added cache_benchmark

.gitignore

@@ -0,0 +1,3 @@
+*.exe
+*.swp
+*.o

cache_benchmark/Makefile

@@ -0,0 +1,2 @@
+build: cache.c
+	gcc cache.c -O2 -o cache.exe

cache_benchmark/cache.c

@@ -0,0 +1,107 @@
+#include<stdio.h>
+#include<stdlib.h>
+#include<time.h>
+
+//prints out usage information
+void usage();
+
+//generate dynamically allocated array, initialized with rand()
+int* rand_array(size_t n);
+
+//converts clock ticks to nano seconds
+double clocks_to_ns(double ct);
+
+//write to /dev/null or nul
+//exits if it fails
+void int_to_dev_null(int trash);
+
+//calculates the clock ticks rand() probably needs
+//from a hot cache
+double clock_ticks_rand();
+
+int main(int argc, char* argv[]) {
+	if(argc != 4) {
+		usage();
+		return 1;
+	}	
+	unsigned n_bytes;
+	unsigned array_size;
+	unsigned n_times_access;
+	unsigned n_loops;
+	sscanf(argv[1], "%u", &n_bytes);
+	sscanf(argv[2], "%u", &n_times_access);
+	sscanf(argv[3], "%u", &n_loops);
+	srand(time(NULL));
+	array_size = n_bytes / 4;
+	int* parry = rand_array(array_size);
+
+	clock_t start;
+	clock_t end;
+	clock_t overall = 0;
+	int tmp = 0;
+	double rc = clock_ticks_rand();
+	for(size_t n=0; n<n_loops; ++n) {
+		start = clock();
+		for(size_t t=0; t<n_times_access; ++t) {
+			//alternative way(without inner rand): tmp += parry[parry[t % array_size] % array_size];
+			tmp += parry[rand() % array_size];
+		}
+		end = clock();
+		overall += (end - start);
+	}
+	double per_instruct_ticks = overall / ((double) n_times_access * n_loops);
+	per_instruct_ticks = per_instruct_ticks - rc;
+	printf("%u, %u, %lf, %lf\n", n_bytes, array_size, clocks_to_ns(per_instruct_ticks), clocks_to_ns(rc));
+	int_to_dev_null(tmp);
+}
+double clocks_to_ns(double ct) {
+	return ct / CLOCKS_PER_SEC * 1000000000.0;
+}
+void int_to_dev_null(int trash) {
+#ifdef _WIN32
+	const char* dev_null = "nul";
+#else
+	const char* dev_null = "/dev/null";
+#endif
+	FILE* fnull = fopen(dev_null, "w");
+	if(! fnull) {
+		perror("unable to open /dev/null or nul");
+		exit(2);
+	}
+	fprintf(fnull, "%d", trash);
+	fclose(fnull);
+}
+
+int* array(size_t n) {
+	int* p=NULL;
+        p = malloc(sizeof(int) * n);
+	if(!p) {
+		perror("malloc failed");
+		exit(2);
+	}
+	return p;
+}
+int* rand_array(size_t n) {
+	int* p = array(n);
+	for(size_t i=0; i<n; ++i) {
+		p[i] = rand();
+	}
+	return p;
+}
+
+void usage() {
+	printf("<prog> array-size n-times-access n-loops\n");
+}
+double clock_ticks_rand() {
+#define N 1000000
+	for(size_t i=0; i<N; ++i) {
+		rand();
+	}
+	clock_t start = clock();
+	for(size_t i=0; i<N; ++i) {
+		rand();
+	}
+	double time_per_rand = (double)(clock() - start) / N;
+	return time_per_rand;
+#undef N
+}

cache_benchmark/execute_cache_test.sh

@@ -0,0 +1,86 @@
+#!/bin/bash
+#usage: <script> filename
+#	filename shouldn't be a path
+#	the file will be automatically added to the output folder
+if [ $# -ne 1 ]
+then
+	echo "missing output-filename"
+	exit 1
+fi
+
+exec >>./output/"$1"
+n_times=10000000
+n_count=5
+
+echo "#n_times = $n_times"
+echo "#n_count = $n_count"
+echo "#n_steps = $n_steps"
+
+step_size_bytes=100
+last_value_bytes=1000
+n_steps=$((last_value_bytes / step_size_bytes))
+
+n_bytes=4
+for i in `seq 0 $n_steps`
+do
+	./cache.exe $n_bytes $n_times $n_count
+	n_bytes=$((n_bytes + step_size_bytes))
+done
+
+step_size_bytes=1000
+last_value_bytes=10000
+n_steps=$((last_value_bytes / step_size_bytes))
+n_bytes=4
+for i in `seq 0 $n_steps`
+do
+	./cache.exe $n_bytes $n_times $n_count
+	n_bytes=$((n_bytes + step_size_bytes))
+done
+
+step_size_bytes=10000
+last_value_bytes=100000
+n_steps=$((last_value_bytes / step_size_bytes))
+n_bytes=4
+for i in `seq 0 $n_steps`
+do
+	./cache.exe $n_bytes $n_times $n_count
+	n_bytes=$((n_bytes + step_size_bytes))
+done
+
+step_size_bytes=100000
+last_value_bytes=1000000
+n_steps=$((last_value_bytes / step_size_bytes))
+n_bytes=4
+for i in `seq 0 $n_steps`
+do
+	./cache.exe $n_bytes $n_times $n_count
+	n_bytes=$((n_bytes + step_size_bytes))
+done
+
+step_size_bytes=1000000
+last_value_bytes=10000000
+n_steps=$((last_value_bytes / step_size_bytes))
+n_bytes=4
+for i in `seq 0 $n_steps`
+do
+	./cache.exe $n_bytes $n_times $n_count
+	n_bytes=$((n_bytes + step_size_bytes))
+done
+step_size_bytes=10000000
+last_value_bytes=100000000
+n_steps=$((last_value_bytes / step_size_bytes))
+n_bytes=4
+for i in `seq 0 $n_steps`
+do
+	./cache.exe $n_bytes $n_times $n_count
+	n_bytes=$((n_bytes + step_size_bytes))
+done
+step_size_bytes=100000000
+last_value_bytes=1000000000
+n_steps=$((last_value_bytes / step_size_bytes))
+n_bytes=4
+for i in `seq 0 $n_steps`
+do
+	./cache.exe $n_bytes $n_times $n_count
+	n_bytes=$((n_bytes + step_size_bytes))
+done

