Beanstalk Longshot test farm
Menu

Report rp_a23e32f86e03a6a5 fail

#2004 WIP: perf(vk): a compute program's c0 layout kept by each thread between the words it reads (RX_VK_CS_LAYOUT_FAST)

any on FusionLord/agent-940fc0,

This report disagrees with its peers.

kokiri-walk#1 run_frametime_max_ms: +23.1% against peers' -0.6%

Machine

Commit
614cc3ee060a1523a896d74b52bb5428d3cb7724
Baseline
9dd41326e71094d8b774459dac5154eb5a3306d9
App
nightly-221-a9505cc3, attempt 1
Game build
e0f3ccfa
os
linux
cpu
AMD Ryzen 7 5800X 8-Core Processor
gpu
AMD Radeon RX 9070 (RADV GFX1201)
tags
linuxamdradvdgpu
driver
26.0.8
ram_gb
32
vram_mb
16,304
gpu_vendor
amd
os_version
Ubuntu 26.04.1 LTS (kernel 7.0.0)

Builds

main (baseline)

1,262.9 s warm

from 9dd41326e710

lift
108.8 s
read
4.1 s
finish
16.2 s
compile
1,125.5 s
shaders
7.6 s

PR

38 s warm

Program cache: warm

Scenarios

kokiri-walk RX_VK_CS_LAYOUT_FAST=0 (PR only)

ok
Metric (median)mainPRΔ
exit_code00±0%
run_hitch_count2726−3.7%
forest.hitch_count00±0%
log_forbidden_hits00±0%
forest.stutter_count00±0%
log_expected_missing00±0%
run_frametime_max_ms352.73373.61+5.9%
forest.frametime_max_ms23.8421.14−11.3%
forest.frametime_p50_ms17.4417.97+3%
forest.frametime_p99_ms19.4719.63+0.8%
forest.frametime_1pct_low_ms19.8820.01+0.7%
screenshot_diff—0.94—

Pictures against main: forest diff 0.9396 (main against itself )

Runs (2)
Metricmain 0pr 1
hangnono
crashednono
exit_code00
run_hitch_count2726
forest.hitch_count00
log_forbidden_hits00
forest.stutter_count00
log_expected_missing00
run_frametime_max_ms352.73373.61
forest.frametime_max_ms23.8421.14
forest.frametime_p50_ms17.4417.97
forest.frametime_p99_ms19.4719.63
forest.frametime_1pct_low_ms19.8820.01
screenshot_diff—0.94

kokiri-walk RX_VK_CS_LAYOUT_FAST=1 (PR only)

ok
Metric (median)mainPRΔ
exit_code00±0%
run_hitch_count2726−3.7%
forest.hitch_count00±0%
log_forbidden_hits00±0%
forest.stutter_count00±0%
log_expected_missing00±0%
run_frametime_max_ms330.03406.25+23.1%
forest.frametime_max_ms21.6320.88−3.5%
forest.frametime_p50_ms16.5317.36+5%
forest.frametime_p99_ms19.0719.48+2.1%
forest.frametime_1pct_low_ms19.4319.83+2.1%
screenshot_diff—0.96—

Pictures against main: forest diff 0.96 (main against itself )

Runs (2)
Metricmain 0pr 1
hangnono
crashednono
exit_code00
run_hitch_count2726
forest.hitch_count00
log_forbidden_hits00
forest.stutter_count00
log_expected_missing00
run_frametime_max_ms330.03406.25
forest.frametime_max_ms21.6320.88
forest.frametime_p50_ms16.5317.36
forest.frametime_p99_ms19.0719.48
forest.frametime_1pct_low_ms19.4319.83
screenshot_diff—0.96

Rules

ScenarioCellRuleLimitmainPRΔ %Status
kokiri-walkRX_VK_CS_LAYOUT_FAST=0 (PR only)crashFalse00—pass
kokiri-walkRX_VK_CS_LAYOUT_FAST=0 (PR only)screenshot.max_diff0.02—0.94—fail
kokiri-walkRX_VK_CS_LAYOUT_FAST=0 (PR only)forest.frametime_p99_ms.max_delta_pct1019.4719.63+0.8%inconclusiveneed two usable starts per side (baseline: 1, PR: 1)
kokiri-walkRX_VK_CS_LAYOUT_FAST=0 (PR only)log_forbidden_hits.max000—pass
kokiri-walkRX_VK_CS_LAYOUT_FAST=1 (PR only)crashFalse00—pass
kokiri-walkRX_VK_CS_LAYOUT_FAST=1 (PR only)screenshot.max_diff0.02—0.96—fail
kokiri-walkRX_VK_CS_LAYOUT_FAST=1 (PR only)forest.frametime_p99_ms.max_delta_pct1019.0719.48+2.1%inconclusiveneed two usable starts per side (baseline: 1, PR: 1)
kokiri-walkRX_VK_CS_LAYOUT_FAST=1 (PR only)log_forbidden_hits.max000—pass
kokiri-walkRX_VK_CS_LAYOUT_FAST=1 (PR only)forest.frametime_p99_ms: RX_VK_CS_LAYOUT_FAST=1 vs RX_VK_CS_LAYOUT_FAST=0 (PR)519.6319.48−0.8%passthe PR's own cells: RX_VK_CS_LAYOUT_FAST=1 against RX_VK_CS_LAYOUT_FAST=0 (RX_VK_CS_LAYOUT_FAST=0 (PR only))

Pictures of the failure

The last frame of a step that got stuck, or main's and the PR's picture where they differed more than the plan allows; shrunk, and kept a day: until .

kokiri-walk-c0-forest-main-failed.png
kokiri-walk-c0-forest-main-failed.png
kokiri-walk-c0-forest-pr-failed.png
kokiri-walk-c0-forest-pr-failed.png
kokiri-walk-c1-forest-main-failed.png
kokiri-walk-c1-forest-main-failed.png
kokiri-walk-c1-forest-pr-failed.png
kokiri-walk-c1-forest-pr-failed.png

Log (PR side, end)

