The 16 MiB merges keep up to 256 MiB of idle snapshot chunks for the submissions in flight. Those chunks prefer device
memory and nothing evicts them, so on a 6 or 8 GiB card they took room the mirror and texture eviction defends. Large
cards keep the full 256 MiB.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Ranges just above the old 4 MiB limit were copied back over PCIe at the end of every submission. A larger snapshot pool
keeps the extra merges from allocating on the replay thread. NX2NAVI_WB_MERGE_MAX sets the limit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HtayeqEG2x5woYzWuEQBGV
At Ultra Performance (an 854x480 scene) the Ocarina frame shape took the first single-target pass wider than the scene
as the output pass: a 960x540 post pass comes before the picture's, so DLSS upscaled to 960x540 and that was stretched
to the output. The output pass must now be at least the scene over the mode's share (widescreen.c dlss_share, exposed
as wide_dlss_share), which skips it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The movie decoder thread decoded each AV1 picture and then debanded and converted it before decoding the next. The
deband alone took 28 ms per 1080p frame on one thread and 20 to 27 ms spread over six stripes while the game ran, so the
movie advanced at 31 to 46 frames a second while its separate sound stream played at full speed.
The deband skips the bounds tests away from the plane's edges and picks the centre or the mean without a branch: the
same output, 16 ms on one thread. In the usual case (8-bit, debanded, the game converts the frame) each stripe debands
the decoder's own planes and writes its NV12 rows in one pass, so a frame takes one round of threads instead of two and
no single-threaded copy. A convert thread now does that work while the decoder thread decodes the next picture. The
pacing report adds the time each step takes.
The debanding now runs in stripes straight from the decoder's planes, and the conversion to the picture runs on a
thread of its own with a two-deep queue, so decoding, debanding and converting no longer run one after another.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
plat_frame_capture was a stub on Linux, so the font rasterizer's first setjmp ended the run (setjmp: no host frame). It now walks the stack with libgcc's _Unwind_Backtrace and fills a PlatJmp of its own (glibc's jmp_buf mangles rsp, rbp and rip, so it cannot be filled by hand); plat_jmp_save and plat_jmp_go are small SysV assembly routines like the Win32 ones.
plat_fiber_entry called the fiber function with rsp 8 bytes off the 16-byte alignment the SysV ABI requires, so the first aligned SSE store on a fiber (movaps in ov_poll) faulted as an access violation at address 0. It now aligns rsp before the call.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sixteen builders on a 32-thread CPU spent about eight cores spinning on NVIDIA's driver lock and never finished the list
in a session; the ini text also said a quarter of the CPUs while the code took half.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HtayeqEG2x5woYzWuEQBGV
tools/build_xlate.py (verify_vendor) requires tools/vendor/ryujinx to match the pinned refs checkout byte for byte and
stops with "tools/vendor/ryujinx differs from MNG in 1 files" since the comment after one title id in TitleIDs.cs was
changed. The file is back to the pinned contents; the translator identity stays 4ced1da5bfc7465d, so shader caches
keep their keys.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The Ampere I2F lowering wrote (size | 4) << 10, so the signed flag landed in bit 12, which the Maxwell decoder
does not read (ISrcFmt = (op >> 11) & 4 | (op >> 10) & 3). I2F.F32.S16 and the other signed sources converted
as unsigned: in Jabu-Jabu's Belly the walls' deformation grid of signed 16-bit offsets turned -5 into 65531 and
the room shell into huge stretched triangles (issue 57). The size code stays in bits 10-11, the flag goes to 13,
as I2FP already does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
make_launcher.py and make_kit.py take --prompt-art (folders holding XGamepad, P5Gamepad and Keyboard_Mouse, or those
folders) and copy only the pictures pad_prompts.py reads into the release's prompt-art folder; tools/prompt_art.py
holds the art tables and that list, with no imaging packages. make_game.py's new last step, prompts, runs
pad_prompts.py on that art into the game folder's romfs-patches-* folders. Its key names everything it reads (the
RomFS, the art, the tools, the bind_ lines of nx2navi.ini, the folders being there), so it runs again only when one of
those changed, and it makes no other step run. A kit without art skips it. The launcher's Python gets fonttools and
pillow, and its jobs show a Draw the Button Prompts row when the version has art.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The launcher took the job's weighted fraction as its progress over time. Compile Changed Files carries over half of an
update's weight and can take seconds, so once the shader step began the job read as most of the way done, and the
estimate came out at a fifth of the time spent so far. The step running now is estimated from its progress over the
last minute (the shader step's translation and its slower build into the driver's cache each fill half of its bar),
and the steps still to come from the seconds per weight of the steps already done. On the shader build of a 42,142
shader update the estimate matches the time left.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The preparation translates the game's shaders in their default bind state and builds their stage libraries and
compute pipelines into the driver's own cache. Variants met in play with another bind state are still translated then,
pipelines are still linked in play, and the driver cache is kept per executable and is lost when the driver clears
it. The launcher's step text, the README, make_game.py and docs/launcher.md now say it makes such stops far rarer.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The draw path called getenv("NX2NAVI_SKIP_PROG") for each of ~2,000 draws a frame, and getenv walks the whole environment block: in a replay-thread profile it was a sixth of the samples. That switch, the polygon offset log and the ASTC alpha probe now read the environment once. The replay's cost per draw went from 3.7-5.0 us to 1.2-2.8 us in play, and the mean frame rate by minute from 60-131 fps to 103-243 fps (different places in each run, so the figures are approximate).
