Compatibilité des jeux avec Lichii Emulator

Related game issues

• [GBug] Ocarina of Time on i7-8700K (6c/12t) + RTX 3060: ~23-25 fps, GPU ~35-50%, all 12 logical CPUs ~50% (open)
• [Crash] Ocarina of Time (0400486024DEC000): Linux NVIDIA GPU hang, Xid 109 CTX SWITCH TIMEOUT after ~5 min, every time (Windows recovers from the same stall) (open)
• [VBug] macOS/MoltenVK (Apple M1): Ocarina of Time (0400486024DEC000) shows only black output from boot, 1 FPS, only 2 draws per frame (open)
• [Regression] OoT black screen with audio on Linux AMD RADV (RX 9070 XT): Build 83 works, Build 95 fails (open)
• [GBug] No Loading build 95 Shadercache (AMD RX 6600) (open)
• [Crash] Ocarina of Time (0400486024DEC000): VK_ERROR_DEVICE_LOST right after the intro cutscene on RX 7800 XT (build-88) (open)
• Title: Feature request: precompile all of a game's shaders at first boot (shader warm-up) (open)
• [GBug] Ocarina of Time (Switch 2): playable on RX 9070 XT with build 93, ~29 FPS in Kokiri Forest, one thread at 99% (open)
• [Crash] Ocarina of Time freeze, I/O spike to very high value, consume a lot of RAM then crash entire emulator (6700XT) (open)
• Ocarina of Time crashes (unhandled vk::Exception) with VK_ERROR_OUT_OF_DEVICE_MEMORY on an 8 GB RX 9060 XT, after the first cutscene (open)
• [GBug] Ocarina of Time 6fps (open)
• [GBug] Ocarina of Time 0 FPS in Lichii Build 95, but in Build 93 it is playable at 10–25 FPS (closed)
• [Crash] Linux RADV, Navi 33 (Steam Machine): GPU reset from mesh draws in OoT, "Illegal opcode in command stream", also MKW on stock build 88 (open)
• [Feature Request] Make pipeline eviction configurable — EDEN_NX2_PRELOAD_ALL_PIPELINES=1 is undone after 9000 frames (open)
• [VBug] Certain blades of grass and pebbles vanish when attacking with sword. (open)
• [GBug] Cutscenes running in slow motion(OoT) (open)
• [GBug] Ocarina of Time (0400486024DEC000): about half of boots run 20-25% slower; the game's job workers stay awake ~45% longer (open)
• [GBug] Ocarina of Time (0400486024DEC000) on Ryzen 5 5500 + RX 7600: test notes, and movie audio voices that get paused (open)
• Ocarina of Time (Switch 2) gets stuck at "Building 9 shaders" on AMD RX 6750 XT (open)
• [VBug] Ocarina of time Black square lines all over hyrule field (open)
• [Perf] OoT (0400486024DEC000): per-fence Preflushable buffer downloads (~430 MB/s) are the biggest GPU back-thread cost; batching them gives +55% draws/s but breaks exposure (open)
• [VBug] Ocarina of Time: distant Kokiri Forest canopy renders neon green in the main scene pass (RX 7900 XTX) (open)
• [AMD] Ocarina of Time: cyan/magenta line on mountain ridges - FS with no color output writes garbage (fix included) (closed)
• [Crash] Linux/RADV (RX 9070 XT): invalid SPIR-V fragment shader (depth image u32 sampled as f32) segfaults in VkPipelineBuild, Ocarina of Time (closed)
• [VBug] OoT (0400486024DEC000): textures flicker with Low GPU accuracy + EDEN_NX2_SPLIT=2 on AMD RX 9070 XT; High fixes it but turns the split off (~9 fps) (closed)
• [VBug] Ocarina of Time is a black screen with the title theme music, and I can't press any buttons (open)
• [GBug] Ocarina of Time on RADV (RX 9070 XT): 28-35 fps, 35 ms frame with no saturated thread (translation/sync bound) (open)
• [VBug] Flashing textures with GPU mode: FAST (closed)
• [VBug] (open)
• [VBug] Ocarina of Time: holes in the distant mountains of Hyrule Field show the sky through them (open)
• [Crash] (open)
• [Crash] (closed)
• [Build] Tree does not compile with clang-cl on Windows (NOMINMAX and FARPROC casts) (closed)
• [Investigation] OoT build-88: Vulkan pipeline stalls + reproducible hidden-window NVIDIA crash (static findings, test candidates) (open)
• [GBug] Ocarina of Time / Audio Desync (open)
• [Crash] Ocarina of Time Crashes on macOS on game loading (after intro video or selecting a saved file) (open)
• [Crash] AMD Ocarina of time Doesn't Start (open)
• [GBug] OoT (0400486024DEC000): the same ~60 cached graphics pipelines are rebuilt on the GPU thread every boot (closed)
