Compatibilité des jeux avec Lichii Emulator

Related game issues

• [GBug] Black Screen In AMD - Zelda Ocarina Of Time (open)
• [Crash] infinite launch screen with legend of zelda ocarina of time (open)
• Release note: Black screen after boot; some reports say sound and the title/logo work.
• Release note: On build-83, one user reports a solid-green displayed image although a screen capture shows the game rendering correctly.
• Release note: On build 83, a user reports device loss at gameplay start with VK_ERROR_DEVICE_LOST.

Compared with Previous Build

Related game issues

• macOS/MoltenVK: Ocarina of Time (0400486024DEC000) kills or hangs the Metal device at the intro-video boundary (open)
• [Crash] (open)
• [Crash] AMD Crash/AudioDesync (open)
• [AMD RX 9070 XT] Black screen with UI visible and GPU driver timeout (Build 88) (open)
• [Crash] Ocarina of Time (0400486024DEC000): freezes in Hyrule Castle Town pot room near the guard (open)
• [Crash] AMD Ocarina of time Doesn't Start (open)
• [GBug] Ocarina of Time (0400486024DEC000): intro freezes for 6 to 18 minutes on AMD while the driver optimizes two compute shaders (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] OoT Slows when booting. (open)
• Ocarina of Time harvest fork: findings and open tasks (open)
• [Crash] Linux/RADV (RX 9070 XT): invalid SPIR-V fragment shader (depth image u32 sampled as f32) segfaults in VkPipelineBuild, Ocarina of Time (open)
• [Bug] Windows build-83 can hang on Closing software after OoT gameplay (open)
• [GBug] OoT (0400486024DEC000): the same ~60 cached graphics pipelines are rebuilt on the GPU thread every boot (open)
• [VBug] Oot is in a "nightmare mode"? also gameplay speed is at like 30% (open)
• [GBug] Ocarina of Time on RADV (RX 9070 XT): 28-35 fps, 35 ms frame with no saturated thread (translation/sync bound) (open)
• [GBug] Zelda: Ocarina of Time (0400486024DEC000): the audio plays a different part of the stream than the picture (open)
• [VBug] [nx2-build-75] Zelda: Ocarina of Time (0400486024DEC000) black screen — no AV1 decoder (open)
• [VBug] OoT (0400486024DEC000): world renders black except areas lit by Navi and around Link, AMD RX 9070 (open)
• Release note: Outdoor areas have far too much fog and glow compared with console; performance is about 12–22 fps on a Ryzen 5900X with an RTX 3080, and three rare shaders remain unsupported.
• Release note: On macOS arm64 through MoltenVK, the Metal device may be lost or the GPU thread may hang at the intro-video/gameplay boundary. A commenter attributes the device loss to an attachment-less framebuffer reaching 32768×32768, above the stated Apple GPU limit of 16384.
• Release note: After a local workaround for device loss, a commenter reports that the 3D scene remains black while sound and subtitles work on an older Eglantine fork; current nx2 behavior was not yet checked.
• Release note: User reports a crash when opening the game on Arch Linux with an AMD Radeon AI PRO R9700 (9070XT); the console reports a segmentation fault.