cfg_get and cfg_value keep each thread's answers per name until the settings version changes (cfg_set, cfg_reset, cfg_load, and cfg_setenv, which the code that sets NX2NAVI_ variables in the process's own environment now uses); NX2NAVI_CFG_MEMO=0 asks every time.
The texture pool creates at most pool_create_budget (NX2NAVI_POOL_CREATE_BUDGET, default 16, was a fixed 64) new texture entries per refresh. The refresh holds the texture lock throughout, and streaming new textures in made 20-40 ms frames; at 16 the lock was held 5 ms per slow frame instead of 24 in a dense area, and the frames over 25 ms in a heavy minute fell from 111-188 to 89, with no pop-in seen.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
lock_time (NX2NAVI_LOCK_TIME) makes the replay threads time each lock they have to wait for, kept per asking site (platform.h plat_lock_timing, plat_mutex_lock_at and plat_rmutex_lock_at carry the site), and the frame-time summary lists the sites with the most wait, the recorder's full-ring and drain waits, and the texture lock's holds over 1 ms with the code that took it. replay_time now also keeps each frame's replay ops: a frame over 25 ms charged to the replay says how many draws, binds and transfers it replayed, their time, its slowest op, and how much of the replay's wall time the thread was on the CPU. NX2NAVI_PROFILE_REPLAY=1 samples only the replay thread without call chains, and the profiler now follows the thread that replays instead of the one that started the back end.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A census of every NVN2 program in the RomFS shader packages found 31 that failed in the SM86 lowering, none met in play yet. Each form was checked with nvdisasm. MUFU.F16 (the upscaler's colour decode: a half lane widened, the F32 function, rounded back to a half) in 11 upscaler programs. ATOMS.POPC.INC with the X16 index stride, the luminance histograms' three counters per 16-byte bin, in 8 programs. LDG and STG .STRONG.GPU, read as plain accesses, in 2. TLD.LZ.MS, a 2x MSAA resolve that reads samples 0 and 1, in 3. ATOMG.CAS for the foliage programs' slot claims in 4, through Maxwell ATOM.CAS, which the translator only logged as not implemented: patch 0037 emits a global compare-and-swap for it. SUATOM.D on a 1D image, the post-effect append counters, in 2. ISBERD.O, a task program reading back its own payload, in 1. HighestGpr also counts the registers these forms read beyond the base fields.
All 31 now pass spirv-val, and the other 42,146 programs give SPIR-V byte-identical to the previous translator. tensor_lowering_check passes 29 of 29, with four new MUFU.F16 cases checked against a model of the lowering (the warp emulator gains MUFU); a wrong half select fails all four.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
texup_log_ms logs a "texup:" line for each frame whose texture work on the render or replay thread
reaches that many milliseconds. The line splits the time into texture creation, staging, storage
copies, ASTC recording, scratch buffers and new device memory, to show where an area load's hitch goes.
astc_scratch_keep keeps the ASTC decoder's scratch buffer across upload batches instead of making a
new 64 MiB buffer for each batch. astc_arena stages GPU-decoded ASTC blocks in the staging arena
instead of a buffer per texture. Both aim at the allocation churn seen during area loads and stay off
until a play test measures them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
NX2NAVI_AUDIO_STATS=1 logs output underruns, late passes of the audio render thread, waits for the renderer lock,
late audio updates from the game and voices that ran out of queued sound, each with the game frame, to find where
a crackle comes from. NX2NAVI_AUDIO_GAME_PRIO raises the thread that sends RequestUpdate while its update is read
(1) or also keeps it above normal (2). Both are off by default, so the audio path behaves as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
NVIDIA 621.01 reuses a pipeline-cache.bin entry only in the executable that wrote it, so every renamed or updated exe
compiled all of its pipelines again (23 s boots, 17 % read from a 470 MB file). With NX2NAVI_PIPE_BINARIES=1, where the
driver has VK_KHR_pipeline_binary, every pipeline, stage library and compute pipeline is looked up by its driver
pipeline key in pipeline-binaries.bin and made from the stored binaries; anything not stored is compiled once with its
data captured and appended. Each binary is stored once under its own key: nearly every graphics pipeline carries the
same 2 MiB binary, which made a first version of the file 7 GB. pipeline-cache.bin is not used while this is on. The
knob is off by default while it is being tested; NX2NAVI_PIPE_BINARIES_MAX_MB caps the file.
