Donkey Kong Bananza PC Recomp est un portage natif de Donkey Kong Bananza pour macOS, distribué en binaire dans ses releases.
Games included
The games this software runs.
Donkey Kong Bananza
Donkey Kong Bananza est le jeu d’origine du portage Donkey Kong Bananza pour macOS.
Architecture
Mac OS X - macOS
Apple Silicon
64-bit
Download and links
- Download Donkey Kong Bananza
- Source code 1 201 GitHub stars
- License : GPL-3.0
Version history
0.5.1
Full changelog
DKB Recomp 0.5.1: a static recompilation of *Donkey Kong Bananza* (Switch 2) to a native Windows exe with a Vulkan renderer. It is still experimental (0.x). A fix release for 0.5.0: on Windows with an NVIDIA GPU, 0.5.0 compiled its compute pipelines again at every start, which held the game near 30 fps for the first minutes.**The packages hold no game data, and no release ever will.** You build the game on your own PC from your own dump of **1.0.0 or the 3.2.0 update**, given as a `.nup` or `.nx` file or as ExeFS and RomFS folders. Other updates are not supported.## What is new since 0.5.0- **NVIDIA shader cache on Windows.** 0.5.0 pointed NVIDIA's shader disk cache at an `nv-cache` folder beside `dkb.exe` but never created that folder. The Windows driver reads that setting and kept no disk cache, so every start compiled the compute pipelines again in the background (about 5 minutes on a GTX 1650, with the game near 30 fps meanwhile; 0.4.2 took 0.2 s). 0.5.1 creates the folder. The first start fills it, and later starts are fast again (here 12.5 s to 0.13 s). Thanks @supermahdi-enjoyer for the four logs that showed it (#437). - **Player 2's rumble.** With two gamepads the game reads both, also in single-player, and sends the rumble for the pad you play on to that pad's own slot. Player 2's slot was ignored, so the rumble came out of the other pad. Player 2's rumble now plays on its own gamepad (two motors), and the log names both pads' ids (#445, thanks @mrflanderson for the two-pad test).## Also in 0.5.0 (since 0.4.2)- **Windows launcher.** `DKB-Launcher.exe` is one small file. It downloads the build tools from the latest release and checks their size and SHA-256, then builds the game from your dump and shows Play. At each start it looks for a new release (a tick box turns that off), shows the notes and updates in place with one click. An update downloads only the new tools zip, rebuilds only the steps whose inputs changed, and keeps your saves (`profile/`) and `dkb.ini` (#420, #140, #402). The setup zip still works as before. - **Performance presets.** `preset = 1..5` (Minimum to Ultra) or `recommended`, which picks a level from the GPU's video memory, the RAM and the processor count at start. A preset only fills in defaults, so any option you set yourself still wins, and the F1 menu has a Preset row. docs/requirements.md lists the resolution and frame rate to expect at each level. - **Upscaler quality.** `dlss_mode` picks the DLSS, FSR 3.1 and FSR 4 mode by name (DLAA, Quality, Balanced, Performance, Ultra Performance). The setup can fetch NVIDIA's DLSS DLL and AMD's FSR 4 from their repositories after you accept their licenses; both are off by default (#423). - **Local co-op.** A second gamepad plays Pauline (Pause > Co-op mode > Share controllers, #240). `coop = 0` turns it off. - **Languages.** `lang = th` plays update 3.2.0 in Thai (#180). Guest `setjmp`/`longjmp` now work, which the Thai text layout needs. - **16:10 screens** (1280x800, 1920x1200, 2560x1600). The view grows taller and the HUD stays whole in a centred 16:9 area. The title menu and loading tips fill the screen instead of leaving purple bands (#249). Tested on 3.2.0 and 1.0.0. - **Button prompts.** PlayStation prompts follow your `bind_` lines, not only `pad_layout` (#416). The game's Switch-letter prompts follow X/Y and A/B swaps, and `pad_layout = xbox_prompts` now works in Bananza too (#440). Run the setup again to rebuild the prompts. - **Start-up.** A start-up screen shows real progress in the game window until the game's first picture (#430; `splash = 0` turns it off). - **Loading and memory.** The staged mirror is on by default on Windows with an NVIDIA GPU. The frame that enters a level or layer takes about half as long (worst frame 444 to 230 ms on an RTX 5060 Ti 8 GB; #442, #424, #429; `mirror_staged = 0` turns it off). - **Fewer CPU threads.** - Recording a draw costs about 14% less. On 6 logical processors, frames over 50 ms in the heaviest scene tested fell from about 22 to 15 a minute (`TOTK_DYN_KEEP=0` brings back the old path). - The replay runs above normal priority while it submits a frame, so loading threads no longer starve it. On 6 logical processors a level transition's replay fell from about 11 ms a frame to 0.25 ms. - **AMD.