Pass over the pools with nothing written when a part that uses the device address table syncs the shadows.

Under shadow storage the prologue of every part whose shaders use the table brings the shadow of every pool up
to date. It ran the full per-pool sync for each of them, 10,700 times a frame outdoors (3,772 pools, about
three such parts), though the write-watch snapshots of all pools are taken at once just before. A pool whose
regions hold no stamp newer than its shadow is now passed over with a compare per region: 1,500 pool syncs a
frame are left, and the copy-ins went from 1.32 to 1.01 ms a frame of the replay (RTX 5090 with
NX2NAVI_SHADOW=1, one standing run each way).

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
This commit is contained in:
chrg6
2026-11-10 23:14:42 -05:00
parent 2f6ff17651
commit 175c1947de

View File

@@ -2074,6 +2074,22 @@ int vk_mem_shadow_storage(void) { return shadow_on; }
static int gpa_shadow_pool(const PoolRec *r) { return r->kind == POOL_IMPORT && r->gpa_rec && r->smap; }
// gen_lock held: nothing of the pool was stamped since its shadow's last copy-in. Its regions have this epoch's
// write-watch snapshot (vk_mem_gpa_sync takes them for all pools at once), none of them is untracked, and none holds a
// stamp newer than the shadow's sgen, which then stays as it is. With cpu the stamps are the CPU-side tables', as
// shadow_sync_pool reads them. A table sync ran shadow_sync_pool for every pool at every part that uses the table,
// 10,700 times a frame outdoors (3,772 pools, about three such parts), each with its own timer reads, snapshot walk
// and region loop; with this test 1,500 pool syncs a frame are left, and the copy-ins went from 1.32 to 1.01 ms a
// frame (RTX 5090 with NX2NAVI_SHADOW=1, 2026-11-10).
static int shadow_pool_clean(const PoolRec *r, uint32_t epoch, int cpu) {
uint64_t r0 = r->import_lo >> SNAP_SHIFT, r1 = (r->import_hi - 1) >> SNAP_SHIFT;
if (!epoch || r1 >= SNAP_REGIONS) return 0;
const uint32_t *mx = cpu ? creg_max : reg_max;
for (uint64_t k = r0; k <= r1; k++)
if (snap_at[k] != epoch || snap_untracked[k] || mx[k] > r->sgen) return 0;
return 1;
}
void vk_mem_gpa_sync(void) {
if (!gpa_part_used) return;
gpa_part_used = 0;
@@ -2101,7 +2117,7 @@ void vk_mem_gpa_sync(void) {
if (have) s_queries += snap_regions(run0, run1, epoch);
for (int i = 0; i < pool_index.n; i++) {
PoolRec *r = pool_index.e[i].buf;
if (gpa_shadow_pool(r)) shadow_sync_pool(r, epoch, cpu), s_syncs++;
if (gpa_shadow_pool(r) && !shadow_pool_clean(r, epoch, cpu)) shadow_sync_pool(r, epoch, cpu), s_syncs++;
}
plat_mutex_unlock(&gen_lock);
// The uploads read the shadows just brought up to date, and the prologue records them ahead of the part, after