[hos host] vk: a capture taken (ms): all 0.015; GPU wait for late pages 0.000 (0 of 600 takes), pools lock 0.000, retires and stamps 0.002, bytes copied 0.000; 24 kept (102 MB), 12 spare buffers, 0 texture pages copied (0 MB)
[hos ModuleSystemWorker5] vk: readbacks of GPU results: 983 of 3000 waited for their write-back, 0.03 ms each; 0 looked at the stopped queues first, 0 waited for their submission's last part (guest memory read in place after theirs), 0 waited for their whole submission (their pages written in place, no mirror)
[hos ModuleSystemWorker5] vk: invalidates since the last readbacks line: 3603 kept their virtual core (nothing there to wait for), 3053 gave it up (counted by 256 a thread; RX_VK_INVALIDATE_QUICK=0: none kept)
[hos host] vk: replay reads of guest memory as submitted: 3740000 (1491598 pages from the captures)
[hos host] vk: queues ordered by their events: 36000 submissions stopped at a wait, 35732 went on once it was reached, 265 waits gone past; made to go by presents 0, releases 0, waits for the replay 0, 50 ms 19 (stopped 50.06 ms at most); waits met 95806, not met 0; captures at a CPU event's wait 5173, 0 of them at a wait met on arrival
[hos host] vk: sparse mirrors: 57, 760 MB with memory of 964 MB (203 MB spared); blocks given memory later 923, given back 19752; bindings served from guest memory meanwhile 0 (0 reaching blocks with memory too), from the mirror past every game pool 0; 93 binds (75.8 ms); 0 made whole for want of memory
[hos host] vk: sparse mirrors' buffers: 6 taken from the spares, 68 made new; 17 kept as spares, 0 destroyed; 11 spares now, 308 MB of address space (RX_VK_SPARSE_SPARE_MB=4096)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5768 dirty pages from their submits' captures, 51507 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 907 dispatches, 41846 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7142 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19949 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: imports: 68 live (1148 MB), 0 retired still imported (0 MB), 9 retired given back (256 MB)
[hos sead::MainThread] nvn: GPU waits in 5 s: 288 sync waits (1410 ms), 0 polls timed out (0 ms), 0 queue finishes (0 ms)
[hos host] vk: replay reads of guest memory as submitted: 3760000 (1499335 pages from the captures)
[hos host] vk: replay reads of guest memory as submitted: 3780000 (1507058 pages from the captures)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5875 dirty pages from their submits' captures, 52035 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 914 dispatches, 42193 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7125 runs, 693.5 MB compared; 42 pages read into the shadows; 260 MB of shadows (122 MB of pages known, in 156 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 20389 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: replay reads of guest memory as submitted: 3800000 (1514804 pages from the captures)
[hos host] vk: a submit's memory work (ms): maps 0.00, capture taken 0.02, writes collected 0.03
[hos host] vk: late pages found at the replay (600 takes): 0 pages in 0 takes (RX_VK_LATE_AT_REPLAY=0: only the ones the capture found)
[hos ModuleSystemWorker2] nvn: CommandBufferSignalEvent #70000: event 0x100d6c801c mode 0 value 69997 (lr 0x802911b8)
[hos host] vk: program picks: 4264613 kept (2314411 by the last picks' textures, a uniform buffer bound since), 22427 new, 1712959 with other textures (of 6000000)
[hos host] vk: replay reads of guest memory as submitted: 3820000 (1522564 pages from the captures)
[hos sead::MainThread] vk: capture: a submit 0.09 ms (lock 0.00, scan 0.02 over 1229 ranges, 1865 MB), 219 pages written, 219 copied (0.9 MB)
[hos host] vk: a capture taken (ms): all 0.015; GPU wait for late pages 0.000 (0 of 600 takes), pools lock 0.000, retires and stamps 0.002, bytes copied 0.000; 25 kept (104 MB), 10 spare buffers, 0 texture pages copied (0 MB)
[hos host] vk: sparse mirrors: 57, 760 MB with memory of 964 MB (203 MB spared); blocks given memory later 923, given back 19752; bindings served from guest memory meanwhile 0 (0 reaching blocks with memory too), from the mirror past every game pool 0; 93 binds (75.8 ms); 0 made whole for want of memory
[hos host] vk: sparse mirrors' buffers: 6 taken from the spares, 68 made new; 17 kept as spares, 0 destroyed; 11 spares now, 308 MB of address space (RX_VK_SPARSE_SPARE_MB=4096)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5726 dirty pages from their submits' captures, 51761 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 911 dispatches, 42007 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7138 runs, 693.6 MB compared; 41 pages read into the shadows; 260 MB of shadows (123 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19984 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: imports: 68 live (1148 MB), 0 retired still imported (0 MB), 9 retired given back (256 MB)
[hos host] vk: late pages' fix (600 recordings with one): of 600 captures with late pages 600 fixed (131.1 pages, 132.1 fixes a capture), waited as before: 0 no pool, 0 no stamps or shadow yet, 0 stamps too new, 0 whole copies, 0 in place or unknown, 0 a page's earlier fix not recorded yet, 0 other; 52 read on the CPU first, 0 with no table, 0 pages whose shadow changed first; stamps 710 blocks (45440 KB); of the reads on the CPU, 0 waited for an earlier part's fix, 0 copied the page again instead (that part not waited for); the paths: reads on the CPU with the fix done 50, of an earlier recording 2, not recorded yet 0, of this part 0; pages fixed from stamps 77489, from snapshots 1200; the shadows: 292421 pages synced from captures, 291793 made known by it, 628 left unknown because the replay or the collection wrote them since
[hos host] vk: replay reads of guest memory as submitted: 3840000 (1530247 pages from the captures)
[hos host] vk: render targets drawn over from another start, let go: 294 so far (1651 MB), 206 of them (1223 MB) between the passes
[hos host] vk: a frame: 56 draws kept their neighbour's descriptors (0 of them through other slots' registrations)
[hos host] vk: a frame: 0.0 draws left out (not ready), 0.00 waits for shaders or pipelines (0.0 ms); translation 3276 ms in all, 0 jobs queued
[hos host] vk: a frame's replay time (ms): submit 7.73, present 0.16, draw 5.00, descriptors 2.05, texture lookups 0.07, uploads 0.00, pipelines 0.17, programs 0.82, commands 1.63, dispatch 0.77, memory sync 0.29, queue submit 0.71, slot wait 0.09
[hos host] vk: a frame: mirrors 74 (760 MB): 5.46 MB synced in 1019.2 copies, 16.92 MB written back
[hos host] vk: 5640 presents, 57.6 fps; a frame: 1600 draws (0 skipped), 145 dispatches, 165 passes, 144 barriers, 6 submits, 0.0 uploads (0.0 MB), 80 ub updates (56 overlays), GPU-bound wait 0.1 ms (recording slots only); 1117 pipelines (97 ms), 661 programs (10 failed), 1659 textures (3805 MB)
[hos host] vk: host time in GPU resource waits: 5.14 ms a frame (fences 0.00, timelines 5.14, idle 0.00, queries 0.00; sum across threads, not frame time; excludes pacing and driver submit/present blocking)
[hos host] vk: a frame: 629.4 full barriers recorded, 301.4 left out (nothing recorded since the one before), 18.2 pass starts ordered in their fragment stages only (draw passes after draw passes)
[hos host] vk: a frame: the game's barriers by their NVN flags: 0x1: 2.0; 0x8: 3.0; 0x9: 1.0; 0x11: 45.5; 0x12: 68.5; 0x13: 2.0; 0x21: 4.0; 0x28: 14.0; 0x92: 3.0; 0x97: 1.0;
[hos host] vk: a frame: indirect mesh draws' arguments (RX_VK_MESH_CONV=2): 41.0 converted one by one, 0.0 from 0.0 batches at the game's barriers (0.0 converted in them; 0.0 dropped after a write, 0.0 the ring came round), 41.0 for the compute fallback (no mesh shaders in use: one by one); scans ended at a barrier 0.0, an op 0.0, the end 0.0, full 0.0; the draws batched by a scan right after an op that ended the one before, by op: none