NX2NAVI_ONESHOT_WAIT_MS sets how long a draw into a target the previous frame did not render waits for a pipeline built
for the first time before it is skipped; the default keeps the old 30 s.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The Button prompts row gains Keyboard and mouse (prompts = pc). tools/pad_prompts.py --style pc writes
romfs-patches-pc: every button glyph shows the first key its bindings give it (Ocarina's defaults, with the bind_
lines of the install's nx2navi.ini over them), the mouse buttons as a mouse, the d-pad and the left stick as their
keys laid out as on the keyboard (1 to 4, W A S D), and the camera as the mouse moving. The action prompt shows the
key's label, the ocarina's notes the light keys. A key the art has no picture of is written on a blank key. The pad
styles draw the same bytes as before.
The prompts description says so in each menu language, and the Chinese one no longer splits words across its lines.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The release carried AMD's FSR 3.1 library but not nvngx_dlss.dll, so DLSS ran only where the NVIDIA driver supplied the model itself; elsewhere players had to fetch the DLL by hand. tools/make_launcher.py and tools/make_kit.py now download NVIDIA's own build from the DLSS SDK repository at a pinned tag (tools/dlss_pin.py: 310.9.1, checked against its size and SHA-256, cached), and put it in the release's bin/ with a notice naming the NVIDIA RTX SDKs License that covers it. --dlss FILE takes a copy on disk instead, --no-dlss leaves it out.
tools/make_game.py copies the DLL and its notice beside the game. A different nvngx_dlss.dll the player put there is kept; the one an earlier build placed (named by the notice's sha256 line) is replaced, so a newer release can bring a newer DLSS. The compile step's key covers make_game.py and the kit version, so existing installs get the DLL with their next update.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Plain http goes only to this PC (the tests' servers), a redirect to http is refused, the Authorization header is
dropped when a redirect leaves its host, a download that runs past its size stops at once, and a manifest file entry
needs a plain file name, a size and a sha256.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A create with oldSwapchain retires it even when it fails. The old one, its views and its semaphores now go at once, on
the frame generation paths too, the next present makes a swapchain from nothing, and the failure line is throttled.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With the replay on the submitting thread (the default), vk_texture_release and entry_of destroy textures on the game's
loader threads, where vk_exec_recording_fence is 0: the deferral took the last submitted value, so the image could go
while another thread's submission, or the recorder behind it, still named it. Such deferrals now wait for issued + 1.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A slot folder holds only progress.sav and caption.sav, and Import Saves took those two files for a whole save: it replaced the save and left them at its top, where the game does not read them, so the game started a new one. A folder with progress.sav at its top now replaces that slot alone. A backup made by the import no longer brings its save-tx folder back with it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The game's ACES tonemap shader gets a GLSL function linked into its SPIR-V where it clamps each channel: the game's SDR result upgraded to the HDR range by luminance and mapped to the display with the RenoDRT curve, exposure, highlights, shadows, contrast, saturation, highlight saturation, blowout, flare and hue correction. The settings reach it as four vec4 the support buffer gains at 1440. The present encode is a port of the swap chain pass: gamma correction, gamut compression, the peak clamp and PQ. Without an HDR10 swapchain the shader gives the game's own SDR picture; native HDR is unchanged.
The menu offers Off, RenoDX and Native, with gamma correction, the tone mapper and its grading rows. The shader build compiles sources marked "keep functions" without optimisation so their functions survive for the splice.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The Upscaling row no longer carries "DLSS on or off: after restart". It shows "after restart" only while the upscaler picked cannot start in this run. Frame generation says so only for a change to or from NVIDIA's, since AMD's applies at once, and the HDR row no longer says it twice.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Large compute shaders that no cache holds are built on one background thread after the boot prebuild, smallest first, so their first dispatch no longer compiles them on the replay thread. NX2NAVI_BIG_BUILD_MAX_MB stops that builder once the process has grown by that much: on a 31 GB machine one of these shaders took the process up by 12.8 GB and the game stuttered while Windows paged. The builder takes the queue under its lock and copies each entry out before compiling it, so a shader queued meanwhile cannot free the arrays under it. The same change reports what pipeline-cache.bin gives back, and starts over when the file grows past NX2NAVI_PIPELINE_CACHE_MAX_MB.
