TotK's compute passes sample a 4x4 R32_SFLOAT texture through a shadow sampler. Vulkan allows depth comparison only
on formats with SAMPLED_IMAGE_DEPTH_COMPARISON, which AMD's drivers do not give R32_SFLOAT, so the result there was
undefined (20 validation errors in 40 s of field play). Such an image now gets a D32_SFLOAT (or D16_UNORM) twin
copied through a buffer right before each dispatch that samples it, and the dispatch samples the twin. Formats the
driver compares as they are keep the colour image. TOTK_DEPTH_TWIN=0 turns it off. Design after Savage Wind Recomp's
depth twins.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
AMD drivers lost the device at boot or at menu opens with the recorder thread; vk_recorder = 0 ended it for three
RX 9070 XT users (issue #5), and RADV saw an illegal opcode at the end of a command buffer (issue #40). It now
defaults to off on vendor 0x1002; TOTK_VK_RECORDER=1 turns it back on.
On AMD's separate driver branch for RDNA 1/2 ("AMD proprietary shader compiler") every device loss came with a cold
pipeline cache and draws waiting on fast-linked pipelines, and the recorder setting did not help there. TOTK_GPL now
has no fixed default: off on that branch, on elsewhere.
VK_AMD_buffer_marker breadcrumbs (vk_crumbs.c) mark the start and end of each submission part, upload batch, ASTC
decode chunk and present; on device lost the log names the last point the GPU reached and the last one it finished.
vk_fsr.c ensure_images held no queue lock around vkDeviceWaitIdle, which needs every queue externally synchronized.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Windows turned the monitor off after its idle timeout during play, since gamepad input does not count as activity (#7).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Above about 63.5 fps the fade step trunc(2.1 * ratio) is 0, so objects that start a fade stay translucent or
invisible until a slower frame moves them on (issues #10, #37). Switching to 60 fps and back cleared it because the
60 fps step is 1. Port of MacNativeGames' high-FPS fade fix as a fade set under TOTK_HIGHFPS_FIX.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
wb_fence_limit (TOTK_WB_FENCE_LIMIT, default 65536) replaces the fixed 64 KiB limit of the write-back done when the CPU waits for a fence. With trace = gfx the log names each dirty target left on the GPU because of its size and each target over 64 KiB that a raised limit wrote back. Meant to test whether the fuse weapon switch bug (#18) and the Purah Pad camera bug (#13, #63) come from a target the game reads on the CPU without an invalidate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
On a 4 GiB GPU the released-texture list (up to 2048 entries, each ASTC texture decoded to RGBA8) filled video memory and the heap spilled into system memory: a Depths lava scene went from 12 ms to 60 ms of GPU time per frame and about 16 fps. Above tex_vram_pressure percent of the device-local budget (default 90, 0 = off), each submit now destroys up to 32 of the oldest released entries.
Only entries that come back intact are dropped: mirrored textures the GPU never wrote, which are uploaded again from their storage. A render target exists only in its image; dropping those made textures come back empty (black flashes in the Depths boss fight, about 170 "left uninitialised" aliases a minute).
Adds the setting to totk.ini and the Advanced tab, and a perf log line with the released count and drops.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The manifest declared no DPI awareness, so Windows ran the process DPI-unaware: on a 4K monitor at 150% scale the
fullscreen window and its swapchain were 2560x1440, the 4K game image was shrunk to it and Windows stretched the
result back up, blurred (issues #25, #56, #31).
The manifest now declares PerMonitorV2 (true/pm for older Windows 10). The window frame is measured at the window's
DPI, a framed window keeps its client size when it moves to a monitor at another scale, WM_DPICHANGED takes the
suggested rect or keeps fullscreen over the monitor, and window_size means screen pixels. The menu's minimum text
size follows the window's DPI, and the text entry dialog is measured at the game window's DPI. --probe-dpi prints
the process awareness and fails unless it is per-monitor v2.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GitHub issue #21 asks for a 90 fps limit. fps_limit = 90 in totk.ini already worked, because frame_rate.c accepts any whole
rate from 30 to 360 and cfg.c does not validate the values column. The menu row and the documented value lists did not
name it. Added 90 to the menu between 60 and 120 (new catalog id UI_MENU_FPS_90 in both languages), to the FPS_LIMIT
row in cfg.c and to totk.ini.example.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Choosing Simplified Chinese in the installer wrote lang = zh-CN to totk.ini and the game showed Japanese: ov_language() handed the code to nn::oe::GetDesiredLanguage unchanged, and TotK only accepts zh-Hans and zh-Hant for Chinese. Reported in #72 and #68.