** `replay_threads` is on by default on every discrete card on AMD's proprietary driver, not only the RX 9070 series (#141, #418). Integrated GPUs, AMDVLK and RADV keep it off, and the RX 5000 series is not measured yet. - **Crashes.** - Picking up some bananas in a busy area crashed 1.0.0. The game's per-frame GPU buffer list walk now stops at the list's 256 entries (#428). - A pool guard at `main+0x429594` (3.2.0) and `main+0x410794` (1.0.0) checks the element it loads, not only the array (#443, #204). - With `realtime = 1`, the frame rate is held to 120 fps at most (`realtime_max_fps`; 0 removes the ceiling). Shorter steps crash the record merge job in the EmeraldRush challenge levels (#411, #421, #441). - **Crash reports.** A second faulting thread waits for the first one's report instead of cutting its register dump short. On Linux and macOS, a fault in a driver or a Vulkan layer is reported by file name (#443). - **Setup.** - "Use the RomFS where it is" (`--romfs-in-place`, #305) reads your extracted `romfs` folder through `romfs_dir` instead of making a ~9 GB copy. Keep the folder where it is afterwards. - The setup no longer asks for the dump folders when they were moved after an install. It verifies and uses the copy already in the install folder (#438). - When the setup's own `bin` folder looks incomplete, it names the folder and the files it misses (#433). - The shader cache, about 118,000 loose files on 3.2.0, is folded into one `shader-cache.pack`, so deleting or copying an install no longer takes tens of minutes. - The finish screen shows where the kept build files are and how big they are, and can delete them. An Uninstall button, or `--uninstall` (`--delete-saves` also removes saves), removes the game and its build files. - **Launchers and frontends** (#402). - `dkb.exe --set KEY=VALUE` sets a `dkb.ini` key for one run, `dkb.exe --help` lists the options, and an unknown option is named. - `DKB-Setup.exe` and the AppImage pass a setup command line on to the command-line setup. `--play [DIR]` starts the finished game, and the setup window takes `--source` and `--dest` to prefill its form. - The AppImage keeps its work folder inside the install folder. - The NVIDIA shader cache on Linux, the flight recorder, and the uv, NuGet and dotnet caches stay in the install or work folder. A test fails if a setup writes anything to the home folder. - The Windows zip ships a RecompLauncher profile (`recomplauncher.json`), and the AppImage's folder also ships as a `.tar.gz`. - **macOS.** The install check finds the Vulkan loader and the headers on Apple silicon (#444). `DKB_HEADLESS=1` runs the game with no window, for captures and scripted runs. CI now compiles the whole runtime on macOS at every push. - **Picture.** Text and the HUD are sharp at 1440p and above with `scene_res = full`. Without it, the HUD is drawn at two thirds of the screen size and stretched (#214).## Pull requests in 0.5.0No pull requests since 0.5.0. Thank you to everyone who sent code for 0.5.0; each one below is new since 0.4.2:- #419 Keep the Bananza installer tests independent of the test machine's free disk space and symlink rights (@wrothgaril) - #420 DKB-Launcher.exe: a small Windows launcher that installs the build tools, checks for releases and updates in place with one click (@wrothgaril) - #422 Stamp release files with the commit time (@GlebShekovtsov) - #423 Let the Bananza setup fetch AMD FSR 4 from AMD's repository once the user accepts AMD's license (@GlebShekovtsov) - #424 Take the staged mirror's bookkeeping off the replay thread, and commit guest memory ahead of use in the background (@wrothgaril) - #425 Name the disk that is short of space in the Bananza installer's check (@cybersecurityhandle) - #429 Yield after each slice of the background commit thread, so a waiter for the memory lock gets in (@wrothgaril) - #430 A start-up screen with real progress in the game window (@wrothgaril) - #436 Two off-by-default scheduler settings against mutex convoys: `sched_handoff = 1` and `mutex_spin = ` (@wrothgaril) - #442 The staged mirror on by default on Windows with an NVIDIA GPU (@wrothgaril) - #444 macOS fixes: the install check, a headless test mode, and the new upscaler hashes (@anthonyyya)## Issues fixed or answered#437 (NVIDIA shader cache, 0.5.1, @supermahdi-enjoyer) and #445 (two gamepads, rumble on the wrong pad, @mrflanderson). In 0.5.0:#428 (banana pickup crash, @prayudigames93-ctrl, with dumps and a save), #443 (pool guard crash, @Felitendo, with the disassembly), #204 (@MedoShinwa), #411, #421 and #441 (EmeraldRush crash with `realtime`, @NoireOvO, @crni20zd, @Nikoroq), #240 (local co-op, @keerttttt), #180 (Thai, @RavenChz, who also tested the mod route), #249 (16:10, @dextert007), #305 (RomFS in place, @drboulakdemhadj-sys), #416 (DualSense prompts, @dancaiman), #440 (prompt swap, @Nikoroq), #438 (@kline1110) and #433 (@sharicov) (setup errors), #402 (launcher and packaging, @wrothgaril), #214 (text at 1440p, @Felitendo), #141 and #418 (AMD frame rate, @GlebShekovtsov, @AcousticBlu3, with runs from @akmuzyka and @Ghos796), #140 (launcher, @Koaxz). Some of these stay open until the reporter retests.