cache_benchmark/output/cache_test_t420.txt

@@ -0,0 +1,80 @@
+#n_times = 10000000
+#n_count = 5
+#n_steps = 
+4, 1, 1.796940, 11.427000
+104, 26, 2.900400, 10.326000
+204, 51, 2.884980, 10.322000
+304, 76, 2.013720, 11.318000
+404, 101, 2.281860, 11.109000
+504, 126, 3.361040, 10.419000
+604, 151, 2.853020, 10.410000
+704, 176, 2.825380, 10.880000
+804, 201, 2.450820, 10.867000
+904, 226, 2.844100, 10.433000
+1004, 251, 2.813580, 10.348000
+4, 1, 2.845820, 10.327000
+1004, 251, 2.675060, 10.475000
+2004, 501, 2.823520, 10.341000
+3004, 751, 2.845340, 10.321000
+4004, 1001, 2.540120, 10.626000
+5004, 1251, 2.811960, 10.348000
+6004, 1501, 2.845200, 10.326000
+7004, 1751, 2.842140, 10.322000
+8004, 2001, 2.852820, 10.322000
+9004, 2251, 2.822840, 10.398000
+10004, 2501, 2.827560, 10.334000
+4, 1, 2.666520, 10.492000
+10004, 2501, 2.814660, 10.359000
+20004, 5001, 2.840020, 10.326000
+30004, 7501, 2.450100, 10.714000
+40004, 10001, 2.802600, 10.356000
+50004, 12501, 2.802320, 10.362000
+60004, 15001, 2.824040, 10.339000
+70004, 17501, 2.834120, 10.321000
+80004, 20001, 2.823640, 10.339000
+90004, 22501, 2.840880, 10.322000
+100004, 25001, 2.842940, 10.326000
+4, 1, 2.865800, 10.329000
+100004, 25001, 2.797480, 10.359000
+200004, 50001, 2.825240, 10.342000
+300004, 75001, 2.817100, 10.347000
+400004, 100001, 2.825280, 10.331000
+500004, 125001, 2.900640, 10.350000
+600004, 150001, 2.827060, 10.345000
+700004, 175001, 2.833460, 10.343000
+800004, 200001, 2.812040, 10.346000
+900004, 225001, 2.562740, 10.644000
+1000004, 250001, 2.820360, 10.343000
+4, 1, 2.829680, 10.334000
+1000004, 250001, 2.826580, 10.347000
+2000004, 500001, 6.918360, 10.342000
+3000004, 750001, 12.378520, 10.587000
+4000004, 1000001, 16.424660, 10.293000
+5000004, 1250001, 21.886400, 10.344000
+6000004, 1500001, 22.119560, 10.812000
+7000004, 1750001, 26.157120, 10.286000
+8000004, 2000001, 25.323320, 10.414000
+9000004, 2250001, 27.974520, 10.294000
+10000004, 2500001, 26.809840, 10.320000
+4, 1, 2.501100, 10.660000
+10000004, 2500001, 26.890300, 10.295000
+20000004, 5000001, 31.671960, 10.288000
+30000004, 7500001, 31.187380, 10.286000
+40000004, 10000001, 33.487660, 10.413000
+50000004, 12500001, 31.705220, 10.305000
+60000004, 15000001, 34.792340, 10.288000
+70000004, 17500001, 34.498320, 10.760000
+80000004, 20000001, 32.972400, 10.309000
+90000004, 22500001, 36.819780, 10.300000
+100000004, 25000001, 33.829780, 10.413000
+4, 1, 2.820420, 10.378000
+100000004, 25000001, 36.639460, 10.292000
+200000004, 50000001, 37.877760, 10.387000
+300000004, 75000001, 41.470580, 10.338000
+400000004, 100000001, 42.162180, 10.371000
+500000004, 125000001, 50.617280, 10.766000
+600000004, 150000001, 48.964480, 10.322000
+700000004, 175000001, 55.075340, 10.292000
+800000004, 200000001, 53.195160, 10.285000
+900000004, 225000001, 54.293140, 10.611000
+1000000004, 250000001, 59.733980, 10.345000