ov_language() now maps both old spellings, so existing totk.ini files keep working and cfg.c still accepts them. The F1 menu offers one row per language under the canonical codes. The installer lists only zh-Hans and zh-Hant, writes them, and reads zh-CN or zh-TW from the CLI or from a recorded totk-install.json as the canonical code, so a repair or update of an old install does not fail with check.lang_unknown and rewrites the ini. --probe-services prints the configured language and checks the four Chinese codes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
vk_device_fault_report queries vkGetDeviceFaultInfoEXT once per process after VK_ERROR_DEVICE_LOST and logs the
description, the faulting addresses with type and precision, and the vendor codes. The submit, upload, occlusion
query and wait fatals call it before rt_fatal, and the present path calls it on a failed acquire, blit submit or
present. Threads that lose the device together wait until the report is written. The extension counts as usable only
when the driver reports the deviceFault feature. No cost unless the device is lost.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every Switch button and the eight stick directions the keys drive get two key cells and a controller cell under the
button layout. A cell takes the next key, mouse button or controller button pressed; a key in use moves over, a
controller button swaps. The tab writes bind_ lines (new: bind_LSTICK_* / bind_RSTICK_*, key:default, pad:default) and
input_win.c builds the tables again when one changes, so remapping applies while the game runs. More keys can be
bound: Enter, Backspace, right Shift and Ctrl, Alt, the numpad, punctuation, F2-F7, F9, F12, mouse buttons 4 and 5.
Alt alone no longer enters the window menu loop.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The gamepad map was by position only, so on an Xbox pad the button labelled A pressed Switch B. pad_layout = auto
(default) keeps the Nintendo layout for a pad whose east button SDL reports labelled A and uses the Xbox layout (south
A, east B, west X, north Y; cross A on a PlayStation pad) for the others; nintendo and xbox force one. It applies at
once; bind_ lines name physical buttons and apply on top of either layout.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Comments cited codes from local notes (port-lessons P1, backlog), task labels (T6c1-T6c3, M4), bare upstream commit
hashes and issue numbers, a file that no longer exists (nvn_api.c, now nvn_api_*.c), a generated file (main_0103.c)
and line numbers into upstream files that refs/ has moved past, so they pointed at unrelated code. Those now name the
file, function or project and say what the cited code does. Attribution of copied code keeps its pinned commits.
Corrected facts: TOTK_REPLAY_ASYNC defaults to 0, and the shader cache, pipeline cache and pipeline keys live next to
totk.exe, not in build/v121.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A player who starts totk.exe from Explorer or a shortcut had no log to attach to a report: the log went only to stderr.
A game run whose stderr is not a file (a console or a pipe) now also writes logs\totk.log, keeps the four earlier
runs as totk.1.log to totk.4.log, and stops the file at 64 MiB. Runs whose stderr is a file (run_game.py, 2> file) and
the tool modes write no extra file. Lines from before the file opens (mem_init comes first so the CRT heap stays out of
the guest address space) wait in a static buffer.
Crash reports, fatal errors, IPC and SVC diagnostics, guest debug output and the boot: lines now go through
rt_logf / rt_log_raw, so they reach the file too. Open log folder in the menu opens the folder of the file.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Write-watch queries skip frozen 2 MiB regions, but a region was frozen only when it lay wholly in imported pools and
wholly committed. NVN pools are small and not 2 MiB aligned, so most regions also hold other guest memory or reserved
holes: in town 340 of the 436 MiB queried every frame lay in regions nothing had written for 300 submissions. On the
GTX 1050 Ti laptop the queries took up to 2.2 ms a frame, 1.4 ms of it in queries that found no write.
mem_freeze now makes the committed pages of a region read-only when all of them are read-write and leaves reserved
holes alone; thaw_locked already restored such regions run by run. The pool rule is gone: the only kernel writes into
guest memory are file reads, which thaw their buffer first, and every user-mode write faults and thaws as before.
TOTK_FREEZE_IDLE defaults to 60 (in a fight 100 of 257 queried MiB lay in regions idle for 30-299 submissions; a thaw
by a write costs about 60 us on the RX 9070 machine), and import mode freezes too, for the texture checks of APUs.
RX 9070, per frame after load: town 436 -> 96 MiB queried, 0.68 -> 0.38 ms; fight 168 MiB, 0.45 ms (257 MiB, 0.57 ms
with the new rule at 300 submissions), about 20 thaws a second. TOTK_MEMORY=verify in the fight: 3.65 million pages compared, 0 mismatches. The gfx-abi probe
covers a region with reserved holes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
TotK rewrites per-object uniform blocks on other threads soon after it submits the frame that reads them, and the
replay records for milliseconds after the submit call, so late draws could read another object's matrix (phantom
rocks, a flat far field). The guard in nvn_api_core.c obs_release_read only knows 1.2.1's producer. runtime/vk_usnap.c
copies every uniform range of imported pools that the submission binds into an arena, after MacNativeGames'
TOTK_UBO_SNAPSHOT, and ubo_info binds the copy; ranges a submission still in flight writes are left out, and GPU writes
within the submission drop the copies they touch. TOTK_UBO_SNAPSHOT=0 binds live memory; =ab alternates.