## ThanksPull requests in 0.5.0: wrothgaril, GlebShekovtsov, cybersecurityhandle, anthonyyya. Issues, logs, saves and recordings: everyone named above. Thank you.## Download| Package | For | Status | |---|---|---| | `DKB-Launcher.exe` | Windows 10/11 x64 | New, the simplest way. Put it in a folder of its own and start it. It downloads the tools zip below, checks its SHA-256, builds the game from your dump and offers each later release with one click (docs/installer.md, "Launcher and updates"). | | `DKB-Setup--windows.zip` | Windows 10/11 x64 | Supported. Carries its own tools and the installer window: unpack, run `DKB-Setup.exe`, pick your dump. | | `dkb-release.txt` | the launcher | The version, file names, sizes and SHA-256 the launcher checks. | | `DKB-Setup--linux-x86_64.tar.gz` | Linux x86-64, kernel 6.7 or later | Preview. Source package with `setup.sh`. You install the system packages it lists (`./setup.sh --check`), then it runs the command-line installer. | | `DKB-Setup--x86_64.AppImage` | Linux x86-64 (SteamOS, any distribution with glibc 2.31 or later) | Preview. Self-contained: bundles Python, LLVM and a sysroot, so no system packages are needed. Make it executable and run it. | | `DKB-Setup--x86_64.tar.gz` | Same as the AppImage | The AppImage's folder as a plain archive, for launchers that cannot mount an AppImage (a Flatpak sandbox). Unpack, run `./setup.sh`, and `./play.sh DIR` starts the finished game. | | `DKB-Setup--macos-arm64.zip` | macOS 14.4 or later, Apple silicon | Preview. Same idea with `setup.command`. Play is not verified, and the destructible terrain is missing because Metal has no geometry stage. |## Requirements (Windows)- NVIDIA GPU with a recent driver (tested on an RTX 5060 Ti and an RTX 3070 Ti Laptop). AMD and Intel are experimental. On AMD's Windows driver the device loss most players hit is fixed (recorder thread off), but RADV on Linux can still hang (#99, #122, #198, #204). Intel Arc: #3. - 16 GB of RAM plus a large page file (the process commits about 9 GB); 32 GB is comfortable. Cards with 4 GB of video memory should use `preset = 1` or `2` (docs/requirements.md). - A controller is recommended.## Known issues- GPU hangs under RADV on Linux (#99, #122, #198, #204), and a driver fault on Intel Arc (#3). - Black shards where salt, acid or muck meets slime (#139, #266, #387, #439). On one PC the muck's vertex normals come out NaN from a GPU compute pass; it does not happen on our test PC. - Crashes reported without a reproduction here: #126 (Lagoon to Hilltop tunnel, retest asked) and #148 (Linux, after the first boss). - Hitches in the Resort and Tempest layers on 4 GB cards, from video memory pressure (#437; try `astc_bc = 1`). - Another device loss on an RTX 5080 at B1000 (#336). - On CPUs with 4 to 6 threads, heavy scenes stay around 40 to 45 fps. A first run with an empty pipeline cache still stalls about 3 s on its first heavy frame. - Checkerboard textures (#5), #112 and unrecorded NVN 2 calls (#6).## Reporting a bugOpen an issue with the log (`dkb.log`), the build line at its top (`dkb 0.5.1+`), your GPU and driver, and what you did before it failed. **Never share generated files, keys or screenshots that show game content.**Roadmap: #10.**Full Changelog**: https://github.com/Lucas-blip-png/dkb-pc-recomp/compare/v0.5.0...v0.5.1
0.5.0
Full changelog
DKB Recomp 0.5.0: a static recompilation of Donkey Kong Bananza (Switch 2) to a native Windows exe with a Vulkan renderer. It is still experimental (0.x). This release adds a one-file Windows launcher with one-click updates, performance presets for weak and strong PCs, local co-op, Thai, 16:10 screens, and crash fixes from player reports.
The packages hold no game data, and no release ever will. You build the game on your own PC from your own dump of 1.0.0 or the 3.2.0 update, given as a .nup or .nx file or as ExeFS and RomFS folders. Other updates are not supported.
Not run in play by us: co-op with two real gamepads (tested with a simulated second pad), the FSR 4 download and Anti-Lag 2 on AMD hardware, the Windows RecompLauncher profile, and play on macOS.
What is new since 0.4.2
- Windows launcher.