• Release note: User reports audio/video desynchronization during the opening cutscene and a system freeze at the Deku Tree cutscene on Vulkan with an AMD RX 9000 series GPU. Comments report crashes on AMD GPUs, including while building shaders. One commenter reports that OpenGL GLSL stops the crashes but leaves only the UI visible.
• Release note: On some AMD GPUs, the UI may remain visible while the 3D world is black; reports also describe AMD driver timeouts or crashes.
• Release note: User-reported reproducible freeze when walking deeper into the pot room in Hyrule Castle Town, near the guard.
• Release note: The issue reporter using an AMD RX 6600 8GB on Windows 10 reports that the emulator starts loading and then remains on a black screen.
• Release note: Commenters report rendering failures, black screens, and AMD driver crashes; one reports that audio continues while rendering is crashed.
• Release note: On AMD's proprietary Windows driver, two compute pipelines can take 6 to 18 minutes to compile during the intro, leaving the game at 0 FPS while audio continues. Rejecting the shader hashes avoids the wait but causes black areas in the intro, title, and castle scenes.
• Release note: In the measurement-only experiment, disabling per-fence Preflushable downloads or batching downloads by 64 fences caused overexposure in Kokiri Forest.
• Release note: A commenter reports that an experimental compute storage write-extent analysis reduced marked GPU-written data and fence copies while preserving correct auto-exposure. It was described as code on a user's branch, not as an upstream or validated release fix.
• Release note: Static review and a separate research comment describe read-trap bookkeeping and dirty-state issues that can leave stale GPU-written data when per-fence downloads are disabled; proposed corrections and batching remain untested or local experiments.
• Release note: Slows when booting.
• Release note: Community comments describe AMD black or corrupted-world output, fragment-pipeline compiler crashes on Windows, later timeouts or TFB stalls, corrupted output, and a RADV depth-image SPIR-V failure. These are separate reported cases; the comments do not establish a universal cause or resolution.
• Release note: User reports a segfault during gameplay on Linux with an AMD RX 9000-series GPU and RADV, when a graphics pipeline is built for an invalid fragment shader.
• Release note: On Windows build-83, shutdown can hang at “Closing software…” after gameplay. Some sessions completed shutdown successfully; the cause is undetermined.
• Release note: Persistent graphics pipeline cache entries are reportedly rebuilt on the GPU thread during gameplay on repeated boots, causing a visible hitch.
• Release note: With GPU Mode: Accurate (High), gameplay may run in slow motion at about 30 FPS; shader compilation can cause stutter on new content.
• Release note: PGO artifact crashes in dynarmic with Unhandled SIGSEGV after 2–10 minutes on the reported machine.
• Release note: Clean audio from a different point of the game plays while the intro image is displayed; audio and picture are substantially out of sync.
• Release note: In gameplay, the world renders almost entirely black, with only the HUD and some areas near light sources visible; some shapes appear blown out with heavy bloom.