• macOS/MoltenVK: Ocarina of Time (0400486024DEC000) kills or hangs the Metal device at the intro-video boundary (open)
• [GBug] OOT Remake Intel Arc 7 Cores - Only 1 Fps in the main menu and first cutscene (open)
• [VBug] Ocarina of Time: dialogue box panel and hint strip flicker with EDEN_NX2_FRAMES_AHEAD=1, gone with =0 (closed)
• [Crash] Ocarina of Time NUP remains on Launching in build 88 (open)
• [VBug] OoT Slows when booting. (closed)
• [VBug] OOT eventually turns into hard black screen outside of captions in handheld mode, renders as though looking through vaseline in docked mode (open)
• [Crash] Ocarina of time crash on Linux AMD (open)
• Ocarina of Time: frame dump findings (overlay source, bloom/exposure, missing vegetation, rejected gathers) (open)
• Release note: White streaks or trails at night, including around Navi, fireflies, and smoke.
• Release note: AMD black foliage and haze.
• Release note: HUD flicker.
• Release note: Cutscene audio drift.
• Release note: First-launch shader stall on Linux.
• Release note: An NVIDIA device-lost crash was reported on build 93; the release says it is awaiting retest on this build.
• Release note: A user reports approximately 23–25 FPS in Kokiri Forest on Fritchii v4, with CPU and GPU not saturated.
• Release note: On Linux with NVIDIA GPUs, gameplay can lead to a GPU hang after roughly five minutes. Reports include Xid 109 CTX SWITCH TIMEOUT or a GPU stuck at 100% utilization without an Xid.
• Release note: On macOS/MoltenVK with an Apple M1, the Switch 2 NSO version boots and plays audio but presents a black image from startup, with approximately 1 FPS and occasional red/green block artifacts.
• Release note: On Build 95 Non-PGO in Linux AMD RADV, the game remains black-screened while intro audio plays, even after shader compilation. The symptom persisted across the reported Wayland/XWayland and environment-variable tests, and after removing Vulkan Pipeline Cache and Cache Storage.
• Release note: Build 117 renders again, but the user reports substantial FPS and frametime variability, including drops to 19 FPS.
• Release note: On build 117, the game shows a black screen while background music plays and does not load, reported with an AMD RX 6000 series or older GPU on Windows 11.
• Release note: AMD GPU users report Vulkan VK_ERROR_DEVICE_LOST crashes around the intro-to-gameplay transition or while loading a save. Reports include RX 7800 XT, RX 9070 XT, and RX 6900 XT systems; the issue is not universal across all reports or tested builds.
• Release note: Some users report crashes or black screens after trying the EDEN_NX2_NO_MESH=1 workaround; results vary by GPU, operating system, and build.
• Release note: Hard lag or freezes when shaders are being built, mostly on first visits to areas, as reported by the issue author.
• Release note: On build 93 with RX 9070 XT, gameplay is slow-motion at reported frame rates, with heavy bloom and overexposure in bright areas and the Great Deku Tree rendered as a dark silhouette in its intro scene.
• Release note: On RX 5700 (RDNA1), Kokiri Forest performance is reported at 14–15 FPS, compared with 22–24 FPS indoors; build 95 is reported to have the same outdoor performance as build 93.
• Release note: On Lichii-AMD 2026.11.07 with RX 9070 XT using GPU Mode Fast, repeated rolling and jumping triggered black game frames while the HUD remained visible; the game continued running. Blackouts recurred in Kokiri Forest and were not accompanied by log output during the reported black periods.
• Release note: With an older configuration retaining GPU accuracy High, GPU thread splitting did not start, leaving work on one GPU thread. The commenter also reports long hitches while native pipelines were built, even with async shaders on.
• Release note: On RX 9070 XT with extracted base-game version 1.0.0, the commenter reports blackouts and a Native compute shader rejected (per-tap gather) log error. They report no blackouts or such error during about 70 minutes of play using version 2.0.0 from a .nup on build 117 PGO.