[hos host] vk: GPU time 13.3 ms a frame (the longest submission 3.8 ms); 0.0 textures made a frame: colour targets 0.0 (0.0 MB), depth 0.0 (0.0 MB), ASTC or BC 0.0 (0.0 MB), other 0.0 (0.0 MB)
[hos host] vk: a frame: 0.00 MB unswizzled on the CPU in 0.00 ms
[hos aal::LowPrioWorkerThread] vcores: 97991 takes, 15505 waits (mean 85 us, longest 5.9 ms, 0 over HOS_VCORE_LOG_MS), 6644 kicks, 7121 yields, 0 starved grants, 0 escapes in 5 s
[hos host] vk: replay reads of guest memory as submitted: 3860000 (1537939 pages from the captures)
[hos Presentation Thread] vi: 5657 frames presented (57.6 fps over the last 5.0 s, interval 1)
[hos Presentation Thread] vi: present intervals: <15.5 ms 0, 15.5-17.8 187, 17.8-22 102, 22-30 0, 30-40 0, over 40 0
[hos host] vk: the recording streams: 48 of 1 MB, the most one held 181 KB; 0 made on demand (0 MB kept now); c1 stages: at most 120 a recording; 0 descriptor pools made when a recording's own ran out (0 slots' pools made whole); a recording's pool took at most 153 sets (ub 1695, sb 926, cis 11724, si 0, utb 0, stb 0)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5717 dirty pages from their submits' captures, 51524 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 907 dispatches, 41865 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7141 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (123 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19965 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: replay reads of guest memory as submitted: 3880000 (1545691 pages from the captures)
[hos rmbl::VibrationThread] hid: 16384 vibration values, strongest 0.00 since the last count
[hos ModuleSystemWorker3] vk: readbacks of GPU results: 982 of 3000 waited for their write-back, 0.04 ms each; 0 looked at the stopped queues first, 0 waited for their submission's last part (guest memory read in place after theirs), 0 waited for their whole submission (their pages written in place, no mirror)
[hos ModuleSystemWorker3] vk: invalidates since the last readbacks line: 3531 kept their virtual core (nothing there to wait for), 2869 gave it up (counted by 256 a thread; RX_VK_INVALIDATE_QUICK=0: none kept)
[hos host] vk: replay reads of guest memory as submitted: 3900000 (1553392 pages from the captures)
[hos host] vk: a submit's memory work (ms): maps 0.00, capture taken 0.02, writes collected 0.03
[hos ModuleSystemWorker3] nvn: CommandBufferWaitEvent #100000: event 0x100d6c8010 mode 0 value 71875 (now 71856) (lr 0x80660e38)
[hos ModuleSystemWorker3] nvn: DrawMeshTasksIndirect: 160000 so far (x1 0x117db40020 x2 0 x3 0x117db40000 x4 0x44b040)
[hos Presentation Thread] vk: capture: a submit 0.09 ms (lock 0.00, scan 0.02 over 1229 ranges, 1865 MB), 216 pages written, 216 copied (0.8 MB)
[hos host] vk: a capture taken (ms): all 0.014; GPU wait for late pages 0.000 (0 of 600 takes), pools lock 0.000, retires and stamps 0.002, bytes copied 0.000; 24 kept (102 MB), 12 spare buffers, 0 texture pages copied (0 MB)
[hos host] vk: replay reads of guest memory as submitted: 3920000 (1561130 pages from the captures)
[hos host] vk: sparse mirrors: 57, 760 MB with memory of 964 MB (203 MB spared); blocks given memory later 923, given back 19752; bindings served from guest memory meanwhile 0 (0 reaching blocks with memory too), from the mirror past every game pool 0; 93 binds (75.8 ms); 0 made whole for want of memory
[hos host] vk: sparse mirrors' buffers: 6 taken from the spares, 68 made new; 17 kept as spares, 0 destroyed; 11 spares now, 308 MB of address space (RX_VK_SPARSE_SPARE_MB=4096)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5753 dirty pages from their submits' captures, 51609 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 907 dispatches, 41885 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7131 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (123 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19967 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: imports: 68 live (1148 MB), 0 retired still imported (0 MB), 9 retired given back (256 MB)
[hos host] vk: replay reads of guest memory as submitted: 3940000 (1568808 pages from the captures)
[hos host] vk: readback pages refreshed from guest memory: 1318324 pages so far (0 held: may be written ahead in their submission)
[hos host] vk: program picks: 4408371 kept (2393376 by the last picks' textures, a uniform buffer bound since), 22427 new, 1769201 with other textures (of 6200000)
[hos host] vk: replay reads of guest memory as submitted: 3960000 (1576625 pages from the captures)
[hos host] vk: write-backs left out (600 submits): 965, 3944.8 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5774 dirty pages from their submits' captures, 51980 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 911 dispatches, 42327 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7135 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (123 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19879 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: a frame: 56 draws kept their neighbour's descriptors (0 of them through other slots' registrations)
[hos host] vk: a frame: 0.0 draws left out (not ready), 0.00 waits for shaders or pipelines (0.0 ms); translation 3276 ms in all, 0 jobs queued
[hos host] vk: a frame's replay time (ms): submit 7.70, present 0.14, draw 4.90, descriptors 2.01, texture lookups 0.07, uploads 0.00, pipelines 0.16, programs 0.81, commands 1.60, dispatch 0.77, memory sync 0.29, queue submit 0.71, slot wait 0.08
[hos host] vk: a frame: mirrors 74 (760 MB): 5.46 MB synced in 1017.6 copies, 16.92 MB written back
[hos host] vk: 5760 presents, 57.9 fps; a frame: 1595 draws (0 skipped), 145 dispatches, 165 passes, 144 barriers, 6 submits, 0.0 uploads (0.0 MB), 80 ub updates (56 overlays), GPU-bound wait 0.1 ms (recording slots only); 1117 pipelines (97 ms), 661 programs (10 failed), 1659 textures (3805 MB)
[hos host] vk: host time in GPU resource waits: 5.00 ms a frame (fences 0.00, timelines 5.00, idle 0.00, queries 0.00; sum across threads, not frame time; excludes pacing and driver submit/present blocking)
[hos host] vk: a frame: 629.2 full barriers recorded, 301.3 left out (nothing recorded since the one before), 18.2 pass starts ordered in their fragment stages only (draw passes after draw passes)
[hos host] vk: a frame: the game's barriers by their NVN flags: 0x1: 2.0; 0x8: 3.0; 0x9: 1.0; 0x11: 45.5; 0x12: 68.5; 0x13: 2.0; 0x21: 4.0; 0x28: 14.0; 0x92: 3.0; 0x97: 1.0;
[hos host] vk: a frame: indirect mesh draws' arguments (RX_VK_MESH_CONV=2): 41.0 converted one by one, 0.0 from 0.0 batches at the game's barriers (0.0 converted in them; 0.0 dropped after a write, 0.0 the ring came round), 41.0 for the compute fallback (no mesh shaders in use: one by one); scans ended at a barrier 0.0, an op 0.0, the end 0.0, full 0.0; the draws batched by a scan right after an op that ended the one before, by op: none
[hos host] vk: GPU time 13.2 ms a frame (the longest submission 3.9 ms); 0.0 textures made a frame: colour targets 0.0 (0.0 MB), depth 0.0 (0.0 MB), ASTC or BC 0.0 (0.0 MB), other 0.0 (0.0 MB)
[hos host] vk: a frame: 0.00 MB unswizzled on the CPU in 0.00 ms
[hos host] vk: replay reads of guest memory as submitted: 3980000 (1584221 pages from the captures)
[hos host] vk: late pages found at the replay (600 takes): 0 pages in 0 takes (RX_VK_LATE_AT_REPLAY=0: only the ones the capture found)
[hos host] vk: replay reads of guest memory as submitted: 4000000 (1592007 pages from the captures)
[hos host] vk: sparse mirrors: 57, 760 MB with memory of 964 MB (203 MB spared); blocks given memory later 923, given back 19752; bindings served from guest memory meanwhile 0 (0 reaching blocks with memory too), from the mirror past every game pool 0; 93 binds (75.8 ms); 0 made whole for want of memory