A tensor-core program is now tested before its shader is asked for, in both dispatch paths. It runs the stand-in while NX2NAVI_NATIVE_TENSOR is off, and translating it anyway recorded four AI-upscaler shaders of 0.87 to 4.4 million SPIR-V words that this game never dispatches natively. Process memory went from 29.3 GB to 24 GB and the multi-gigabyte builds during play are gone.
The frame-time summary says which part a slow frame went to, and with NX2NAVI_REPLAY_TIME how the replay splits between draws, binds and transfers; NX2NAVI_SVC_SLOW_MS logs the kernel services that block, with the thread they blocked on. prof_report.py reads the map this build writes and adds the image base to a runtime Rip, so a profile resolves to function names instead of question marks.
build_xlate.py applies the patches with GIT_CEILING_DIRECTORIES set to the build folder's parent. Inside a checkout git read 0021-prmt-byte-permute.patch's paths from the repository's top, skipped every file of it and still exited 0, so PRMT was missing from locally built translators and every byte permute was a no-op. A Skipped patch line now fails the build. The note in AmpereLowering.cs that said the translator emits nothing for PRMT is corrected.
Performance: draws reuse the bindings that did not change and record only the dynamic states that differ from what
the command buffer holds; small uploads come from a staging arena; finished submissions are held briefly and handed to
the queue in batches; the recorder thread can keep a CPU core of its own; ASTC block-linear textures are deswizzled on
the GPU; dynamic resolution steps the upscaler mode to hold a frame-time target. Each has a setting in nx2navi.ini
(bind_reuse, dyn_delta, stg_arena, submit_batch, recorder_core, recorder_prio, astc_gpu_deswizzle, dyn_res) and the
staging cache now frees what it holds at shutdown.
Capture: NX2NAVI_DUMP=key saves the frame after each press of Insert, with mesh draws listed by task, mesh and fragment
shader, depth test and face culling, and mesh_prims.txt giving each mesh draw's primitive, clipping and fragment counts.
All of it is off unless NX2NAVI_DUMP is set.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
In a test run with the budget taken as 16 GiB, 113 pools refused for the budget asked again 1,795 times in 6,000 frames, every 16 submissions, though nothing could make room: the world's textures were all in use. Now the idle mirrors go only when together they free what the new mirror needs, and a refused pool asks the textures for room and comes back soon once; after that it asks again every 600 submissions until it gets its mirror.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A test run with the budget taken as 16 GiB showed two faults. The evictions measured the driver's heap use, which keeps counting a VMA block whole while one allocation lives in it, so freeing textures and mirrors in the middle of blocks did not move it: they went on until 5700 of 7232 textures and 395 of 410 mirrors were gone (9.2 GiB free inside blocks). And mirrors evicted under pressure during a menu were refused again when the world came back, so the draws read the pools over PCIe at 4 fps. The evictions now count the free space inside VMA's blocks as room (vk_mem_vram_live), mirrors are no longer evicted to make room for textures, and a mirror that does not fit after the idle mirrors made room asks idle textures for the rest (vk_texture_want_room) and is asked again 16 submissions later.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The pool refresh looks every registered slot up only in submissions that draw through the texture pool. Counted in submissions, a pause screen or a fade that draws nothing through the pool made every texture of the world idle within seconds (the census showed all 3456 decoded textures idle during one such stretch), and under VRAM pressure the eviction would have dropped them all, to be uploaded again when the world came back. The clock now moves only for a submission whose slot checks reached half of the registered pool slots.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A pool whose mirror did not fit the VRAM budget stayed imported for good, however hot, while mirrors nothing had used for minutes kept their memory. Now a new mirror that does not fit takes the room of mirrors no submission synced for 120 submissions, least recently synced first; a pool refused for the budget asks again after 600 submissions; and above the texture pressure threshold (90% of the budget) mirrors idle for 600 submissions go every 8 submissions so the textures get the room. A victim has nothing in the submission being recorded and its GPU writes are already back in guest memory; its device address table pages move back to the imported buffer (vk_gpa_remap_plain) and the mirror is destroyed once the submission is done. It is mirrored again from a whole copy at its next buffer use. NX2NAVI_MIRROR_EVICT=0 turns it off.
Measured with the budget taken as 16 GiB, same route: off, 691 mirrors (910 MiB) and 133 pools refused for the budget; on, 405 mirrors (799 MiB): 240 idle ones (98 MiB, title-screen and loading pools) evicted, 114 pools refused. That exe still evicted idle mirrors under pressure for the textures, which a later commit removes, and asked refused pools again every 16 submissions (1,795 times), which another commit spaces out.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>