Compared with Previous Build

Related game issues

• macOS/MoltenVK: Ocarina of Time (0400486024DEC000) kills or hangs the Metal device at the intro-video boundary (open)
• [Crash] (open)
• [Crash] AMD Crash/AudioDesync (open)
• [AMD RX 9070 XT] Black screen with UI visible and GPU driver timeout (Build 88) (open)
• [Crash] Ocarina of Time (0400486024DEC000): freezes in Hyrule Castle Town pot room near the guard (open)
• [Crash] AMD Ocarina of time Doesn't Start (open)
• [GBug] Ocarina of Time (0400486024DEC000): intro freezes for 6 to 18 minutes on AMD while the driver optimizes two compute shaders (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] OoT Slows when booting. (open)
• Ocarina of Time harvest fork: findings and open tasks (open)
• [Crash] Linux/RADV (RX 9070 XT): invalid SPIR-V fragment shader (depth image u32 sampled as f32) segfaults in VkPipelineBuild, Ocarina of Time (open)
• [Bug] Windows build-83 can hang on Closing software after OoT gameplay (open)
• [GBug] OoT (0400486024DEC000): the same ~60 cached graphics pipelines are rebuilt on the GPU thread every boot (open)
• [VBug] Oot is in a "nightmare mode"? also gameplay speed is at like 30% (open)
• [GBug] Ocarina of Time on RADV (RX 9070 XT): 28-35 fps, 35 ms frame with no saturated thread (translation/sync bound) (open)
• [GBug] Zelda: Ocarina of Time (0400486024DEC000): the audio plays a different part of the stream than the picture (open)
• [VBug] [nx2-build-75] Zelda: Ocarina of Time (0400486024DEC000) black screen — no AV1 decoder (open)
• [VBug] OoT (0400486024DEC000): world renders black except areas lit by Navi and around Link, AMD RX 9070 (open)
• Release note: Outdoor areas have far too much fog and glow compared with console; performance is about 12–22 fps on a Ryzen 5900X with an RTX 3080, and three rare shaders remain unsupported.
• Release note: On macOS arm64 through MoltenVK, the Metal device may be lost or the GPU thread may hang at the intro-video/gameplay boundary. A commenter attributes the device loss to an attachment-less framebuffer reaching 32768×32768, above the stated Apple GPU limit of 16384.
• Release note: After a local workaround for device loss, a commenter reports that the 3D scene remains black while sound and subtitles work on an older Eglantine fork; current nx2 behavior was not yet checked.
• Release note: User reports a crash when opening the game on Arch Linux with an AMD Radeon AI PRO R9700 (9070XT); the console reports a segmentation fault.
• Release note: User reports audio/video desynchronization during the opening cutscene and a system freeze at the Deku Tree cutscene on Vulkan with an AMD RX 9000 series GPU. Comments report crashes on AMD GPUs, including while building shaders. One commenter reports that OpenGL GLSL stops the crashes but leaves only the UI visible.
• Release note: On some AMD GPUs, the UI may remain visible while the 3D world is black; reports also describe AMD driver timeouts or crashes.
• Release note: User-reported reproducible freeze when walking deeper into the pot room in Hyrule Castle Town, near the guard.
• Release note: The issue reporter using an AMD RX 6600 8GB on Windows 10 reports that the emulator starts loading and then remains on a black screen.
• Release note: Commenters report rendering failures, black screens, and AMD driver crashes; one reports that audio continues while rendering is crashed.
• Release note: On AMD's proprietary Windows driver, two compute pipelines can take 6 to 18 minutes to compile during the intro, leaving the game at 0 FPS while audio continues. Rejecting the shader hashes avoids the wait but causes black areas in the intro, title, and castle scenes.
• Release note: In the measurement-only experiment, disabling per-fence Preflushable downloads or batching downloads by 64 fences caused overexposure in Kokiri Forest.
• Release note: A commenter reports that an experimental compute storage write-extent analysis reduced marked GPU-written data and fence copies while preserving correct auto-exposure. It was described as code on a user's branch, not as an upstream or validated release fix.
• Release note: Static review and a separate research comment describe read-trap bookkeeping and dirty-state issues that can leave stale GPU-written data when per-fence downloads are disabled; proposed corrections and batching remain untested or local experiments.
• Release note: Slows when booting.
• Release note: Community comments describe AMD black or corrupted-world output, fragment-pipeline compiler crashes on Windows, later timeouts or TFB stalls, corrupted output, and a RADV depth-image SPIR-V failure. These are separate reported cases; the comments do not establish a universal cause or resolution.
• Release note: User reports a segfault during gameplay on Linux with an AMD RX 9000-series GPU and RADV, when a graphics pipeline is built for an invalid fragment shader.
• Release note: On Windows build-83, shutdown can hang at “Closing software…” after gameplay. Some sessions completed shutdown successfully; the cause is undetermined.
• Release note: Persistent graphics pipeline cache entries are reportedly rebuilt on the GPU thread during gameplay on repeated boots, causing a visible hitch.
• Release note: With GPU Mode: Accurate (High), gameplay may run in slow motion at about 30 FPS; shader compilation can cause stutter on new content.
• Release note: PGO artifact crashes in dynarmic with Unhandled SIGSEGV after 2–10 minutes on the reported machine.
• Release note: Clean audio from a different point of the game plays while the intro image is displayed; audio and picture are substantially out of sync.
• Release note: In gameplay, the world renders almost entirely black, with only the HUD and some areas near light sources visible; some shapes appear blown out with heavy bloom.

Compared with Previous Build

Current: build-83 (1 record); previous: build-75 (0 records).