[hos host] vk: sparse mirrors' buffers: 6 taken from the spares, 68 made new; 17 kept as spares, 0 destroyed; 11 spares now, 308 MB of address space (RX_VK_SPARSE_SPARE_MB=4096)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5738 dirty pages from their submits' captures, 51536 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 907 dispatches, 41800 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7150 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19937 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: imports: 68 live (1148 MB), 0 retired still imported (0 MB), 9 retired given back (256 MB)
[hos Presentation Thread] vk: capture: a submit 0.09 ms (lock 0.00, scan 0.02 over 1229 ranges, 1865 MB), 218 pages written, 218 copied (0.9 MB)
[hos host] vk: a capture taken (ms): all 0.015; GPU wait for late pages 0.000 (0 of 600 takes), pools lock 0.000, retires and stamps 0.002, bytes copied 0.000; 24 kept (102 MB), 11 spare buffers, 0 texture pages copied (0 MB)
[hos host] vk: a submit's memory work (ms): maps 0.00, capture taken 0.02, writes collected 0.03
[hos host] vk: replay reads of guest memory as submitted: 4020000 (1601685 pages from the captures)
[hos ModuleSystemWorker3] vk: readbacks of GPU results: 980 of 3000 waited for their write-back, 0.03 ms each; 0 looked at the stopped queues first, 0 waited for their submission's last part (guest memory read in place after theirs), 0 waited for their whole submission (their pages written in place, no mirror)
[hos ModuleSystemWorker3] vk: invalidates since the last readbacks line: 3673 kept their virtual core (nothing there to wait for), 2983 gave it up (counted by 256 a thread; RX_VK_INVALIDATE_QUICK=0: none kept)
[hos host] vk: replay reads of guest memory as submitted: 4040000 (1609192 pages from the captures)
[hos ModuleSystemWorker5] nvn: 600000 recordings made, 49023 live copies (15.9 MB), 351 texture copies; 4 command buffers' segments (4.2 MB)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5852 dirty pages from their submits' captures, 51649 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 908 dispatches, 41922 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7136 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19923 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: replay reads of guest memory as submitted: 4060000 (1616961 pages from the captures)
[hos sead::MainThread] nvn: GPU waits in 5 s: 289 sync waits (1396 ms), 0 polls timed out (0 ms), 0 queue finishes (0 ms)
[hos host] vk: program picks: 4552253 kept (2472371 by the last picks' textures, a uniform buffer bound since), 22427 new, 1825319 with other textures (of 6400000)
[hos host] vk: replay reads of guest memory as submitted: 4080000 (1624613 pages from the captures)
[hos ModuleSystemWorker1] nvn: CommandBufferSignalEvent #75000: event 0x100d6c8030 mode 0 value 74999 (lr 0x806612a8)
[hos host] vk: the recording streams: 48 of 1 MB, the most one held 181 KB; 0 made on demand (0 MB kept now); c1 stages: at most 120 a recording; 0 descriptor pools made when a recording's own ran out (0 slots' pools made whole); a recording's pool took at most 153 sets (ub 1695, sb 926, cis 11724, si 0, utb 0, stb 0)
[hos host] vk: sparse mirrors: 57, 760 MB with memory of 964 MB (203 MB spared); blocks given memory later 923, given back 19752; bindings served from guest memory meanwhile 0 (0 reaching blocks with memory too), from the mirror past every game pool 0; 93 binds (75.8 ms); 0 made whole for want of memory
[hos host] vk: sparse mirrors' buffers: 6 taken from the spares, 68 made new; 17 kept as spares, 0 destroyed; 11 spares now, 308 MB of address space (RX_VK_SPARSE_SPARE_MB=4096)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5728 dirty pages from their submits' captures, 51465 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 907 dispatches, 41811 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7231 runs, 709.6 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19922 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: imports: 68 live (1148 MB), 0 retired still imported (0 MB), 9 retired given back (256 MB)
[hos host] vk: replay reads of guest memory as submitted: 4100000 (1632287 pages from the captures)
[hos host] vk: a frame: 56 draws kept their neighbour's descriptors (0 of them through other slots' registrations)
[hos host] vk: a frame: 0.0 draws left out (not ready), 0.00 waits for shaders or pipelines (0.0 ms); translation 3276 ms in all, 0 jobs queued
[hos host] vk: a frame's replay time (ms): submit 7.70, present 0.14, draw 4.89, descriptors 2.00, texture lookups 0.07, uploads 0.00, pipelines 0.17, programs 0.81, commands 1.59, dispatch 0.75, memory sync 0.29, queue submit 0.70, slot wait 0.09
[hos host] vk: a frame: mirrors 74 (760 MB): 5.46 MB synced in 1018.6 copies, 16.92 MB written back
[hos host] vk: 5880 presents, 57.5 fps; a frame: 1594 draws (0 skipped), 145 dispatches, 166 passes, 144 barriers, 6 submits, 0.0 uploads (0.0 MB), 80 ub updates (56 overlays), GPU-bound wait 0.1 ms (recording slots only); 1117 pipelines (97 ms), 661 programs (10 failed), 1659 textures (3805 MB)
[hos host] vk: host time in GPU resource waits: 4.93 ms a frame (fences 0.00, timelines 4.93, idle 0.00, queries 0.00; sum across threads, not frame time; excludes pacing and driver submit/present blocking)
[hos host] vk: a frame: 629.9 full barriers recorded, 301.4 left out (nothing recorded since the one before), 18.2 pass starts ordered in their fragment stages only (draw passes after draw passes)
[hos host] vk: a frame: the game's barriers by their NVN flags: 0x1: 2.0; 0x8: 3.0; 0x9: 1.0; 0x11: 45.5; 0x12: 68.5; 0x13: 2.0; 0x21: 4.0; 0x28: 14.0; 0x92: 3.0; 0x97: 1.0;
[hos host] vk: a frame: indirect mesh draws' arguments (RX_VK_MESH_CONV=2): 41.0 converted one by one, 0.0 from 0.0 batches at the game's barriers (0.0 converted in them; 0.0 dropped after a write, 0.0 the ring came round), 41.0 for the compute fallback (no mesh shaders in use: one by one); scans ended at a barrier 0.0, an op 0.0, the end 0.0, full 0.0; the draws batched by a scan right after an op that ended the one before, by op: none
[hos host] vk: GPU time 13.2 ms a frame (the longest submission 3.8 ms); 0.0 textures made a frame: colour targets 0.0 (0.0 MB), depth 0.0 (0.0 MB), ASTC or BC 0.0 (0.0 MB), other 0.0 (0.0 MB)
[hos host] vk: a frame: 0.00 MB unswizzled on the CPU in 0.00 ms
[hos sead::MainThread] vk: capture: a submit 0.09 ms (lock 0.00, scan 0.02 over 1229 ranges, 1865 MB), 219 pages written, 219 copied (0.9 MB)
[hos host] vk: a capture taken (ms): all 0.015; GPU wait for late pages 0.000 (0 of 600 takes), pools lock 0.000, retires and stamps 0.002, bytes copied 0.000; 25 kept (104 MB), 11 spare buffers, 0 texture pages copied (0 MB)
[hos host] vk: replay reads of guest memory as submitted: 4120000 (1639979 pages from the captures)
[hos host] vk: a submit's memory work (ms): maps 0.00, capture taken 0.02, writes collected 0.03
[hos ModuleSystemWorker4] nvn: CommandBufferWaitEvent #105000: event 0x100d6c800c mode 0 value 75674 (now 75655) (lr 0x80660e38)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5738 dirty pages from their submits' captures, 51404 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 908 dispatches, 41806 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7166 runs, 694.2 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19865 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: replay reads of guest memory as submitted: 4140000 (1647608 pages from the captures)
[hos BaseProcExecutor1] vcores: 98005 takes, 15616 waits (mean 82 us, longest 6.0 ms, 0 over HOS_VCORE_LOG_MS), 6712 kicks, 7191 yields, 0 starved grants, 0 escapes in 5 s
[hos host] vk: readback pages refreshed from guest memory: 1391860 pages so far (0 held: may be written ahead in their submission)
[hos host] vk: replay reads of guest memory as submitted: 4160000 (1655329 pages from the captures)
[hos Presentation Thread] vi: 5946 frames presented (57.7 fps over the last 5.0 s, interval 1)
[hos Presentation Thread] vi: present intervals: <15.5 ms 1, 15.5-17.8 189, 17.8-22 99, 22-30 0, 30-40 0, over 40 0
[hos host] vk: late pages found at the replay (600 takes): 0 pages in 0 takes (RX_VK_LATE_AT_REPLAY=0: only the ones the capture found)