Town on an RX 9070 (in-launch A/B, 70 s): 2235 ranges, 1.9 MiB per submission in 0.49 ms; interval +0.02 ms, replay
+0.19 ms. The sort of the ranges was half of the cost with qsort (0.53 ms for about 10,000 bindings) and is a radix
sort now. fight2 interval within noise. The four probes passed on 1.4.3.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The generator emits a guest poll such as `ldr w9, [x19]; cmp w9, #1; b.eq` as a plain C load, and clang hoists it out
of the loop: the lock wait in f_7001fa0e2c compiled to `cmp; jne; jmp .`, so a thread that found the lock held spun
forever after its holder released it. A fight run hung this way in 3 of 13 launches on an RX 9070, with the UBO
snapshot on and off; gdb showed the thread on that `jmp` in every sample. The gate is an opaque call on every
back-edge of the 868 loops tools/spin_waits.py lists, so each pass reloads memory: the exe goes from 182 self-jumps to
148, and 144 of those are the game's own `b .` instructions. 8 of 8 fight runs then passed; tests/diff/func.py with
--patches passed 5470 cases on 878 functions, and the four probes passed on 1.4.3.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The lifted decoder ran its copy loops as chains of small functions. On the island and fight routes TOTK_ZSTD_CHECK=1 decoded every call both ways with no difference; the host took 0.72 s where the guest decoder took 1.99 s. Load time and fps on an RX 9070 and a 7900X did not change: the gain is CPU time on the loading threads, for machines with fewer cores. The caller at main+0x547a04 stays on the guest decoder: it resets a static DCtx with which the game's zstd rejects some frames libzstd decodes, and the game relies on that result.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The FP helpers compared c->fpcr with a cached copy before every operation to resync MXCSR lazily. MXCSR now
follows FPCR.FZ at once: MSR FPCR calls rt_fpcr_sync, and so do rt_call (thread, fiber and host entry into guest
code), fiber resume (FIBER_FLAG_FLOAT_SWITCH fibers keep their own MXCSR) and the differential test harness per
case. op and vop only test FPCR.RMode for the slow path, and their redo check is one ordered-not-equal compare of
|r| with the smallest normal, the same NaN-or-smallest-normal set as before.
Hot guest code shrinks by about a quarter (132 functions holding half the guest cycles, 1.4.3); fight frame time
moves 1-3%, inside the run-to-run spread. lift, simd, func, atomic and flags_check pass on 1.4.3; simd.py fails
without the harness sync.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With TOTK_TRACE=sched, TOTK_CRIT=<lo>-<hi> records every state change and wake of the guest threads and every
present over those frames (runtime/kernel.c crit_add from acct() and wake(), crit_frame at the present) and writes
them to TOTK_CRIT_OUT at the last frame. tools/crit_path.py walks back from each present through the waits to the
thread whose signal ended them and splits the frame interval by thread and state; the sums equal the interval.
A 1.4.3 fight: with the replay on the submitting thread the path is 7.3 ms of replay on the render thread and 4.3 ms
of guest code; with TOTK_REPLAY_ASYNC=1 it is 10.1 ms of guest code through the fork-join of the main thread and five
workers, at the same frame interval.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every flag-setting instruction packed N, Z, C and V into c->nzcv and every reader unpacked them. recomp/flags.py
now runs a reaching-definitions pass per function: a compare stores its operands in a local Flags F, and a reader
whose definitions are all of one kind computes its condition from them, so cmp and b.cond become one host compare.
CCMP, FCMP, MSR NZCV and ADCS write the packed form, which readers reached by several kinds also use. Calls and
returns publish nothing (AAPCS64 neither passes nor preserves NZCV); tail branches, falls into the next function and
br through rt_dispatch publish c->nzcv, since discovery can start a function inside another one. Functions with an
unresolved computed br keep flags in c->nzcv.
1.4.3 main: 492k direct and 89k packed flag reads, .text 342 -> 299 MiB. Fight bench (bench-fight2-2026-10-27b,
frames 5400-7500, uncapped) 13.15 and 13.63 ms vs 13.9 before. tests/flags_check.py checks the new helpers against
the packed ones; tests/diff/func.py no longer compares NZCV, which generated code leaves untouched.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
memo_match compared the vertex and fragment stages' 64 texture and 64 image handles on every draw. Every OP_BIND_TEX and now every OP_BIND_IMG bumps tex_serial; a key whose serial is current has the same handles, and a key that matched by the compare takes the current serial. memo_match went from 0.95% to 0.27% of the fight replay.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A fight frame of about 2,000 draws made 15 parts on the 7900X at 64 draws a part, each split costing the render thread a submit, a new command buffer and a mirror prologue (vk_sync_split, 7.6% of the replay). maybe_split now also waits until TOTK_SPLIT_US (1000, 0 off) has passed since the part began, so a frame makes 7-9 parts there. A millisecond of recording is well under a millisecond of GPU work, so the GPU still starts early in the frame.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
vk_exec_pass_begin created every attachment view again in each submission and freed them at its fence: 2% of the fight replay on the 7900X, more on drivers with a slower vkCreateImageView. view_of now takes the view from vk_exec_target_view, which keeps it in the sampled-view tables under the texture lock, and vk_exec_image_gone releases it with the image. Images that are not registered textures keep the per-submission views.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>