• Release note: User reports random freezes after 5–15 minutes of play on Linux with an AMD RX 6000 series or older GPU, followed by I/O spikes up to 6GB/s, rising RAM use, and an out-of-memory emulator crash.
• Release note: Crashes during play with VK_ERROR_OUT_OF_DEVICE_MEMORY on 8 GB AMD GPUs; reports describe unhandled Vulkan exceptions.
• Release note: A user reports performance of up to 6 FPS and a crash at Kokiri Tree after removing the pipeline cache.
• Release note: On Build 95, the user reports sound from the intro but no video and 0 FPS on Linux.
• Release note: A commenter says performance is worse on Build 116 than on Build 93.
• Release note: On Linux RADV systems, user reports describe GPU resets or device loss during mesh draws, sometimes with an illegal opcode in the command stream. EDEN_NX2_NO_MESH=1 avoided resets in several tests but caused missing terrain, plants, or other geometry.
• Release note: With EDEN_NX2_PRELOAD_ALL_PIPELINES=1, pipelines are evicted after 9000 frames, causing runtime builds, long held draws, and a reported freeze.
• Release note: Grass and pebbles disappear or flicker during sword attacks; disabling persistent pipeline cache was reported to make grass disappear permanently, while enabling it left flickering.
• Release note: Cutscenes play in slow motion, lagging behind audio; the report describes them as locked to 49fps with 99% GPU usage.
• Release note: The same slow-motion cutscene problem is reported.
• Release note: The same slow-motion cutscene problem is reported.
• Release note: When fastmem could not be created, guest code ran about 1.4x slower; Kokiri Forest ran at 22.2–26.6 FPS rather than 28.5–33.8 FPS. The report attributes the slow boots to failure to allocate 14336 MiB of backing memory.
• Release note: During movies, including the title-screen background, the six movie audio voices can pause while the picture continues, resulting in late or silent audio. The issue varied between sessions and was also reported on builds without the movie-sync patch.
• Release note: On an AMD RX 6750 XT, the emulator freezes at “Building 9 shaders” after launching; background audio continues and the game does not progress.
• Release note: On an RX 6650 XT setup, asynchronous shader compilation on was associated with a black screen or a freeze while building shaders; asynchronous shader compilation off allowed the game to load and play on that setup.
• Release note: A commenter reports a freeze with background music on an AMD Ryzen 5 5500U with Radeon Graphics, regardless of the asynchronous shader compilation toggle.
• Release note: Black square lines appear across Hyrule Field and disappear when approached.
• Release note: Skipping or batching Preflushable buffer downloads can cause the scene to be overexposed because guest reads may occur before downloaded GPU-written data is available. The issue also reports potentially stale exposure data with the no-preflush experiment.
• Release note: A commenter reports that Kokiri Forest and the Deku Tree can run much more slowly after watching the New Game intro than when loading a save directly; the commenter reports a storage-write extent branch improved the intro-first result.
• Release note: A commenter reports that the per-fence download cost remains and describes an unmerged pull request that limits GPU-written marking to shader store extents.
• Release note: On an AMD RX 7900 XTX, the distant Kokiri Forest canopy renders neon green and cyan.
• Release note: On AMD Radeon RX 9070 XT with driver 2.0.416, a dotted cyan/magenta line appeared along mountain and sky silhouettes in Hyrule Field.
• Release note: On Linux with AMD RX 9000-series graphics using RADV, an invalid SPIR-V fragment shader can cause a segfault during Vulkan graphics-pipeline construction during gameplay.
• Release note: On earlier builds, small details such as leaves, pebbles, and grass flickered or appeared black with Low GPU accuracy and the split GPU thread enabled.
• Release note: A user reports a black screen with title music and unresponsive controls until shaders loaded; after entering gameplay, they reported severe glitches and lag, freezes lasting about 20 seconds during shader building, and a steady 18 FPS.
• Release note: The official PGO artifact of build-80 / nx2-latest crashed in dynarmic with Unhandled SIGSEGV after 2-10 minutes on the reporting machine; another report found a PGO AppImage produced audio but no video output.
• Release note: White flashes were reported before PR #345 in the Deku Tree interior; the sky/lighting colour result from a separate misaligned 64/128-bit SSBO split fix was not verified.
• Release note: In a Windows AMD-driver follow-up on Frizti v3, performance was about 20 fps in a busy view and about 60 fps looking at the floor; the reporter said keeping float16 on improved the busy-view result from about 20 to 24-25 fps.
• Release note: A commenter reports white flickering throughout the 3D world in the Frizti fork.