[hos BaseProcExecutor3] sync: object cache (HOS_OBJ_CACHE=16384 a thread, 35 made): 14858171 lookups, 123373 missed (0.83 %)
[hos host] vk: replay reads of guest memory as submitted: 4180000 (1662903 pages from the captures)
[hos host] vk: sparse mirrors: 57, 760 MB with memory of 964 MB (203 MB spared); blocks given memory later 923, given back 19752; bindings served from guest memory meanwhile 0 (0 reaching blocks with memory too), from the mirror past every game pool 0; 93 binds (75.8 ms); 0 made whole for want of memory
[hos host] vk: sparse mirrors' buffers: 6 taken from the spares, 68 made new; 17 kept as spares, 0 destroyed; 11 spares now, 308 MB of address space (RX_VK_SPARSE_SPARE_MB=4096)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5768 dirty pages from their submits' captures, 51561 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 908 dispatches, 41842 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7134 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19950 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: imports: 68 live (1148 MB), 0 retired still imported (0 MB), 9 retired given back (256 MB)
[hos ModuleSystemWorker3] vk: readbacks of GPU results: 980 of 3000 waited for their write-back, 0.05 ms each; 0 looked at the stopped queues first, 0 waited for their submission's last part (guest memory read in place after theirs), 0 waited for their whole submission (their pages written in place, no mirror)
[hos ModuleSystemWorker3] vk: invalidates since the last readbacks line: 3731 kept their virtual core (nothing there to wait for), 3181 gave it up (counted by 256 a thread; RX_VK_INVALIDATE_QUICK=0: none kept)
[hos sead::MainThread] vk: capture: a submit 0.09 ms (lock 0.00, scan 0.02 over 1229 ranges, 1865 MB), 216 pages written, 216 copied (0.8 MB)
[hos host] vk: a capture taken (ms): all 0.015; GPU wait for late pages 0.000 (0 of 600 takes), pools lock 0.000, retires and stamps 0.002, bytes copied 0.000; 25 kept (96 MB), 11 spare buffers, 0 texture pages copied (0 MB)
[hos host] vk: program picks: 4695828 kept (2550927 by the last picks' textures, a uniform buffer bound since), 22427 new, 1881744 with other textures (of 6600000)
[hos host] vk: replay reads of guest memory as submitted: 4200000 (1670592 pages from the captures)
[hos host] vk: replay reads of guest memory as submitted: 4220000 (1678315 pages from the captures)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5741 dirty pages from their submits' captures, 51472 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 907 dispatches, 41789 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7139 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19890 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: memory: working set 6379 MB (peak 6423), private 1237 MB, staging 32 MB in 2 arenas, 20 made so far (0 ms)
[hos host] vk: memory heaps: heap 0 (system): 14 blocks 648 MB, 39 allocations 624 MB; used 1812 of 15328 MB; heap 1 (device): 176 blocks 5501 MB, 197164 allocations 5143 MB; used 5674 of 14541 MB;
[hos host] vk: image blocks: 1 blocks 32 MB, 5 images 16 MB; 1 free ranges, the largest 15 MB
[hos host] vk: image blocks (render targets): 16 blocks 512 MB, 97 images 294 MB; 46 free ranges, the largest 32 MB
[hos host] vk: image blocks (textures, compacted): 70 blocks 2240 MB, 1548 images 2208 MB; 289 free ranges, the largest 27 MB
[hos host] vk: render target images: 177 reused from the pool, 284 made new, 184 destroyed from it, 23 reused by size (on a pooled one's memory, RX_VK_TARGET_REUSE_MEMORY=1), 6 missed after their image was taken by size; it holds 0 (0 MB)
[hos host] vk: render targets not reused by size: 187 with the pool empty (0 of them with targets still in the dead queue), 20 with none in it a present yet, 1 with none of a kind they can use, 20 with all smaller (the largest 55% short on average), 14 with all bigger (over 1.5 times), 42 with some of each, 0 over a block, 0 refused by the driver, 0 not bound; the pool let go of 0 over its bound and 184 idle
[hos host] vk: VRAM: 5674 MB used of 14541 MB allowed (358 MB of it free in the blocks); textures 3805 MB (budget 10909 MB), 1659 textures
[hos host] vk: budget: VRAM 5501 of 14541 MB (5674 MB by the driver, 172 MB unaccounted): textures 3805, mirrors 760, blocks free 358, shadows 260, other buffers 221, other images 63, staging 32; RAM 1798 MB (the process's private memory 1237 MB): imports 1148, capture spares 518, captures 100, capture table 26, other buffers 6, others 0
[hos host] vk: textures: decoded ASTC 2441 MB, rendered 877 MB (102 drawn in one frame, kept: 54738 KB), other 373 MB; used in the last 8 presents 2743 MB
[hos host] vk: decoded ASTC used again: 0 textures (0 MB) copied from ones the game finalized, not decoded; 0 of those retired, 3 listed; 503 MB of blocks hashed in 45 ms
[hos host] vk:   288 MB: 0x1147c16000 12288x12288x1 x1 levels 1 NVN 0x32 rendered, last used 1 presents ago
[hos host] vk:   214 MB: 0x10a22aa600 1365x1365x1 x90 levels 11 NVN 0x8e ASTC, last used 1 presents ago
[hos host] vk:   214 MB: 0x10a5b22c00 1365x1365x1 x90 levels 11 NVN 0x80 ASTC, last used 1 presents ago
[hos host] vk:   64 MB: 0x113cba9000 512x512x64 x1 levels 1 NVN 0x3d rendered, last used 1 presents ago
[hos host] vk:   64 MB: 0x113755c000 4096x4096x1 x1 levels 1 NVN 0xa, last used 3972 presents ago
[hos host] vk:   64 MB: 0x113755c000 4096x4096x1 x1 levels 1 NVN 0x34 rendered, last used 1 presents ago
[hos host] vk:   54 MB: 0x10005192200 3072x3072x1 x3 levels 1 NVN 0x32 rendered, last used 1 presents ago
[hos host] vk:   47 MB: 0x10b4ad6c00 3072x3072x1 x1 levels 12 NVN 0x7b ASTC, last used 1 presents ago
[hos host] vk: finalized textures the GPU drew once, kept for being made again: 2 (177 KB): 8x8x1 NVN 0x3f: 1 (2 KB); 260x260x1 NVN 0xd: 1 (175 KB);
[hos host] vk: a submit's memory work (ms): maps 0.00, capture taken 0.02, writes collected 0.03
[hos Presentation Thread] vk: presents after the CPU's signal (600): 600 waited for it, 5.04 ms each, 0 let go after RX_VK_FRAME_GATE_MS
[hos host] vk: status: 600 frames, the slowest 19.9 ms; VRAM 5.5 of 14.2 GB
[hos host] vk: program census (play): created 0 (0 with stages; 0 with no present for 100 ms before, 0 in a frame of 100 ms or more; 0.0 ms of the game's threads), freed 0 (0 never used); first uses 0: program ready 0, loaded 0, translated 0, failed 0; pipeline ready 0, from a cache 0, compiled 0, made 0, left out 0, none 0; 0 with no create call seen, 0 whose stages are not the create call's; after the create call: under 16.7 ms 0, under 100 ms 0, under 1 s 0, longer 0; needing work 0 (0 of them 100 ms or more after the create call, 0 created in a load or a long frame); in play: 0 program records loaded (0.0 ms), 814 new graphics pipelines, pipelines from the driver's cache 995, from the start's cache 0, compiled 0; 2957 draws and dispatches left out, 562 waits (2525.5 ms)
[hos host] vk: program loads at create (play): 0 queued, 0 loaded (median 0.00 ms, 99th percentile 0.00 ms, worst 0.00 ms, 0 ms and 0.0 MB read in all); 0 programs (0 of them compute variants without their code, 0 graphics programs no note names), 0 graphics pipelines (0 not in a cache), 0 compute pipelines (0 not in a cache); a first use ran 0 itself and waited for 0 (0.0 ms); 0 pipelines made at create taken by draws; 7015 create calls with nothing kept for their stages, 0 before the index; units 7015 (0 ready, 0 loading), 12472 programs alive; 0 noted graphics pipeline probes deferred to first use
[hos host] vk: program loads at create (play): 0 shader modules let go once their programs' noted pipelines were built, 0 programs' modules made again for a new pipeline state (RX_VK_DROP_LOADED_MODULES=0 keeps them)
[hos host] vk: module identifiers (play): 0 known, 0 learned in this run; 0 compute pipelines made by them at a create call, 0 asked that way were in no cache (their code read, as before)
[hos host] vk: a frame: 56 draws kept their neighbour's descriptors (0 of them through other slots' registrations)
