if(strlen(toks[column[2]]) - (size_t)(endptr-toks[column[2]]) > 1) {
printf("couldnt parse token #%d: ->%s<-\n", column[2], toks[column[2]]);
return 1;
}
if(errno) {
printf("Couldn't parse token #%d: ->%s<-", column[2],toks[column[2]]);
perror("strtoll error is");
return 1;
}
scratch.set2(&t);
}
if(3 < PAGELAYOUT::FMT::TUP::NN) {
char * endptr;
errno = 0;
typename PAGELAYOUT::FMT::TUP::TYP3 t = strtoll(toks[column[3]], &endptr, 0);
if(strlen(toks[column[3]]) - (size_t)(endptr-toks[column[3]]) > 1) {
printf("couldnt parse token #%d: ->%s<-\n", column[3], toks[column[3]]);
return 1;
}
if(errno) {
printf("Couldn't parse token #%d: ->%s<-", column[3],toks[column[3]]);
perror("strtoll error is");
return 1;
}
scratch.set3(&t);
}
if(4 < PAGELAYOUT::FMT::TUP::NN) {
char * endptr;
errno = 0;
typename PAGELAYOUT::FMT::TUP::TYP4 t = strtoll(toks[column[4]], &endptr, 0);
if(strlen(toks[column[4]]) - (size_t)(endptr-toks[column[4]]) > 1) {
printf("couldnt parse token #%d: ->%s<-\n", column[4], toks[column[4]]);
return 1;
}
if(errno) {
printf("Couldn't parse token #%d: ->%s<-", column[4],toks[column[4]]);
perror("strtoll error is");
return 1;
}
scratch.set4(&t);
}
if(5 < PAGELAYOUT::FMT::TUP::NN) {
char * endptr;
errno = 0;
typename PAGELAYOUT::FMT::TUP::TYP5 t = strtoll(toks[column[5]], &endptr, 0);
if(strlen(toks[column[5]]) - (size_t)(endptr-toks[column[5]]) > 1) {
printf("couldnt parse token #%d: ->%s<-\n", column[5], toks[column[5]]);
return 1;
}
if(errno) {
printf("Couldn't parse token #%d: ->%s<-", column[5],toks[column[5]]);
perror("strtoll error is");
return 1;
}
scratch.set5(&t);
}
if(6 < PAGELAYOUT::FMT::TUP::NN) {
char * endptr;
errno = 0;
typename PAGELAYOUT::FMT::TUP::TYP6 t = strtoll(toks[column[6]], &endptr, 0);
if(strlen(toks[column[6]]) - (size_t)(endptr-toks[column[6]]) > 1) {
printf("couldnt parse token #%d: ->%s<-\n", column[6], toks[column[6]]);
return 1;
}
if(errno) {
printf("Couldn't parse token #%d: ->%s<-", column[6],toks[column[6]]);
perror("strtoll error is");
return 1;
}
scratch.set6(&t);
}
if(7 < PAGELAYOUT::FMT::TUP::NN) {
char * endptr;
errno = 0;
typename PAGELAYOUT::FMT::TUP::TYP7 t = strtoll(toks[column[7]], &endptr, 0);
if(strlen(toks[column[7]]) - (size_t)(endptr-toks[column[7]]) > 1) {
printf("couldnt parse token #%d: ->%s<-\n", column[7], toks[column[7]]);
return 1;
}
if(errno) {
printf("Couldn't parse token #%d: ->%s<-", column[7],toks[column[7]]);
perror("strtoll error is");
return 1;
}
scratch.set7(&t);
}
if(8 < PAGELAYOUT::FMT::TUP::NN) {
char * endptr;
errno = 0;
typename PAGELAYOUT::FMT::TUP::TYP8 t = strtoll(toks[column[8]], &endptr, 0);
if(strlen(toks[column[8]]) - (size_t)(endptr-toks[column[8]]) > 1) {
printf("couldnt parse token #%d: ->%s<-\n", column[8], toks[column[8]]);
return 1;
}
if(errno) {
printf("Couldn't parse token #%d: ->%s<-", column[8],toks[column[8]]);
perror("strtoll error is");
return 1;
}
scratch.set8(&t);
}
if(9 < PAGELAYOUT::FMT::TUP::NN) {
char * endptr;
errno = 0;
typename PAGELAYOUT::FMT::TUP::TYP9 t = strtoll(toks[column[9]], &endptr, 0);
if(strlen(toks[column[9]]) - (size_t)(endptr-toks[column[9]]) > 1) {
printf("couldnt parse token #%d: ->%s<-\n", column[9], toks[column[9]]);
return 1;
}
if(errno) {
printf("Couldn't parse token #%d: ->%s<-", column[9],toks[column[9]]);
perror("strtoll error is");
return 1;
}
scratch.set9(&t);
}
// abort();
TlsmTableInsert(h,scratch);
count --;
if(!count) {
count = COUNT;
gettimeofday(&now_tv,0);
now = tv_to_double(now_tv);
printf("Wrote %d tuples "
"%9.3f Mtup/sec (avg) %9.3f Mtup/sec (cur) "
"%9.3f Mbyte/sec (avg) %9.3f Mbyte/sec (cur)\n",
inserts, //((inserts+1) * 100) / INSERTS,
((double)inserts/1000000.0)/(now-start),
((double)count/1000000.0)/(now-last_start),
(((double)PAGELAYOUT::FMT::TUP::sizeofBytes())*(double)inserts/1000000.0)/(now-start),
(((double)PAGELAYOUT::FMT::TUP::sizeofBytes())*(double)count/1000000.0)/(now-last_start)
);
last_start = now;
}
}
}
printf("insertions done.\n");
} else {
for(long int i = 0; i < INSERTS; i++) {
getTuple<PAGELAYOUT>(i,t);
TlsmTableInsert(h,t);
@ -105,6 +337,7 @@ namespace rose {
#endif
}
printf("insertions done.\n"); fflush(stdout);
count = COUNT;
gettimeofday(&start_tv,0);
@ -134,7 +367,7 @@ namespace rose {
);
last_start = now;
}
}
}
TlsmTableStop<PAGELAYOUT>(h);

2
src/stasis/page/compression/staticTuple.h
View file

@ -11,7 +11,7 @@ namespace rose {
static const char NORMAL = 0;
static const char TOMBSTONE = 1;
static const int TUPLE_ID = 1;
static const int NN = N;
typedef TYPE0 TYP0;
typedef TYPE1 TYP1;
typedef TYPE2 TYP2;