• Release note: User reports intermittent audio loss, brief frame-rate drops, an excessively bright missing sky, and a missing map.
• Release note: User reported sky-coloured holes in distant mountains of Hyrule Field on nx2 a7a28175 (same content as build-93) + local patches; a later user comment says they could no longer reproduce them on a newer local build.
• Release note: Heavy stuttering during shader compilation, sometimes dropping to 0 FPS, with poor reported performance.
• Release note: On Radeon RX 580 (Polaris / GCN4), the game hangs at 0 FPS with a black screen while building compute pipelines as the intro video starts.
• Release note: A user reported that local ClangCL and MSVC builds stalled with no frames, while the compared build-75 release presented frames. The cause was not determined; a maintainer noted the report came from an old commit and may no longer happen.
• Release note: Two hidden/minimized Windows CLI launches crashed in nvoglv64.dll with exception 0xc0000094 at offset 0xEFC9BB. Static findings identify stale SDL visibility state and no zero-extent guard as possible leads, but neither is proven to cause the crash.
• Release note: Audio and video become desynchronized on the title screen, intro, and initial dream cutscene; a commenter reports slow-motion live-rendered video and an audio lead of approximately 3 seconds.
• Release note: User reports freezing while loading a game and a subsequent emulator crash on macOS; in Build 93, pressing A on a controller reportedly does nothing when attempting to return to the title screen.
• Release note: AMD GPUs may show a black screen or crash the graphics driver after loading, the intro, or a cutscene.
• Release note: On a local nx2 build, white or yellow haze outdoors and black water or foliage after an area change were reported as remaining graphics issues.
• Release note: On build-83, repeated graphics-pipeline builds on the GPU thread across boots cause a visible gameplay hitch.
• Release note: A user reported repeated GPU-thread pipeline builds and tessellation-related warnings in one Linux session; recurrence across launches was not confirmed.
• Release note: On macOS arm64 with Vulkan via MoltenVK, the Metal device may be lost with a GPU address fault or device error during intro video playback, or the GPU thread may hang at the video-to-gameplay transition in fence waiting or buffer download.
• Release note: On Apple GPUs, attachment-less framebuffer draws using a 32768×32768 surface clip can exceed device framebuffer limits and cause GPU failure/device loss. A user also reports a stream-buffer constant-buffer tail read issue on Metal.
• Release note: User reports that rejecting shader programs containing native opcode 0x4a4 avoids a Metal page fault, but may leave the run stalled at the end of the video with semaphore waits and GPU-thread abort.
• Release note: Users report very low frame rates in the main menu and opening cutscene on Intel Arc GPUs; a Linux user also reports a prolonged shader stall on first boot and a frozen Kokiri intro cutscene.
• Release note: Dialogue box background panel and hint strip flicker in about half of presented frames with EDEN_NX2_FRAMES_AHEAD=1; other HUD elements may also flicker.
• Release note: A user reports that the game remains on the Launching screen and does not start in nx2 build 88 on Windows 11.
• Release note: The game runs slowly when booting.
• Release note: Black screens during the introductory cinematic, GDT sequence, or gameplay; reports include intermittent black flashes and audio playing ahead of video.
• Release note: Docked mode may show a black screen except for captions or HUD, while handheld mode may appear excessively bright, low resolution, or vaseline-like.
• Release note: A commenter says many crashes with AMD graphics cards are related to mesh shaders.
• Release note: A user reports a crash when opening the game on Linux with an AMD Radeon AI PRO R9700 (9070XT).
• Release note: The Deku Tree exterior cutscene can render black except for a small circle around Navi, even when the title screen, intro, and earlier cinematics render correctly. A later report also observed a deterministic AMD driver crash while building graphics pipelines in the same scene.
• Release note: A half-resolution layer can produce dark blotches and translucent patches over Link, terrain, and other objects. User reports traced the issue to a depth image being cleared after a depth copy; one report says PR #185 addresses the ordering, but reports do not establish a confirmed released fix.
• Release note: River water may be dark or lack visible lit contribution, and one user reported flickering that depends on the sun's position relative to the camera.
• Release note: Outdoor scenes may remain severely white/yellow and washed out on clean upstream nx2, despite substantial improvements to vegetation and atmosphere.
• Release note: Texture-format and texture-index assertions were reported during gameplay; a separate report associated a period of such assertions with terrain and subtitle flickering, without establishing causation.