[hos host] vk: a frame: 0.0 draws left out (not ready), 0.00 waits for shaders or pipelines (0.0 ms); translation 3276 ms in all, 0 jobs queued
[hos host] vk: a frame's replay time (ms): submit 7.74, present 0.16, draw 4.85, descriptors 2.00, texture lookups 0.07, uploads 0.00, pipelines 0.17, programs 0.81, commands 1.55, dispatch 0.76, memory sync 0.29, queue submit 0.71, slot wait 0.10
[hos host] vk: a frame: mirrors 74 (760 MB): 5.45 MB synced in 1017.1 copies, 16.92 MB written back
[hos host] vk: 6000 presents, 57.7 fps; a frame: 1599 draws (0 skipped), 145 dispatches, 166 passes, 144 barriers, 6 submits, 0.0 uploads (0.0 MB), 80 ub updates (56 overlays), GPU-bound wait 0.1 ms (recording slots only); 1117 pipelines (97 ms), 661 programs (10 failed), 1659 textures (3805 MB)
[hos host] vk: host time in GPU resource waits: 5.13 ms a frame (fences 0.00, timelines 5.13, idle 0.00, queries 0.00; sum across threads, not frame time; excludes pacing and driver submit/present blocking)
[hos host] vk: a frame: 630.3 full barriers recorded, 301.3 left out (nothing recorded since the one before), 18.2 pass starts ordered in their fragment stages only (draw passes after draw passes)
[hos host] vk: a frame: the game's barriers by their NVN flags: 0x1: 2.0; 0x8: 3.0; 0x9: 1.0; 0x11: 45.5; 0x12: 68.5; 0x13: 2.0; 0x21: 4.0; 0x28: 14.0; 0x92: 3.0; 0x97: 1.0;
[hos host] vk: a frame: indirect mesh draws' arguments (RX_VK_MESH_CONV=2): 41.0 converted one by one, 0.0 from 0.0 batches at the game's barriers (0.0 converted in them; 0.0 dropped after a write, 0.0 the ring came round), 41.0 for the compute fallback (no mesh shaders in use: one by one); scans ended at a barrier 0.0, an op 0.0, the end 0.0, full 0.0; the draws batched by a scan right after an op that ended the one before, by op: none
[hos host] vk: GPU time 13.2 ms a frame (the longest submission 3.9 ms); 0.0 textures made a frame: colour targets 0.0 (0.0 MB), depth 0.0 (0.0 MB), ASTC or BC 0.0 (0.0 MB), other 0.0 (0.0 MB)
[hos host] vk: a frame: 0.00 MB unswizzled on the CPU in 0.00 ms
[hos host] vk: the render thread (600 presents, 17.34 ms a frame): its jobs 8.09 ms a frame (submits 0.63, presents 0.16, others 7.29), idle 9.25; the game waited for room at 0 presents, 0.00 ms each (0.00 a frame; 1.00 jobs queued at a present); 0 catch-ups (a finish, a texture read back, a write-back not recorded yet: 0), 0.00 ms each; 0 presents held for the CPU's signal, 0.00 ms each (0 let go by RX_VK_HOLD_MS); 600 captures at a CPU signal; the submits' waits read in 0.038 ms a frame; readbacks: 3000 next to a texture's storage (no catch-up), 0 in one, 0 read exactly (after the work handed over: among the catch-ups); a present made 7.35 ms after its hand-over (9.11 at most)
[hos host] vk: frame overlap (600 presents): the GPU between recordings, ms a frame, idle + in the next one's uploads: up to a frame's first recording with a draw 0.77 + 0.24 (the window's copy is in the idle), after it 2.49 + 0.62; after a frame's last recording that reads or writes the game's memory come 1.00 recordings (0.02 ms of GPU time), after its last that reads it 1.00 (0.02 ms)
[hos host] vk: frame ahead (600 presents): 7.62 parts a frame recorded in the signal's job, 60.2% of them took nothing that came with the signal; of the others, 3.04 a frame took pages of the signal's capture into a mirror (140.5 pages a frame) and 0.00 pages read as guest memory had them (0.0 pages a frame); a frame's first part took nothing in 0 of 600 frames, its first three in 0
[hos host] vk: frame ahead, what came with the signal: 600 signals with a capture, each of 172.9 pages in 10.0 pools; between a frame's last submit and that signal 0.00 texture and sampler registrations, 0.00 releases and 0.00 other CPU signals a signal (0 of them with any)
[hos host] vk: frame overlap, by place: idle + uploads before a frame's 2nd to 12th recording with a draw, then all later ones: 1.14+0.30, 0.71+0.12, 0.06+0.09, 0.56+0.02, 0.01+0.02, 0.00+0.03, 0.01+0.03, 0.00+0.01, 0.00+0.00, 0.00+0.00, 0.00+0.00, 0.00+0.00
[hos Presentation Thread] vi: real-time step on: ratio 1.005, game's frame rate 60
[hos host] vk: replay reads of guest memory as submitted: 4240000 (1685966 pages from the captures)
[hos host] vk: replay reads of guest memory as submitted: 4260000 (1693622 pages from the captures)
[hos host] vk: recorder: 5550962 packets run in 10.0 s (busy 1.39 s, spun idle 0.12 s, slept 11594 times, 4608 of them under 200 us), 0 calls made directly (no room); 17503 batches over 578 frames (30.3 a frame, 317 packets a batch), 8439 parts ended and submitted by it; the render thread (the replay lock's holder) waited for it 0 times, 0.00 ms (0.0 times, 0.00 ms a frame); every thread 12 times, 1.21 ms in all, the longest 0.39 ms; vkQueueSubmit2 10/1.20 ms (max 0.39), the queue's lock 2/0.02 ms (max 0.02)
[hos host] vk: recorder: 10 waits given up under the queue's lock, the recorder waiting for it
[hos host] vk: recorder: of its busy 1.39 s, 0.26 s in parts' ends and submits (the longest 0.25 ms), 0.00 s of that blocked on the queue's lock (14 times, the longest 0.17 ms), so 0.26 s in the ends, submits and holds; 1.13 s running the other packets; 9604 packets a frame
[hos host] vk: sparse mirrors: 57, 760 MB with memory of 964 MB (203 MB spared); blocks given memory later 923, given back 19752; bindings served from guest memory meanwhile 0 (0 reaching blocks with memory too), from the mirror past every game pool 0; 93 binds (75.8 ms); 0 made whole for want of memory
[hos host] vk: sparse mirrors' buffers: 6 taken from the spares, 68 made new; 17 kept as spares, 0 destroyed; 11 spares now, 308 MB of address space (RX_VK_SPARSE_SPARE_MB=4096)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5724 dirty pages from their submits' captures, 51489 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 907 dispatches, 41808 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7128 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19953 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: imports: 68 live (1148 MB), 0 retired still imported (0 MB), 9 retired given back (256 MB)
[hos host] vk: replay reads of guest memory as submitted: 4280000 (1701282 pages from the captures)
[hos Presentation Thread] vk: capture: a submit 0.09 ms (lock 0.00, scan 0.02 over 1229 ranges, 1865 MB), 218 pages written, 218 copied (0.9 MB)
[hos host] vk: a capture taken (ms): all 0.015; GPU wait for late pages 0.000 (0 of 600 takes), pools lock 0.000, retires and stamps 0.002, bytes copied 0.000; 24 kept (102 MB), 12 spare buffers, 0 texture pages copied (0 MB)
[hos host] vk: replay reads of guest memory as submitted: 4300000 (1708953 pages from the captures)
[hos host] vk: the recording streams: 48 of 1 MB, the most one held 181 KB; 0 made on demand (0 MB kept now); c1 stages: at most 120 a recording; 0 descriptor pools made when a recording's own ran out (0 slots' pools made whole); a recording's pool took at most 153 sets (ub 1695, sb 926, cis 11724, si 0, utb 0, stb 0)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5742 dirty pages from their submits' captures, 51457 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 907 dispatches, 41793 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7152 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19932 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: replay reads of guest memory as submitted: 4320000 (1716623 pages from the captures)
[hos host] vk: program picks: 4839442 kept (2629393 by the last picks' textures, a uniform buffer bound since), 22427 new, 1938130 with other textures (of 6800000)
[hos host] vk: a submit's memory work (ms): maps 0.00, capture taken 0.02, writes collected 0.03
[hos ModuleSystemWorker5] vk: readbacks of GPU results: 980 of 3000 waited for their write-back, 0.04 ms each; 0 looked at the stopped queues first, 0 waited for their submission's last part (guest memory read in place after theirs), 0 waited for their whole submission (their pages written in place, no mirror)