DKB-Launcher.exeis one small file. It downloads the build tools from the latest release and checks their size and SHA-256, then builds the game from your dump and shows Play. At each start it looks for a new release (a tick box turns that off), shows the notes and updates in place with one click. An update downloads only the new tools zip, rebuilds only the steps whose inputs changed, and keeps your saves (profile/) anddkb.ini(#420, #140, #402). The setup zip still works as before. - Performance presets.
preset = 1..5(Minimum to Ultra) orrecommended, which picks a level from the GPU's video memory, the RAM and the processor count at start. A preset only fills in defaults, so any option you set yourself still wins, and the F1 menu has a Preset row. docs/requirements.md lists the resolution and frame rate to expect at each level. - Upscaler quality.
dlss_modepicks the DLSS, FSR 3.1 and FSR 4 mode by name (DLAA, Quality, Balanced, Performance, Ultra Performance). The setup can fetch NVIDIA's DLSS DLL and AMD's FSR 4 from their repositories after you accept their licenses; both are off by default (#423). - Local co-op. A second gamepad plays Pauline (Pause > Co-op mode > Share controllers, #240).
coop = 0turns it off. - Languages.
lang = thplays update 3.2.0 in Thai (#180). Guestsetjmp/longjmpnow work, which the Thai text layout needs. - 16:10 screens (1280x800, 1920x1200, 2560x1600). The view grows taller and the HUD stays whole in a centred 16:9 area. The title menu and loading tips fill the screen instead of leaving purple bands (#249). Tested on 3.2.0 and 1.0.0.
- Button prompts. PlayStation prompts follow your
bind_lines, not onlypad_layout(#416). The game's Switch-letter prompts follow X/Y and A/B swaps, andpad_layout = xbox_promptsnow works in Bananza too (#440). Run the setup again to rebuild the prompts. - Start-up. A start-up screen shows real progress in the game window until the game's first picture (#430;
splash = 0turns it off). - Loading and memory. The staged mirror is on by default on Windows with an NVIDIA GPU. The frame that enters a level or layer takes about half as long (worst frame 444 to 230 ms on an RTX 5060 Ti 8 GB; #442, #424, #429;
mirror_staged = 0turns it off). - Fewer CPU threads.
- Recording a draw costs about 14% less. On 6 logical processors, frames over 50 ms in the heaviest scene tested fell from about 22 to 15 a minute (
TOTK_DYN_KEEP=0brings back the old path). - The replay runs above normal priority while it submits a frame, so loading threads no longer starve it. On 6 logical processors a level transition's replay fell from about 11 ms a frame to 0.25 ms.
- Recording a draw costs about 14% less. On 6 logical processors, frames over 50 ms in the heaviest scene tested fell from about 22 to 15 a minute (
- AMD.
replay_threadsis on by default on every discrete card on AMD's proprietary driver, not only the RX 9070 series (#141, #418). Integrated GPUs, AMDVLK and RADV keep it off, and the RX 5000 series is not measured yet. - Crashes.
- Picking up some bananas in a busy area crashed 1.0.0. The game's per-frame GPU buffer list walk now stops at the list's 256 entries (#428).
- A pool guard at
main+0x429594(3.2.0) andmain+0x410794(1.0.0) checks the element it loads, not only the array (#443, #204). - With
realtime = 1, the frame rate is held to 120 fps at most (realtime_max_fps; 0 removes the ceiling). Shorter steps crash the record merge job in the EmeraldRush challenge levels (#411, #421, #441).
- Crash reports. A second faulting thread waits for the first one's report instead of cutting its register dump short. On Linux and macOS, a fault in a driver or a Vulkan layer is reported by file name (#443).
- Setup.
- "Use the RomFS where it is" (
--romfs-in-place, #305) reads your extractedromfsfolder throughromfs_dirinstead of making a ~9 GB copy. Keep the folder where it is afterwards. - The setup no longer asks for the dump folders when they were moved after an install. It verifies and uses the copy already in the install folder (#438).
- When the setup's own
binfolder looks incomplete, it names the folder and the files it misses (#433). - The shader cache, about 118,000 loose files on 3.2.0, is folded into one
shader-cache.pack, so deleting or copying an install no longer takes tens of minutes. - The finish screen shows where the kept build files are and how big they are, and can delete them. An Uninstall button, or
--uninstall(--delete-savesalso removes saves), removes the game and its build files.
- "Use the RomFS where it is" (
- Launchers and frontends (#402).
dkb.exe --set KEY=VALUEsets adkb.inikey for one run,dkb.exe --helplists the options, and an unknown option is named.DKB-Setup.exeand the AppImage pass a setup command line on to the command-line setup.--play [DIR]starts the finished game, and the setup window takes--sourceand--destto prefill its form.- The AppImage keeps its work folder inside the install folder.
- The NVIDIA shader cache on Linux, the flight recorder, and the uv, NuGet and dotnet caches stay in the install or work folder. A test fails if a setup writes anything to the home folder.
- The Windows zip ships a RecompLauncher profile (
recomplauncher.json), and the AppImage's folder also ships as a.tar.gz.
- macOS. The install check finds the Vulkan loader and the headers on Apple silicon (#444).
