52386401b3
Major Features: - Debug counter infrastructure for Refill Stage tracking - Free Pipeline counters (ss_local, ss_remote, tls_sll) - Diagnostic counters for early return analysis - Unified larson.sh benchmark runner with profiles - Phase 6-3 regression analysis documentation Bug Fixes: - Fix SuperSlab disabled by default (HAKMEM_TINY_USE_SUPERSLAB) - Fix profile variable naming consistency - Add .gitignore patterns for large files Performance: - Phase 6-3: 4.79 M ops/s (has OOM risk) - With SuperSlab: 3.13 M ops/s (+19% improvement) This is a clean repository without large log files. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
83 lines
2.4 KiB
C
83 lines
2.4 KiB
C
#include "hakmem_ace_stats.h"
|
|
#include <stdlib.h>
|
|
|
|
static _Atomic uint64_t g_mid_hit = 0;
|
|
static _Atomic uint64_t g_mid_miss = 0;
|
|
static _Atomic uint64_t g_large_hit = 0;
|
|
static _Atomic uint64_t g_large_miss = 0;
|
|
static _Atomic uint64_t g_l1_fallback = 0;
|
|
static _Atomic uint64_t g_mid_free = 0;
|
|
static _Atomic uint64_t g_large_free = 0;
|
|
|
|
static _Atomic uint32_t g_sample_mask = 0; // 0=off
|
|
static _Atomic uint64_t g_seq = 0;
|
|
|
|
void hkm_ace_stats_init(void) {
|
|
const char* env = getenv("HAKMEM_ACE_SAMPLE");
|
|
if (env) {
|
|
int n = atoi(env);
|
|
if (n > 0 && n < 31) {
|
|
uint32_t mask = (1u << n) - 1u; // sample every 2^n calls
|
|
atomic_store(&g_sample_mask, mask);
|
|
}
|
|
}
|
|
}
|
|
|
|
void hkm_ace_set_sample_mask(uint32_t mask) {
|
|
atomic_store(&g_sample_mask, mask);
|
|
}
|
|
|
|
static inline int should_sample(void) {
|
|
uint32_t mask = atomic_load(&g_sample_mask);
|
|
if (mask == 0) return 0;
|
|
uint64_t x = atomic_fetch_add(&g_seq, 1);
|
|
return ((x & mask) == 0);
|
|
}
|
|
|
|
void hkm_ace_stat_mid_attempt(int hit) {
|
|
if (!should_sample()) return;
|
|
if (hit) atomic_fetch_add(&g_mid_hit, 1);
|
|
else atomic_fetch_add(&g_mid_miss, 1);
|
|
}
|
|
|
|
void hkm_ace_stat_large_attempt(int hit) {
|
|
if (!should_sample()) return;
|
|
if (hit) atomic_fetch_add(&g_large_hit, 1);
|
|
else atomic_fetch_add(&g_large_miss, 1);
|
|
}
|
|
|
|
void hkm_ace_stat_l1_fallback(void) {
|
|
if (!should_sample()) return;
|
|
atomic_fetch_add(&g_l1_fallback, 1);
|
|
}
|
|
|
|
void hkm_ace_stat_mid_free(void) {
|
|
if (!should_sample()) return;
|
|
atomic_fetch_add(&g_mid_free, 1);
|
|
}
|
|
|
|
void hkm_ace_stat_large_free(void) {
|
|
if (!should_sample()) return;
|
|
atomic_fetch_add(&g_large_free, 1);
|
|
}
|
|
|
|
void hkm_ace_stats_snapshot(hkm_ace_stats_snapshot_t* out, int reset) {
|
|
if (!out) return;
|
|
out->mid_hit = atomic_load(&g_mid_hit);
|
|
out->mid_miss = atomic_load(&g_mid_miss);
|
|
out->large_hit = atomic_load(&g_large_hit);
|
|
out->large_miss = atomic_load(&g_large_miss);
|
|
out->l1_fallback = atomic_load(&g_l1_fallback);
|
|
out->mid_free = atomic_load(&g_mid_free);
|
|
out->large_free = atomic_load(&g_large_free);
|
|
if (reset) {
|
|
atomic_store(&g_mid_hit, 0);
|
|
atomic_store(&g_mid_miss, 0);
|
|
atomic_store(&g_large_hit, 0);
|
|
atomic_store(&g_large_miss, 0);
|
|
atomic_store(&g_l1_fallback, 0);
|
|
atomic_store(&g_mid_free, 0);
|
|
atomic_store(&g_large_free, 0);
|
|
}
|
|
}
|