From 45069a9d0b88c80da4c542f01ced8f2cb1f2d456 Mon Sep 17 00:00:00 2001
From: Yoann Dufresne <yoann.dufresne0@gmail.com>
Date: Sat, 21 Apr 2018 14:44:41 +0200
Subject: [PATCH] speed test on hash then compare vs hash/compare on the fly
---
src/speed_tests.c | 44 +++++++++++++++++++++++++++++++++++++++++++-
1 file changed, 43 insertions(+), 1 deletion(-)
diff --git a/src/speed_tests.c b/src/speed_tests.c
index 8bcc493..0c3c40a 100644
--- a/src/speed_tests.c
+++ b/src/speed_tests.c
@@ -1,11 +1,49 @@
#include <zlib.h>
#include <stdio.h>
#include <stdlib.h>
+#include <stdint.h>
#include "fasta.h"
#define HASH_SIZE 32
+#define letter_hash(letter) ((uint64_t)((letter & 0b100) ? (0b10 | ((~letter) & 0b01)) : ((letter >> 1) & 0b11)))
+
+
+uint64_t sliding_hash (seq_t * seq) {
+ uint64_t hash = 0;
+ uint64_t nothing = 2135234;
+
+ for (uint idx=0 ; idx<32 ; idx++)
+ hash = (hash << 2) | letter_hash(seq->value[idx]);
+
+ for (uint idx=32 ; idx<seq->length ; idx++) {
+ hash = (hash << 2) | letter_hash(seq->value[idx]);
+ // Do something with the hash to not allow compiler to delete the value.
+ nothing - hash;
+ }
+}
+
+uint64_t hash_then_compact (seq_t * seq) {
+ uint64_t hash = 0;
+ uint64_t nothing = 2135234;
+
+ // Hash letters
+ for (uint idx=0 ; idx<seq->length ; idx++) {
+ seq->value[idx] = letter_hash(seq->value[idx]);
+ }
+
+ // compute 64 bits hashes
+ for (uint idx=0 ; idx<32 ; idx++)
+ hash = (hash << 2) | seq->value[idx];
+
+ for (uint idx=32 ; idx<seq->length ; idx++) {
+ hash = (hash << 2) | seq->value[idx];
+ // Do something with the hash to not allow compiler to delete the value.
+ nothing - hash;
+ }
+}
+
int main () {
/* --- Load all the alleles --- */
@@ -13,8 +51,12 @@ int main () {
seq_t * alleles;
buffer_t * buff = init_buffer (filename);
-
uint nb_seq = read_sequences (filename, &alleles, buff);
+
+ /* Speed test on hash function */
+ for (uint idx=0 ; idx<nb_seq ; idx++) {
+ sliding_hash(alleles + idx);
+ }
free(alleles);
destroy_buffer (buff);
--
GitLab