DKB_HEADLESS=1runs the game with no window, for captures and scripted runs. CI now compiles the whole runtime on macOS at every push. - Picture. Text and the HUD are sharp at 1440p and above with
scene_res = full. Without it, the HUD is drawn at two thirds of the screen size and stretched (#214).
Pull requests in this release
Thank you to everyone who sent code. Each one below is new since 0.4.2:
- #419 Keep the Bananza installer tests independent of the test machine's free disk space and symlink rights (@wrothgaril)
- #420 DKB-Launcher.exe: a small Windows launcher that installs the build tools, checks for releases and updates in place with one click (@wrothgaril)
- #422 Stamp release files with the commit time (@GlebShekovtsov)
- #423 Let the Bananza setup fetch AMD FSR 4 from AMD's repository once the user accepts AMD's license (@GlebShekovtsov)
- #424 Take the staged mirror's bookkeeping off the replay thread, and commit guest memory ahead of use in the background (@wrothgaril)
- #425 Name the disk that is short of space in the Bananza installer's check (@cybersecurityhandle)
- #429 Yield after each slice of the background commit thread, so a waiter for the memory lock gets in (@wrothgaril)
- #430 A start-up screen with real progress in the game window (@wrothgaril)
- #436 Two off-by-default scheduler settings against mutex convoys:
sched_handoff = 1andmutex_spin =(@wrothgaril) - #442 The staged mirror on by default on Windows with an NVIDIA GPU (@wrothgaril)
- #444 macOS fixes: the install check, a headless test mode, and the new upscaler hashes (@anthonyyya)
Issues fixed or answered by this release
#428 (banana pickup crash, @prayudigames93-ctrl, with dumps and a save), #443 (pool guard crash, @Felitendo, with the disassembly), #204 (@MedoShinwa), #411, #421 and #441 (EmeraldRush crash with realtime, @NoireOvO, @crni20zd, @Nikoroq), #240 (local co-op, @keerttttt), #180 (Thai, @RavenChz, who also tested the mod route), #249 (16:10, @dextert007), #305 (RomFS in place, @drboulakdemhadj-sys), #416 (DualSense prompts, @dancaiman), #440 (prompt swap, @Nikoroq), #438 (@kline1110) and #433 (@sharicov) (setup errors), #402 (launcher and packaging, @wrothgaril), #214 (text at 1440p, @Felitendo), #141 and #418 (AMD frame rate, @GlebShekovtsov, @AcousticBlu3, with runs from @akmuzyka and @Ghos796), #140 (launcher, @Koaxz). Some of these stay open until the reporter retests.
Thanks
Pull requests since 0.4.2: wrothgaril, GlebShekovtsov, cybersecurityhandle, anthonyyya. Issues, logs, saves and recordings: everyone named above. Thank you.
Download
| Package | For | Status |
|---|---|---|
DKB-Launcher.exe |
Windows 10/11 x64 | New, the simplest way. Put it in a folder of its own and start it. It downloads the tools zip below, checks its SHA-256, builds the game from your dump and offers each later release with one click (docs/installer.md, "Launcher and updates"). |
DKB-Setup--windows.zip |
Windows 10/11 x64 | Supported. Carries its own tools and the installer window: unpack, run DKB-Setup.exe, pick your dump. |
dkb-release.txt |
the launcher | The version, file names, sizes and SHA-256 the launcher checks. |
DKB-Setup--linux-x86_64.tar.gz |
Linux x86-64, kernel 6.7 or later | Preview. Source package with setup.sh. You install the system packages it lists (./setup.sh --check), then it runs the command-line installer. |
DKB-Setup--x86_64.AppImage |
Linux x86-64 (SteamOS, any distribution with glibc 2.31 or later) | Preview. Self-contained: bundles Python, LLVM and a sysroot, so no system packages are needed. Make it executable and run it. |
DKB-Setup--x86_64.tar.gz |
Same as the AppImage | The AppImage's folder as a plain archive, for launchers that cannot mount an AppImage (a Flatpak sandbox). Unpack, run ./setup.sh, and ./play.sh DIR starts the finished game. |
DKB-Setup--macos-arm64.zip |
macOS 14.4 or later, Apple silicon | Preview. Same idea with setup.command. Play is not verified, and the destructible terrain is missing because Metal has no geometry stage. |
Requirements (Windows)
- NVIDIA GPU with a recent driver (tested on an RTX 5060 Ti and an RTX 3070 Ti Laptop). AMD and Intel are experimental. On AMD's Windows driver the device loss most players hit is fixed (recorder thread off), but RADV on Linux can still hang (#99, #122, #198, #204). Intel Arc: #3.
- 16 GB of RAM plus a large page file (the process commits about 9 GB); 32 GB is comfortable. Cards with 4 GB of video memory should use
preset = 1or2(docs/requirements.md). - A controller is recommended.
Known issues
- GPU hangs under RADV on Linux (#99, #122, #198, #204), and a driver fault on Intel Arc (#3).
- Black shards where salt, acid or muck meets slime (#139, #266, #387, #439). On one PC the muck's vertex normals come out NaN from a GPU compute pass; it does not happen on our test PC.