Related game issues

• macOS/MoltenVK: Ocarina of Time (0400486024DEC000) kills or hangs the Metal device at the intro-video boundary (open)
• [Crash] (open)
• [Crash] AMD Crash/AudioDesync (open)
• [AMD RX 9070 XT] Black screen with UI visible and GPU driver timeout (Build 88) (open)
• [Crash] Ocarina of Time (0400486024DEC000): freezes in Hyrule Castle Town pot room near the guard (open)
• [Crash] AMD Ocarina of time Doesn't Start (open)
• [GBug] Ocarina of Time (0400486024DEC000): intro freezes for 6 to 18 minutes on AMD while the driver optimizes two compute shaders (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] OoT Slows when booting. (open)
• Ocarina of Time harvest fork: findings and open tasks (open)
• [Crash] Linux/RADV (RX 9070 XT): invalid SPIR-V fragment shader (depth image u32 sampled as f32) segfaults in VkPipelineBuild, Ocarina of Time (open)
• [Bug] Windows build-83 can hang on Closing software after OoT gameplay (open)
• [GBug] OoT (0400486024DEC000): the same ~60 cached graphics pipelines are rebuilt on the GPU thread every boot (open)
• [VBug] Oot is in a "nightmare mode"? also gameplay speed is at like 30% (open)
• [GBug] Ocarina of Time on RADV (RX 9070 XT): 28-35 fps, 35 ms frame with no saturated thread (translation/sync bound) (open)
• [GBug] Zelda: Ocarina of Time (0400486024DEC000): the audio plays a different part of the stream than the picture (open)
• [VBug] [nx2-build-75] Zelda: Ocarina of Time (0400486024DEC000) black screen — no AV1 decoder (open)
• [VBug] OoT (0400486024DEC000): world renders black except areas lit by Navi and around Link, AMD RX 9070 (open)
• Release note: Outdoor areas have far too much fog and glow compared with console; performance is about 12–22 fps on a Ryzen 5900X with an RTX 3080, and three rare shaders remain unsupported.
• Release note: On macOS arm64 through MoltenVK, the Metal device may be lost or the GPU thread may hang at the intro-video/gameplay boundary. A commenter attributes the device loss to an attachment-less framebuffer reaching 32768×32768, above the stated Apple GPU limit of 16384.
• Release note: After a local workaround for device loss, a commenter reports that the 3D scene remains black while sound and subtitles work on an older Eglantine fork; current nx2 behavior was not yet checked.
• Release note: User reports a crash when opening the game on Arch Linux with an AMD Radeon AI PRO R9700 (9070XT); the console reports a segmentation fault.
• Release note: User reports audio/video desynchronization during the opening cutscene and a system freeze at the Deku Tree cutscene on Vulkan with an AMD RX 9000 series GPU. Comments report crashes on AMD GPUs, including while building shaders. One commenter reports that OpenGL GLSL stops the crashes but leaves only the UI visible.
• Release note: On some AMD GPUs, the UI may remain visible while the 3D world is black; reports also describe AMD driver timeouts or crashes.
• Release note: User-reported reproducible freeze when walking deeper into the pot room in Hyrule Castle Town, near the guard.
• Release note: The issue reporter using an AMD RX 6600 8GB on Windows 10 reports that the emulator starts loading and then remains on a black screen.
• Release note: Commenters report rendering failures, black screens, and AMD driver crashes; one reports that audio continues while rendering is crashed.
• Release note: On AMD's proprietary Windows driver, two compute pipelines can take 6 to 18 minutes to compile during the intro, leaving the game at 0 FPS while audio continues. Rejecting the shader hashes avoids the wait but causes black areas in the intro, title, and castle scenes.
• Release note: In the measurement-only experiment, disabling per-fence Preflushable downloads or batching downloads by 64 fences caused overexposure in Kokiri Forest.
• Release note: A commenter reports that an experimental compute storage write-extent analysis reduced marked GPU-written data and fence copies while preserving correct auto-exposure. It was described as code on a user's branch, not as an upstream or validated release fix.
• Release note: Static review and a separate research comment describe read-trap bookkeeping and dirty-state issues that can leave stale GPU-written data when per-fence downloads are disabled; proposed corrections and batching remain untested or local experiments.
• Release note: Slows when booting.
• Release note: Community comments describe AMD black or corrupted-world output, fragment-pipeline compiler crashes on Windows, later timeouts or TFB stalls, corrupted output, and a RADV depth-image SPIR-V failure. These are separate reported cases; the comments do not establish a universal cause or resolution.
• Release note: User reports a segfault during gameplay on Linux with an AMD RX 9000-series GPU and RADV, when a graphics pipeline is built for an invalid fragment shader.
• Release note: On Windows build-83, shutdown can hang at “Closing software…” after gameplay. Some sessions completed shutdown successfully; the cause is undetermined.
• Release note: Persistent graphics pipeline cache entries are reportedly rebuilt on the GPU thread during gameplay on repeated boots, causing a visible hitch.
• Release note: With GPU Mode: Accurate (High), gameplay may run in slow motion at about 30 FPS; shader compilation can cause stutter on new content.
• Release note: PGO artifact crashes in dynarmic with Unhandled SIGSEGV after 2–10 minutes on the reported machine.
• Release note: Clean audio from a different point of the game plays while the intro image is displayed; audio and picture are substantially out of sync.
• Release note: In gameplay, the world renders almost entirely black, with only the HUD and some areas near light sources visible; some shapes appear blown out with heavy bloom.