Compared with Previous Build

Current: build-93 (4 records); previous: build-88 (1 record).

Related game issues

• [Regression] OoT black screen with audio on Linux AMD RADV (RX 9070 XT): Build 83 works, Build 95 fails (open)
• [Crash] Ocarina of Time (0400486024DEC000): VK_ERROR_DEVICE_LOST right after the intro cutscene on RX 7800 XT (build-88) (open)
• Title: Feature request: precompile all of a game's shaders at first boot (shader warm-up) (open)
• [GBug] Ocarina of Time (Switch 2): playable on RX 9070 XT with build 93, ~29 FPS in Kokiri Forest, one thread at 99% (open)
• Ocarina of Time (Switch 2) gets stuck at "Building 9 shaders" on AMD RX 6750 XT (open)
• [Crash] (open)
• [Crash] (closed)
• [Investigation] OoT build-88: Vulkan pipeline stalls + reproducible hidden-window NVIDIA crash (static findings, test candidates) (open)
• macOS/MoltenVK: Ocarina of Time (0400486024DEC000) kills or hangs the Metal device at the intro-video boundary (open)
• Release note: On Build 95 Non-PGO in Linux AMD RADV, the game remains black-screened while intro audio plays, even after shader compilation. The symptom persisted across the reported Wayland/XWayland and environment-variable tests, and after removing Vulkan Pipeline Cache and Cache Storage.
• Release note: Build 117 renders again, but the user reports substantial FPS and frametime variability, including drops to 19 FPS.
• Release note: AMD GPU users report Vulkan VK_ERROR_DEVICE_LOST crashes around the intro-to-gameplay transition or while loading a save. Reports include RX 7800 XT, RX 9070 XT, and RX 6900 XT systems; the issue is not universal across all reports or tested builds.
• Release note: Some users report crashes or black screens after trying the EDEN_NX2_NO_MESH=1 workaround; results vary by GPU, operating system, and build.
• Release note: Hard lag or freezes when shaders are being built, mostly on first visits to areas, as reported by the issue author.
• Release note: On build 93 with RX 9070 XT, gameplay is slow-motion at reported frame rates, with heavy bloom and overexposure in bright areas and the Great Deku Tree rendered as a dark silhouette in its intro scene.
• Release note: On RX 5700 (RDNA1), Kokiri Forest performance is reported at 14–15 FPS, compared with 22–24 FPS indoors; build 95 is reported to have the same outdoor performance as build 93.
• Release note: On Lichii-AMD 2026.11.07 with RX 9070 XT using GPU Mode Fast, repeated rolling and jumping triggered black game frames while the HUD remained visible; the game continued running. Blackouts recurred in Kokiri Forest and were not accompanied by log output during the reported black periods.
• Release note: With an older configuration retaining GPU accuracy High, GPU thread splitting did not start, leaving work on one GPU thread. The commenter also reports long hitches while native pipelines were built, even with async shaders on.
• Release note: On RX 9070 XT with extracted base-game version 1.0.0, the commenter reports blackouts and a Native compute shader rejected (per-tap gather) log error. They report no blackouts or such error during about 70 minutes of play using version 2.0.0 from a .nup on build 117 PGO.
• Release note: On an AMD RX 6750 XT, the emulator freezes at “Building 9 shaders” after launching; background audio continues and the game does not progress.
• Release note: Heavy stuttering during shader compilation, sometimes dropping to 0 FPS, with poor reported performance.
• Release note: On Radeon RX 580 (Polaris / GCN4), the game hangs at 0 FPS with a black screen while building compute pipelines as the intro video starts.
• Release note: Two hidden/minimized Windows CLI launches crashed in nvoglv64.dll with exception 0xc0000094 at offset 0xEFC9BB. Static findings identify stale SDL visibility state and no zero-extent guard as possible leads, but neither is proven to cause the crash.
• Release note: On macOS arm64 with Vulkan via MoltenVK, the Metal device may be lost with a GPU address fault or device error during intro video playback, or the GPU thread may hang at the video-to-gameplay transition in fence waiting or buffer download.
• Release note: On Apple GPUs, attachment-less framebuffer draws using a 32768×32768 surface clip can exceed device framebuffer limits and cause GPU failure/device loss. A user also reports a stream-buffer constant-buffer tail read issue on Metal.
• Release note: User reports that rejecting shader programs containing native opcode 0x4a4 avoids a Metal page fault, but may leave the run stalled at the end of the video with semaphore waits and GPU-thread abort.