- Crashes reported without a reproduction here: #126 (Lagoon to Hilltop tunnel, retest asked) and #148 (Linux, after the first boss).
- Hitches in the Resort and Tempest layers on 4 GB cards, from video memory pressure (#437; try
astc_bc = 1). - Another device loss on an RTX 5080 at B1000 (#336).
- On CPUs with 4 to 6 threads, heavy scenes stay around 40 to 45 fps. A first run with an empty pipeline cache still stalls about 3 s on its first heavy frame.
- Checkerboard textures (#5), #112 and unrecorded NVN 2 calls (#6).
Reporting a bug
Open an issue with the log (dkb.log), the build line at its top (dkb 0.5.0+), your GPU and driver, and what you did before it failed. Never share generated files, keys or screenshots that show game content.
Roadmap: #10.
What's Changed
- Keep the Bananza installer tests independent of the test machine's free disk space and symlink rights by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/419
- DKB-Launcher.exe: a small Windows launcher that installs the build tools, checks for releases and updates in place with one click by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/420
- Stamp release files with the commit time by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/422
- Name the disk that is short of space in the Bananza installer's check by @cybersecurityhandle in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/425
- Let the Bananza setup fetch AMD FSR 4 from AMD's repository once the user accepts AMD's license by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/423
- Take the staged mirror's bookkeeping off the replay thread: steady play with TOTK_MIRROR_STAGED=all is as fast as without it by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/424
- Yield after each slice of the background commit thread, so a waiter for the memory lock gets in by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/429
- Show a start-up screen with real progress in the game window, from its opening to the game's first picture by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/430
- Turn the staged mirror on by default on Windows with an NVIDIA GPU, and test the paths #424 left untested by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/442
- Add two off-by-default scheduler settings against mutex convoys: a core hand-off as an SVC returns and a short poll before a wait sleeps by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/436
- macOS fixes: hostcheck repair, headless test mode, and the new upscaler hashes by @anthonyyya in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/444
New Contributors
- @cybersecurityhandle made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/425
- @anthonyyya made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/444
Full Changelog: https://github.com/Lucas-blip-png/dkb-pc-recomp/compare/v0.4.2...v0.5.0
0.4.2
Full changelog
DKB Recomp 0.4.2: a static recompilation of Donkey Kong Bananza (Switch 2) to a native Windows exe with a Vulkan renderer. It is still experimental (0.x). A small release: fixes and community pull requests from the day after 0.4.1, with most of the work on frame drops in heavy scenes, on first runs and on PCs with fewer CPU threads.
The packages hold no game data, and no release ever will. You build the game on your own PC from your own dump of 1.0.0 or the 3.2.0 update, given as a .nup or .nx file or as ExeFS and RomFS folders. Other updates are not supported.
Not run in play by us: Switch 2 controllers over Bluetooth (switch2 = 1), HD rumble on the pads that reported problems, and the lookahead on AMD hardware.
What is new since 0.4.1
- Heavy scenes. The layer -200 scene, the one most players reported, holds 60 fps on a Ryzen 7 5700X and RTX 5060 Ti with warm caches (first minute about 57 fps while it loads, second minute 59.9, 1% low 51; 0.4.0 gave 54 and 27). Most of it comes from the pointer guards no longer asking Windows about memory (#401) and the other replay and memory changes listed below.
- Fewer CPU threads. PCs with 4 to 7 logical processors (many laptops, 6-core CPUs without SMT) record big render passes on 3 threads by default instead of one: on layer -200, frames over 50 ms in the first minute fell from about 160 to 16 on 6 logical processors and from 203 to 47 on 4 (docs/dkb-lowend.md).
- First runs. With an empty pipeline cache, a draw that waits for its pipeline builds it itself instead of queueing behind dozens of skipped draws (longest wait 2.4 s to 0.23 s), and before it waits it queues the pipelines of the frame's next draws, so the first heavy frame stalls about 3 s instead of about 10 s (#309;
TOTK_LOOKAHEAD=0turns it off). - Long sessions. Frame drops that come after an hour or more of play (#317, #391, #380, #397) are much smaller with #401.
- Crashes and device loss. The Landfill Layer device loss on NVIDIA 50-series (#304) is avoided by skipping one vertex stage on NVIDIA.
- Picture. The radial blur a smash starts ends above 60 fps with
realtime = 1on 1.0.0 too, as it already did on 3.2.0 (#389, #390). - Input. HD rumble is off by default (
rumble_hd = autoturns it on) because three third-party Switch Pro pads broke with it; it sends neutral values while the motors are silent and falls back to two motors when a pad drops (#363, #383). Switch 2 controllers can connect over Bluetooth LE withswitch2 = 1(#413, off by default). - Installer and setup. DLC from a container whose base package has title id 0 installs, and an add-on that cannot be placed is reported (#388).
console = 0also closes a Windows Terminal window (#238, #364). A refusedoutput_ressays which sizes work (#403). The installer's dotnet no longer adds a dead folder to your PATH on every install. - Testing.