Compared with Previous Build

Current: build-83 (1 record); previous: build-75 (0 records).

Related game issues

• macOS/MoltenVK: Ocarina of Time (0400486024DEC000) kills or hangs the Metal device at the intro-video boundary (open)
• [Crash] (open)
• [Crash] AMD Crash/AudioDesync (open)
• [AMD RX 9070 XT] Black screen with UI visible and GPU driver timeout (Build 88) (open)
• [Crash] Ocarina of Time (0400486024DEC000): freezes in Hyrule Castle Town pot room near the guard (open)
• [Crash] AMD Ocarina of time Doesn't Start (open)
• [GBug] Ocarina of Time (0400486024DEC000): intro freezes for 6 to 18 minutes on AMD while the driver optimizes two compute shaders (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] OoT Slows when booting. (open)
• Ocarina of Time harvest fork: findings and open tasks (open)
• [Crash] Linux/RADV (RX 9070 XT): invalid SPIR-V fragment shader (depth image u32 sampled as f32) segfaults in VkPipelineBuild, Ocarina of Time (open)
• [Bug] Windows build-83 can hang on Closing software after OoT gameplay (open)
• [GBug] OoT (0400486024DEC000): the same ~60 cached graphics pipelines are rebuilt on the GPU thread every boot (open)
• [VBug] Oot is in a "nightmare mode"? also gameplay speed is at like 30% (open)
• [GBug] Ocarina of Time on RADV (RX 9070 XT): 28-35 fps, 35 ms frame with no saturated thread (translation/sync bound) (open)
• [GBug] Zelda: Ocarina of Time (0400486024DEC000): the audio plays a different part of the stream than the picture (open)
• [VBug] [nx2-build-75] Zelda: Ocarina of Time (0400486024DEC000) black screen — no AV1 decoder (open)
• [VBug] OoT (0400486024DEC000): world renders black except areas lit by Navi and around Link, AMD RX 9070 (open)
• Release note: Outdoor areas have far too much fog and glow compared with console; performance is about 12–22 fps on a Ryzen 5900X with an RTX 3080, and three rare shaders remain unsupported.
• Release note: On macOS arm64 through MoltenVK, the Metal device may be lost or the GPU thread may hang at the intro-video/gameplay boundary. A commenter attributes the device loss to an attachment-less framebuffer reaching 32768×32768, above the stated Apple GPU limit of 16384.
• Release note: After a local workaround for device loss, a commenter reports that the 3D scene remains black while sound and subtitles work on an older Eglantine fork; current nx2 behavior was not yet checked.
• Release note: User reports a crash when opening the game on Arch Linux with an AMD Radeon AI PRO R9700 (9070XT); the console reports a segmentation fault.
• Release note: User reports audio/video desynchronization during the opening cutscene and a system freeze at the Deku Tree cutscene on Vulkan with an AMD RX 9000 series GPU. Comments report crashes on AMD GPUs, including while building shaders. One commenter reports that OpenGL GLSL stops the crashes but leaves only the UI visible.
• Release note: On some AMD GPUs, the UI may remain visible while the 3D world is black; reports also describe AMD driver timeouts or crashes.
• Release note: User-reported reproducible freeze when walking deeper into the pot room in Hyrule Castle Town, near the guard.
• Release note: The issue reporter using an AMD RX 6600 8GB on Windows 10 reports that the emulator starts loading and then remains on a black screen.
• Release note: Commenters report rendering failures, black screens, and AMD driver crashes; one reports that audio continues while rendering is crashed.
• Release note: On AMD's proprietary Windows driver, two compute pipelines can take 6 to 18 minutes to compile during the intro, leaving the game at 0 FPS while audio continues. Rejecting the shader hashes avoids the wait but causes black areas in the intro, title, and castle scenes.
• Release note: In the measurement-only experiment, disabling per-fence Preflushable downloads or batching downloads by 64 fences caused overexposure in Kokiri Forest.
• Release note: A commenter reports that an experimental compute storage write-extent analysis reduced marked GPU-written data and fence copies while preserving correct auto-exposure. It was described as code on a user's branch, not as an upstream or validated release fix.
• Release note: Static review and a separate research comment describe read-trap bookkeeping and dirty-state issues that can leave stale GPU-written data when per-fence downloads are disabled; proposed corrections and batching remain untested or local experiments.
• Release note: Slows when booting.
• Release note: Community comments describe AMD black or corrupted-world output, fragment-pipeline compiler crashes on Windows, later timeouts or TFB stalls, corrupted output, and a RADV depth-image SPIR-V failure. These are separate reported cases; the comments do not establish a universal cause or resolution.
• Release note: User reports a segfault during gameplay on Linux with an AMD RX 9000-series GPU and RADV, when a graphics pipeline is built for an invalid fragment shader.
• Release note: On Windows build-83, shutdown can hang at “Closing software…” after gameplay. Some sessions completed shutdown successfully; the cause is undetermined.
• Release note: Persistent graphics pipeline cache entries are reportedly rebuilt on the GPU thread during gameplay on repeated boots, causing a visible hitch.
• Release note: With GPU Mode: Accurate (High), gameplay may run in slow motion at about 30 FPS; shader compilation can cause stutter on new content.
• Release note: PGO artifact crashes in dynarmic with Unhandled SIGSEGV after 2–10 minutes on the reported machine.
• Release note: Clean audio from a different point of the game plays while the intro image is displayed; audio and picture are substantially out of sync.
• Release note: In gameplay, the world renders almost entirely black, with only the HUD and some areas near light sources visible; some shapes appear blown out with heavy bloom.