[hos ModuleSystemWorker5] vk: invalidates since the last readbacks line: 3424 kept their virtual core (nothing there to wait for), 2976 gave it up (counted by 256 a thread; RX_VK_INVALIDATE_QUICK=0: none kept)
[hos host] vk: replay reads of guest memory as submitted: 4340000 (1725761 pages from the captures)
[hos ModuleSystemWorker4] nvn: CommandBufferWaitEvent #110000: event 0x100d6c800c mode 0 value 79474 (now 79455) (lr 0x80660e38)
[hos host] vk: a frame: 57 draws kept their neighbour's descriptors (0 of them through other slots' registrations)
[hos host] vk: a frame: 0.0 draws left out (not ready), 0.00 waits for shaders or pipelines (0.0 ms); translation 3276 ms in all, 0 jobs queued
[hos host] vk: a frame's replay time (ms): submit 7.74, present 0.16, draw 4.99, descriptors 2.05, texture lookups 0.07, uploads 0.00, pipelines 0.17, programs 0.82, commands 1.63, dispatch 0.76, memory sync 0.29, queue submit 0.70, slot wait 0.09
[hos host] vk: a frame: mirrors 74 (760 MB): 5.46 MB synced in 1018.0 copies, 16.92 MB written back
[hos host] vk: 6120 presents, 57.2 fps; a frame: 1608 draws (0 skipped), 145 dispatches, 166 passes, 144 barriers, 6 submits, 0.0 uploads (0.0 MB), 80 ub updates (56 overlays), GPU-bound wait 0.1 ms (recording slots only); 1117 pipelines (97 ms), 661 programs (10 failed), 1659 textures (3805 MB)
[hos host] vk: host time in GPU resource waits: 5.01 ms a frame (fences 0.00, timelines 5.01, idle 0.00, queries 0.00; sum across threads, not frame time; excludes pacing and driver submit/present blocking)
[hos host] vk: a frame: 629.6 full barriers recorded, 301.3 left out (nothing recorded since the one before), 18.2 pass starts ordered in their fragment stages only (draw passes after draw passes)
[hos host] vk: a frame: the game's barriers by their NVN flags: 0x1: 2.0; 0x8: 3.0; 0x9: 1.0; 0x11: 45.6; 0x12: 68.5; 0x13: 2.0; 0x21: 4.0; 0x28: 14.0; 0x92: 3.0; 0x97: 1.0;
[hos host] vk: a frame: indirect mesh draws' arguments (RX_VK_MESH_CONV=2): 41.0 converted one by one, 0.0 from 0.0 batches at the game's barriers (0.0 converted in them; 0.0 dropped after a write, 0.0 the ring came round), 41.0 for the compute fallback (no mesh shaders in use: one by one); scans ended at a barrier 0.0, an op 0.0, the end 0.0, full 0.0; the draws batched by a scan right after an op that ended the one before, by op: none
[hos host] vk: GPU time 13.3 ms a frame (the longest submission 3.8 ms); 0.0 textures made a frame: colour targets 0.0 (0.0 MB), depth 0.0 (0.0 MB), ASTC or BC 0.0 (0.0 MB), other 0.0 (0.0 MB)
[hos host] vk: a frame: 0.00 MB unswizzled on the CPU in 0.00 ms
[hos host] vk: replay reads of guest memory as submitted: 4360000 (1734110 pages from the captures)
[hos host] vk: sparse mirrors: 57, 760 MB with memory of 964 MB (203 MB spared); blocks given memory later 923, given back 19752; bindings served from guest memory meanwhile 0 (0 reaching blocks with memory too), from the mirror past every game pool 0; 93 binds (75.8 ms); 0 made whole for want of memory
[hos host] vk: sparse mirrors' buffers: 6 taken from the spares, 68 made new; 17 kept as spares, 0 destroyed; 11 spares now, 308 MB of address space (RX_VK_SPARSE_SPARE_MB=4096)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5885 dirty pages from their submits' captures, 51733 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 909 dispatches, 41964 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7141 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19933 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: imports: 68 live (1148 MB), 0 retired still imported (0 MB), 9 retired given back (256 MB)
[hos sead::MainThread] nvn: CommandBufferSignalEvent #80000: event 0x100d6c803c mode 0 value 79999 (lr 0x80661bfc)
[hos host] vk: late pages found at the replay (600 takes): 0 pages in 0 takes (RX_VK_LATE_AT_REPLAY=0: only the ones the capture found)
[hos sead::MainThread] nvn: GPU waits in 5 s: 288 sync waits (1388 ms), 0 polls timed out (0 ms), 0 queue finishes (0 ms)
[hos Presentation Thread] vk: capture: a submit 0.09 ms (lock 0.00, scan 0.02 over 1229 ranges, 1865 MB), 219 pages written, 219 copied (0.9 MB)
[hos host] vk: a capture taken (ms): all 0.015; GPU wait for late pages 0.000 (0 of 600 takes), pools lock 0.000, retires and stamps 0.002, bytes copied 0.000; 24 kept (102 MB), 12 spare buffers, 0 texture pages copied (0 MB)
[hos host] vk: replay reads of guest memory as submitted: 4380000 (1741764 pages from the captures)
[hos host] vk: readback pages refreshed from guest memory: 1464972 pages so far (0 held: may be written ahead in their submission)
[hos host] vk: replay reads of guest memory as submitted: 4400000 (1749432 pages from the captures)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5742 dirty pages from their submits' captures, 51460 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 907 dispatches, 41809 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7135 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19926 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos ModuleSystemWorker1] nvn: DrawMeshTasksIndirect: 180000 so far (x1 0x117df8b070 x2 0 x3 0x117df8b040 x4 0x44b040)
[hos host] vk: replay reads of guest memory as submitted: 4420000 (1757326 pages from the captures)
[hos host] vk: replay reads of guest memory as submitted: 4440000 (1764950 pages from the captures)
[hos host] vk: a submit's memory work (ms): maps 0.00, capture taken 0.02, writes collected 0.03
[hos host] vk: sparse mirrors: 57, 760 MB with memory of 964 MB (203 MB spared); blocks given memory later 923, given back 19752; bindings served from guest memory meanwhile 0 (0 reaching blocks with memory too), from the mirror past every game pool 0; 93 binds (75.8 ms); 0 made whole for want of memory
[hos host] vk: sparse mirrors' buffers: 6 taken from the spares, 68 made new; 17 kept as spares, 0 destroyed; 11 spares now, 308 MB of address space (RX_VK_SPARSE_SPARE_MB=4096)
[hos host] vk: write-backs left out (600 submits): 944, 3854.3 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5749 dirty pages from their submits' captures, 51497 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 909 dispatches, 41817 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7141 runs, 693.5 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19900 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: imports: 68 live (1148 MB), 0 retired still imported (0 MB), 9 retired given back (256 MB)
[hos host] vk: program picks: 4983205 kept (2708313 by the last picks' textures, a uniform buffer bound since), 22427 new, 1994367 with other textures (of 7000000)
[hos host] vk: replay reads of guest memory as submitted: 4460000 (1772623 pages from the captures)
[hos rmbl::VibrationThread] vcores: 97972 takes, 15107 waits (mean 80 us, longest 6.2 ms, 0 over HOS_VCORE_LOG_MS), 6524 kicks, 6966 yields, 0 starved grants, 0 escapes in 5 s
[hos sead::MainThread] vk: capture: a submit 0.09 ms (lock 0.00, scan 0.02 over 1229 ranges, 1865 MB), 219 pages written, 219 copied (0.9 MB)
[hos host] vk: a capture taken (ms): all 0.014; GPU wait for late pages 0.000 (0 of 600 takes), pools lock 0.000, retires and stamps 0.002, bytes copied 0.000; 25 kept (108 MB), 10 spare buffers, 0 texture pages copied (0 MB)
[hos host] vk: late pages' fix (600 recordings with one): of 600 captures with late pages 600 fixed (131.3 pages, 132.3 fixes a capture), waited as before: 0 no pool, 0 no stamps or shadow yet, 0 stamps too new, 0 whole copies, 0 in place or unknown, 0 a page's earlier fix not recorded yet, 0 other; 47 read on the CPU first, 0 with no table, 0 pages whose shadow changed first; stamps 710 blocks (45440 KB); of the reads on the CPU, 0 waited for an earlier part's fix, 0 copied the page again instead (that part not waited for); the paths: reads on the CPU with the fix done 40, of an earlier recording 7, not recorded yet 0, of this part 0; pages fixed from stamps 77557, from snapshots 1200; the shadows: 292162 pages synced from captures, 291562 made known by it, 600 left unknown because the replay or the collection wrote them since
