041 runs.sh: the suite three ways (the compiled code, the executor with batches, the executor alone) and icount.c's instructions and cycles; stage 1 at 10a50dd: knapsack 13.57 ms with batches (1,040.9 alone, 7.52 compiled, C 6.23), the others as they were; the executor's instructions as e2e56dd's or fewer (roots 2.756 to 2.760 billion, cycles alike), knapsack's 21.7 to 0.46 billion with batches

1 files changed, 27 insertions, 0 deletions

diff --git a/research/experiments/041-batches/icount.c b/research/experiments/041-batches/icount.c
new file mode 100644
index 0000000..d8e8943
--- /dev/null
+++ b/research/experiments/041-batches/icount.c
@@ -0,0 +1,27 @@
+/* icount CMD...: user-mode instructions and cycles of CMD and its children (perf_event_open, inherited). */
+#define _GNU_SOURCE
+#include <linux/perf_event.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <string.h>
+#include <sys/ioctl.h>
+#include <sys/syscall.h>
+#include <sys/wait.h>
+#include <unistd.h>
+static int counter(uint64_t config) {
+    struct perf_event_attr a;
+    memset(&a, 0, sizeof(a));
+    a.size = sizeof(a), a.type = PERF_TYPE_HARDWARE, a.config = config, a.disabled = 1, a.exclude_kernel = 1, a.exclude_hv = 1, a.inherit = 1;
+    return (int)syscall(SYS_perf_event_open, &a, 0, -1, -1, 0);
+}
+int main(int argc, char **argv) {
+    (void)argc;
+    int ins = counter(PERF_COUNT_HW_INSTRUCTIONS), cyc = counter(PERF_COUNT_HW_CPU_CYCLES);
+    ioctl(ins, PERF_EVENT_IOC_ENABLE, 0), ioctl(cyc, PERF_EVENT_IOC_ENABLE, 0);
+    pid_t p = fork();
+    if (!p) { execvp(argv[1], argv + 1); _exit(127); }
+    int st; waitpid(p, &st, 0);
+    uint64_t i = 0, c = 0;
+    read(ins, &i, 8), read(cyc, &c, 8);
+    fprintf(stderr, "{\"instructions\":%llu,\"cycles\":%llu}\n", (unsigned long long)i, (unsigned long long)c);
+}