Compared with Previous Build

Related game issues

• macOS/MoltenVK: Ocarina of Time (0400486024DEC000) kills or hangs the Metal device at the intro-video boundary (open)
• [Crash] (open)
• [Crash] AMD Crash/AudioDesync (open)
• [AMD RX 9070 XT] Black screen with UI visible and GPU driver timeout (Build 88) (open)
• [Crash] Ocarina of Time (0400486024DEC000): freezes in Hyrule Castle Town pot room near the guard (open)
• [Crash] AMD Ocarina of time Doesn't Start (open)
• [GBug] Ocarina of Time (0400486024DEC000): intro freezes for 6 to 18 minutes on AMD while the driver optimizes two compute shaders (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] OoT Slows when booting. (open)
• Ocarina of Time harvest fork: findings and open tasks (open)
• [Crash] Linux/RADV (RX 9070 XT): invalid SPIR-V fragment shader (depth image u32 sampled as f32) segfaults in VkPipelineBuild, Ocarina of Time (open)
• [Bug] Windows build-83 can hang on Closing software after OoT gameplay (open)
• [GBug] OoT (0400486024DEC000): the same ~60 cached graphics pipelines are rebuilt on the GPU thread every boot (open)
• [VBug] Oot is in a "nightmare mode"? also gameplay speed is at like 30% (open)
• [GBug] Ocarina of Time on RADV (RX 9070 XT): 28-35 fps, 35 ms frame with no saturated thread (translation/sync bound) (open)
• [GBug] Zelda: Ocarina of Time (0400486024DEC000): the audio plays a different part of the stream than the picture (open)
• [VBug] [nx2-build-75] Zelda: Ocarina of Time (0400486024DEC000) black screen — no AV1 decoder (open)
• [VBug] OoT (0400486024DEC000): world renders black except areas lit by Navi and around Link, AMD RX 9070 (open)
• Release note: Outdoor areas have far too much fog and glow compared with console; performance is about 12–22 fps on a Ryzen 5900X with an RTX 3080, and three rare shaders remain unsupported.
• Release note: On macOS arm64 through MoltenVK, the Metal device may be lost or the GPU thread may hang at the intro-video/gameplay boundary. A commenter attributes the device loss to an attachment-less framebuffer reaching 32768×32768, above the stated Apple GPU limit of 16384.
• Release note: After a local workaround for device loss, a commenter reports that the 3D scene remains black while sound and subtitles work on an older Eglantine fork; current nx2 behavior was not yet checked.
• Release note: User reports a crash when opening the game on Arch Linux with an AMD Radeon AI PRO R9700 (9070XT); the console reports a segmentation fault.
• Release note: User reports audio/video desynchronization during the opening cutscene and a system freeze at the Deku Tree cutscene on Vulkan with an AMD RX 9000 series GPU. Comments report crashes on AMD GPUs, including while building shaders. One commenter reports that OpenGL GLSL stops the crashes but leaves only the UI visible.
• Release note: On some AMD GPUs, the UI may remain visible while the 3D world is black; reports also describe AMD driver timeouts or crashes.