Compared with Previous Build

Related game issues

• [Regression] OoT black screen with audio on Linux AMD RADV (RX 9070 XT): Build 83 works, Build 95 fails (open)
• [Crash] Ocarina of Time (0400486024DEC000): VK_ERROR_DEVICE_LOST right after the intro cutscene on RX 7800 XT (build-88) (open)
• Title: Feature request: precompile all of a game's shaders at first boot (shader warm-up) (open)
• [GBug] Ocarina of Time (Switch 2): playable on RX 9070 XT with build 93, ~29 FPS in Kokiri Forest, one thread at 99% (open)
• Ocarina of Time (Switch 2) gets stuck at "Building 9 shaders" on AMD RX 6750 XT (open)
• [Crash] (open)
• [Crash] (closed)
• [Investigation] OoT build-88: Vulkan pipeline stalls + reproducible hidden-window NVIDIA crash (static findings, test candidates) (open)
• macOS/MoltenVK: Ocarina of Time (0400486024DEC000) kills or hangs the Metal device at the intro-video boundary (open)
• Release note: On Build 95 Non-PGO in Linux AMD RADV, the game remains black-screened while intro audio plays, even after shader compilation. The symptom persisted across the reported Wayland/XWayland and environment-variable tests, and after removing Vulkan Pipeline Cache and Cache Storage.
• Release note: Build 117 renders again, but the user reports substantial FPS and frametime variability, including drops to 19 FPS.
• Release note: AMD GPU users report Vulkan VK_ERROR_DEVICE_LOST crashes around the intro-to-gameplay transition or while loading a save. Reports include RX 7800 XT, RX 9070 XT, and RX 6900 XT systems; the issue is not universal across all reports or tested builds.
• Release note: Some users report crashes or black screens after trying the EDEN_NX2_NO_MESH=1 workaround; results vary by GPU, operating system, and build.
• Release note: Hard lag or freezes when shaders are being built, mostly on first visits to areas, as reported by the issue author.
• Release note: On build 93 with RX 9070 XT, gameplay is slow-motion at reported frame rates, with heavy bloom and overexposure in bright areas and the Great Deku Tree rendered as a dark silhouette in its intro scene.
• Release note: On RX 5700 (RDNA1), Kokiri Forest performance is reported at 14–15 FPS, compared with 22–24 FPS indoors; build 95 is reported to have the same outdoor performance as build 93.
• Release note: On Lichii-AMD 2026.11.07 with RX 9070 XT using GPU Mode Fast, repeated rolling and jumping triggered black game frames while the HUD remained visible; the game continued running. Blackouts recurred in Kokiri Forest and were not accompanied by log output during the reported black periods.
• Release note: With an older configuration retaining GPU accuracy High, GPU thread splitting did not start, leaving work on one GPU thread. The commenter also reports long hitches while native pipelines were built, even with async shaders on.
• Release note: On RX 9070 XT with extracted base-game version 1.0.0, the commenter reports blackouts and a Native compute shader rejected (per-tap gather) log error. They report no blackouts or such error during about 70 minutes of play using version 2.0.0 from a .nup on build 117 PGO.
• Release note: On an AMD RX 6750 XT, the emulator freezes at “Building 9 shaders” after launching; background audio continues and the game does not progress.
• Release note: Heavy stuttering during shader compilation, sometimes dropping to 0 FPS, with poor reported performance.
• Release note: On Radeon RX 580 (Polaris / GCN4), the game hangs at 0 FPS with a black screen while building compute pipelines as the intro video starts.
• Release note: Two hidden/minimized Windows CLI launches crashed in nvoglv64.dll with exception 0xc0000094 at offset 0xEFC9BB. Static findings identify stale SDL visibility state and no zero-extent guard as possible leads, but neither is proven to cause the crash.
• Release note: On macOS arm64 with Vulkan via MoltenVK, the Metal device may be lost with a GPU address fault or device error during intro video playback, or the GPU thread may hang at the video-to-gameplay transition in fence waiting or buffer download.
• Release note: On Apple GPUs, attachment-less framebuffer draws using a 32768×32768 surface clip can exceed device framebuffer limits and cause GPU failure/device loss. A user also reports a stream-buffer constant-buffer tail read issue on Metal.
• Release note: User reports that rejecting shader programs containing native opcode 0x4a4 avoids a Metal page fault, but may leave the run stalled at the end of the video with semaphore waits and GPU-thread abort.

Compared with Previous Build