TOTK_VRAM_LIMIT=caps the video memory budget the runtime reads, to see how a smaller card behaves.
Pull requests in this release
Thank you to everyone who sent code. Each one below is new since 0.4.1:
- #354 Read indirect-reset pointer lists from guest memory instead of through a GPU read-back (TOTK_RESET_CPU_LIST) (@thatbananaman08, landed with a host-pool check)
- #357 Slice passes whose draws sample an image through a depth twin (@GlebShekovtsov)
- #360 Scripted game runs with a verdict: tools/scenario.py, save_warp.py and perf_check.py (@thatbananaman08)
- #361 Scan a pool's pages for its mirror sync under the pool's own lock, and take mirror_lock only to queue the copies (@thatbananaman08)
- #365 Skip the submit scans of copies that hold nothing for them (@thatbananaman08)
- #366 Log the scheduler state when no frame has been presented for a while (TOTK_STALL_DUMP, default 5 s) (@thatbananaman08)
- #367 Opt-in orders for consecutive dispatches: the game's barriers alone, or compute barriers (TOTK_CS_BARRIER) (@GlebShekovtsov)
- #368 Apply the follow-ups from the review of #338: Anti-Lag 2 across a device restart, Auto's flight thumbnails, and the frame_gen_by text (@roverdrubber)
- #372 Leave the translator's emulated compare-exchange loop after 1024 attempts (@roverdrubber)
- #373 Skip the futex wake of a Linux condition variable signal when no thread waits (@roverdrubber)
- #374 Give the heap's freed pages back under Linux memory pressure (@roverdrubber)
- #376 Share one sampler per create info and retire the least used near the driver's limit (@roverdrubber)
- #377 Keep an opaque ASTC texture as BC1 instead of BC3 with
astc_bc1 = 1(off by default) (@roverdrubber) - #378 Lazy write-back of shader storage writes on Linux, off by default (@roverdrubber)
- #379 Copy the full-size 3D through native anti-aliasing's upscale step (@roverdrubber)
- #385 Stamp the GPU writes of a command buffer in batches at its fence (@wrothgaril)
- #386 Write the GPU writes of a held part back to guest memory once the part is submitted (@wrothgaril)
- #392 Let a recording thread note its GPU writes in a list of its own (@wrothgaril)
- #393 Copy the indirect reset's disjoint records in address order (@wrothgaril)
- #399 Keep the window scaling filter row open when the window is the picture size (@wrothgaril)
- #400 Log where a large pool's first GPU use spends its time, and resume write-watch queries where a full answer stopped (@wrothgaril)
- #401 Answer the pointer guards' readability checks without VirtualQuery (@wrothgaril)
- #405 Thaw a pool's whole frozen regions without VirtualQuery (@wrothgaril)
- #406 Split the replay of a slow frame into its parts in the perf trace (@wrothgaril)
- #407 Create the DLSS feature and the motion pass's pipeline ahead of the first upscale (@wrothgaril)
- #408 Skip mem_touch's VirtualQuery for lazy-heap chunks known to be committed (@wrothgaril)
- #409 Save TOTK_SHOTS and --shot pictures of an RGB10A2 (HDR) presented image (@wrothgaril)
- #410 TOTK_MIRROR_STAGED (off by default): feed pools' VRAM mirror through staging (@wrothgaril)
- #413 Switch 2 controllers over Bluetooth LE (
switch2 = 1, off by default), ported from mkw-recomp (@zigzagulouszagger) - #414 Start a new guest thread on its ideal core alone, as Horizon does (@wrothgaril)
Issues fixed or answered by this release
#238 and #364 (console window, @mashakos, @JiangYunWeiShu), #304 (Landfill Layer device loss, @Dimiduh), #309 (late models and textures on a first run, @Th0rfinnKarlsefni), #317 (@li-beihai), #391 (@1665879844), #380 (@supermahdi-enjoyer) and #397 (@crni20zd) (frame drops over time and in heavy scenes), #363 and #383 (HD rumble, @mrflanderson, @zero-zhouxiang), #388 (DLC with a zero title id, found and fixed by @Ki1z), #389 and #390 (smash blur above 60 fps, @elitalos, @Chadman246), #403 (custom resolution, @NicNamSam). Some of these stay open until the reporter retests.
Thanks
Pull requests since 0.4.1: wrothgaril, roverdrubber, thatbananaman08, GlebShekovtsov, zigzagulouszagger. Issues, logs, saves and recordings: everyone named above. Thank you.