• Release note: User-reported reproducible freeze when walking deeper into the pot room in Hyrule Castle Town, near the guard.
• Release note: The issue reporter using an AMD RX 6600 8GB on Windows 10 reports that the emulator starts loading and then remains on a black screen.
• Release note: Commenters report rendering failures, black screens, and AMD driver crashes; one reports that audio continues while rendering is crashed.
• Release note: On AMD's proprietary Windows driver, two compute pipelines can take 6 to 18 minutes to compile during the intro, leaving the game at 0 FPS while audio continues. Rejecting the shader hashes avoids the wait but causes black areas in the intro, title, and castle scenes.
• Release note: In the measurement-only experiment, disabling per-fence Preflushable downloads or batching downloads by 64 fences caused overexposure in Kokiri Forest.
• Release note: A commenter reports that an experimental compute storage write-extent analysis reduced marked GPU-written data and fence copies while preserving correct auto-exposure. It was described as code on a user's branch, not as an upstream or validated release fix.
• Release note: Static review and a separate research comment describe read-trap bookkeeping and dirty-state issues that can leave stale GPU-written data when per-fence downloads are disabled; proposed corrections and batching remain untested or local experiments.
• Release note: Slows when booting.
• Release note: Community comments describe AMD black or corrupted-world output, fragment-pipeline compiler crashes on Windows, later timeouts or TFB stalls, corrupted output, and a RADV depth-image SPIR-V failure. These are separate reported cases; the comments do not establish a universal cause or resolution.
• Release note: User reports a segfault during gameplay on Linux with an AMD RX 9000-series GPU and RADV, when a graphics pipeline is built for an invalid fragment shader.
• Release note: On Windows build-83, shutdown can hang at “Closing software…” after gameplay. Some sessions completed shutdown successfully; the cause is undetermined.
• Release note: Persistent graphics pipeline cache entries are reportedly rebuilt on the GPU thread during gameplay on repeated boots, causing a visible hitch.
• Release note: With GPU Mode: Accurate (High), gameplay may run in slow motion at about 30 FPS; shader compilation can cause stutter on new content.
• Release note: PGO artifact crashes in dynarmic with Unhandled SIGSEGV after 2–10 minutes on the reported machine.
• Release note: Clean audio from a different point of the game plays while the intro image is displayed; audio and picture are substantially out of sync.
• Release note: In gameplay, the world renders almost entirely black, with only the HUD and some areas near light sources visible; some shapes appear blown out with heavy bloom.

Compared with Previous Build

From build-83 to build-88: follow the gameplay changes described for these two builds.