[hos Presentation Thread] vi: 6234 frames presented (57.5 fps over the last 5.0 s, interval 1)
[hos Presentation Thread] vi: present intervals: <15.5 ms 0, 15.5-17.8 178, 17.8-22 110, 22-30 0, 30-40 0, over 40 0
[hos host] vk: replay reads of guest memory as submitted: 4480000 (1780447 pages from the captures)
[hos ModuleSystemWorker5] nvn: 650000 recordings made, 49094 live copies (16.4 MB), 351 texture copies; 5 command buffers' segments (4.2 MB)
[hos host] vk: render targets drawn over from another start, let go: 294 so far (1651 MB), 206 of them (1223 MB) between the passes
[hos host] vk: a frame: 56 draws kept their neighbour's descriptors (0 of them through other slots' registrations)
[hos host] vk: a frame: 0.0 draws left out (not ready), 0.00 waits for shaders or pipelines (0.0 ms); translation 3276 ms in all, 0 jobs queued
[hos host] vk: a frame's replay time (ms): submit 7.76, present 0.15, draw 4.93, descriptors 2.02, texture lookups 0.06, uploads 0.00, pipelines 0.17, programs 0.81, commands 1.62, dispatch 0.76, memory sync 0.28, queue submit 0.70, slot wait 0.09
[hos host] vk: a frame: mirrors 74 (760 MB): 5.45 MB synced in 1017.9 copies, 16.92 MB written back
[hos host] vk: 6240 presents, 57.6 fps; a frame: 1596 draws (0 skipped), 145 dispatches, 166 passes, 144 barriers, 6 submits, 0.0 uploads (0.0 MB), 80 ub updates (56 overlays), GPU-bound wait 0.1 ms (recording slots only); 1117 pipelines (97 ms), 661 programs (10 failed), 1659 textures (3805 MB)
[hos host] vk: host time in GPU resource waits: 5.11 ms a frame (fences 0.00, timelines 5.11, idle 0.00, queries 0.00; sum across threads, not frame time; excludes pacing and driver submit/present blocking)
[hos host] vk: a frame: 629.4 full barriers recorded, 301.3 left out (nothing recorded since the one before), 18.2 pass starts ordered in their fragment stages only (draw passes after draw passes)
[hos host] vk: a frame: the game's barriers by their NVN flags: 0x1: 2.0; 0x8: 3.0; 0x9: 1.0; 0x11: 45.5; 0x12: 68.5; 0x13: 2.0; 0x21: 4.0; 0x28: 14.0; 0x92: 3.0; 0x97: 1.0;
[hos host] vk: a frame: indirect mesh draws' arguments (RX_VK_MESH_CONV=2): 41.0 converted one by one, 0.0 from 0.0 batches at the game's barriers (0.0 converted in them; 0.0 dropped after a write, 0.0 the ring came round), 41.0 for the compute fallback (no mesh shaders in use: one by one); scans ended at a barrier 0.0, an op 0.0, the end 0.0, full 0.0; the draws batched by a scan right after an op that ended the one before, by op: none
[hos host] vk: GPU time 13.2 ms a frame (the longest submission 3.8 ms); 0.0 textures made a frame: colour targets 0.0 (0.0 MB), depth 0.0 (0.0 MB), ASTC or BC 0.0 (0.0 MB), other 0.0 (0.0 MB)
[hos host] vk: a frame: 0.00 MB unswizzled on the CPU in 0.00 ms
[hos ModuleSystemWorker5] vk: readbacks of GPU results: 980 of 3000 waited for their write-back, 0.04 ms each; 0 looked at the stopped queues first, 0 waited for their submission's last part (guest memory read in place after theirs), 0 waited for their whole submission (their pages written in place, no mirror)
[hos ModuleSystemWorker5] vk: invalidates since the last readbacks line: 3811 kept their virtual core (nothing there to wait for), 2845 gave it up (counted by 256 a thread; RX_VK_INVALIDATE_QUICK=0: none kept)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5745 dirty pages from their submits' captures, 51448 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 907 dispatches, 41827 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7127 runs, 693.6 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19876 in captures, 0 only in the collection, 42 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: replay reads of guest memory as submitted: 4500000 (1788079 pages from the captures)
[hos host] vk: replay reads of guest memory as submitted: 4520000 (1795841 pages from the captures)
[hos host] vk: the recording streams: 48 of 1 MB, the most one held 181 KB; 0 made on demand (0 MB kept now); c1 stages: at most 120 a recording; 0 descriptor pools made when a recording's own ran out (0 slots' pools made whole); a recording's pool took at most 153 sets (ub 1695, sb 926, cis 11724, si 0, utb 0, stb 0)
[hos host] vk: sparse mirrors: 57, 760 MB with memory of 964 MB (203 MB spared); blocks given memory later 923, given back 19752; bindings served from guest memory meanwhile 0 (0 reaching blocks with memory too), from the mirror past every game pool 0; 93 binds (75.8 ms); 0 made whole for want of memory
[hos host] vk: sparse mirrors' buffers: 6 taken from the spares, 68 made new; 17 kept as spares, 0 destroyed; 11 spares now, 308 MB of address space (RX_VK_SPARSE_SPARE_MB=4096)
[hos host] vk: write-backs left out (600 submits): 943, 3852.6 MB; 26 bindings followed, 40 ranges the CPU read, 0 resets; CPU reads or writes of one left out: 0 so far
[hos host] vk: mirror syncs (600 submits): 5761 dirty pages from their submits' captures, 51553 from earlier ones', 0 from guest memory as it was at the replay; 25 captures kept; 0 pages the GPU wrote copied over (0 live); 0 kept for the GPU (RX_VK_PROTECT_BARE)
[hos host] vk: mirror syncs' pass (600 submits): 908 dispatches, 41834 chunks; 0 syncs as copies (no table room)
[hos host] vk: write-back diff (600 submits): 7140 runs, 693.7 MB compared; 41 pages read into the shadows; 260 MB of shadows (124 MB of pages known, in 157 MB of 1 MB parts); pages the shadows knew that the CPU wrote: 19975 in captures, 0 only in the collection, 41 only by the replay itself
[hos host] vk:   read live, by pool:
[hos host] vk: imports: 68 live (1148 MB), 0 retired still imported (0 MB), 9 retired given back (256 MB)
[hos host] control: quit asked by a client
[hos host] vk: program census (close): created 0 (0 with stages; 0 with no present for 100 ms before, 0 in a frame of 100 ms or more; 0.0 ms of the game's threads), freed 0 (0 never used); first uses 0: program ready 0, loaded 0, translated 0, failed 0; pipeline ready 0, from a cache 0, compiled 0, made 0, left out 0, none 0; 0 with no create call seen, 0 whose stages are not the create call's; after the create call: under 16.7 ms 0, under 100 ms 0, under 1 s 0, longer 0; needing work 0 (0 of them 100 ms or more after the create call, 0 created in a load or a long frame); in play: 0 program records loaded (0.0 ms), 814 new graphics pipelines, pipelines from the driver's cache 995, from the start's cache 0, compiled 0; 2957 draws and dispatches left out, 562 waits (2525.5 ms)
[hos host] vk: program loads at create (close): 0 queued, 0 loaded (median 0.00 ms, 99th percentile 0.00 ms, worst 0.00 ms, 0 ms and 0.0 MB read in all); 0 programs (0 of them compute variants without their code, 0 graphics programs no note names), 0 graphics pipelines (0 not in a cache), 0 compute pipelines (0 not in a cache); a first use ran 0 itself and waited for 0 (0.0 ms); 0 pipelines made at create taken by draws; 7015 create calls with nothing kept for their stages, 0 before the index; units 7015 (0 ready, 0 loading), 12472 programs alive; 0 noted graphics pipeline probes deferred to first use
[hos host] vk: program loads at create (close): 0 shader modules let go once their programs' noted pipelines were built, 0 programs' modules made again for a new pipeline state (RX_VK_DROP_LOADED_MODULES=0 keeps them)
[hos host] vk: module identifiers (close): 0 known, 0 learned in this run; 0 compute pipelines made by them at a create call, 0 asked that way were in no cache (their code read, as before)
[hos host] vk: indexed descriptor arrays (close): 40 in 40 of 1305 program stages, 0 of them near a limit
[hos host] vk: indexed descriptor arrays (close), fragment stages: 40 texture (17-64: 40); the longest 64
[hos host] vk: page table census (close): no pool outside its windows, 64-72 GB and 508-512 GB (4524 pools in the table now)
[hos host] control: exiting
[hos host] stop: close work done (a file the game was writing waited for, at most 3 s)