Download
| Package | For | Status |
|---|---|---|
DKB-Setup--windows.zip |
Windows 10/11 x64 | Supported. Carries its own tools and the installer window: unpack, run DKB-Setup.exe, pick your dump. |
DKB-Setup--linux-x86_64.tar.gz |
Linux x86-64, kernel 6.7 or later | Preview. Source package with setup.sh: you install the system packages it lists (./setup.sh --check), then it runs the command-line installer. |
DKB-Setup--macos-arm64.zip |
macOS 14.4 or later, Apple silicon | Preview. Same idea with setup.command. Play is not verified and the destructible terrain is missing (Metal has no geometry stage). |
Requirements (Windows)
- NVIDIA GPU with a recent driver (tested on an RTX 5060 Ti and an RTX 3070 Ti Laptop). AMD and Intel are experimental: on AMD's Windows driver the device loss most players hit is fixed (recorder thread off), and RADV on Linux can still hang (#99, #122, #198, #204). Intel Arc: #3.
- 16 GB of RAM plus a large page file (the process commits about 9 GB), 32 GB is comfortable.
- A controller is recommended.
Known issues
- GPU hangs under RADV on Linux (#99, #122, #198, #204), and a driver fault on Intel Arc (#3).
- Crashes reported without a reproduction here: #126 (Lagoon to Hilltop tunnel, retest asked), #136 (Forest snow tunnel landing: needs a guard on a call, after this release), #148 (Linux, after the first boss: the next log will name the function).
- Black shards where salt or acid meets slime (#139, #266, #387).
- Text is soft at 1440p and above: the game's fonts are drawn at a fixed size (#214;
DKB_FONT_SCALE=1helps small text). - Another device loss on an RTX 5080 at B1000 (#336); retests of the Landfill Layer skip (#304) and of long sessions (#317) are welcome.
- On CPUs with 4 to 6 threads heavy scenes stay around 40 fps, bound by the replay thread. A first run with an empty pipeline cache still stalls about 3 s on its first heavy frame, and area loads can stutter about 0.1 to 0.4 s.
- Checkerboard textures (#5), #112 and unrecorded NVN 2 calls (#6).
Reporting a bug
Open an issue with the log (dkb.log), the build line at its top (dkb 0.4.1+), your GPU and driver, and what you did before it failed. Never share generated files, keys or screenshots that show game content.
Roadmap: #10.
What's Changed
- Scan a pool's pages for its mirror sync under the pool's own lock, and take mirror_lock only to queue the copies by @thatbananaman08 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/361
- Skip the submit scans of copies that hold nothing for them: WaitSync scan without OP_WAIT_SYNC, deferred resolution without deferred records by @thatbananaman08 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/365
- Read indirect-reset pointer lists from guest memory instead of through a GPU read-back (TOTK_RESET_CPU_LIST) by @thatbananaman08 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/354
- Log the scheduler state when no frame has been presented for a while (TOTK_STALL_DUMP, default 5 s) by @thatbananaman08 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/366
- Opt-in orders for consecutive dispatches: the game's barriers alone, or compute barriers by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/367
- Apply the follow-ups from the review of #338: Anti-Lag 2 across a device restart, Auto's flight thumbnails, and the frame_gen_by text by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/368
- Leave the translator's emulated compare-exchange loop after 1024 attempts by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/372
- Slice passes whose draws sample an image through a depth twin by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/357
- Scripted game runs with a verdict: tools/scenario.py, save_warp.py and perf_check.py by @thatbananaman08 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/360
- Skip the futex wake of a Linux condition variable signal when no thread waits by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/373
- Share one sampler per create info and retire the least used near the driver's sampler limit by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/376
- Copy the full-size 3D through native anti-aliasing's upscale step when no upscaler runs it (#375) by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/379
- Stamp the GPU writes of a command buffer in batches at its fence by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/385
- Write the GPU writes of a held part back to guest memory once the part is submitted, before the submission's end by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/386
- Let a recording thread note its GPU writes in a list of its own and take the uniform snapshot's lock only for a write that drops a copy by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/392
- Keep the window scaling filter row open when the window is the picture's size, and say when Sharpness has no effect by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/399
- Answer the pointer guards' readability checks without VirtualQuery, which stalled write-watch queries (#317) by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/401
- Give the heap's freed pages back under Linux memory pressure and after the shader prebuild by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/374
- Copy the indirect reset's disjoint records in address order and note their GPU writes once per run of adjacent records by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/393
- Log where a large pool's first GPU use spends its time, and resume write-watch queries instead of rescanning by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/400
- Save TOTK_SHOTS and --shot pictures of an RGB10A2 presented texture by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/409
- Add a lazy write-back of shader storage writes on Linux, off by default by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/378
- Thaw a pool's whole frozen regions without VirtualQuery, which stalled the replay for 65 to 105 ms at each layer exit by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/405
- Split the replay of a slow frame into its parts, and keep the sampling profiler sampling while it lists the threads by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/406
- Make the DLSS feature and the motion pass's pipeline at the title screen, and match the game's upscaler before hashing shaders by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/407
- Skip mem_touch's VirtualQuery for lazy-heap chunks known to be committed, and log each first-use pool import by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/408
- Keep an opaque ASTC texture as BC1 instead of BC3 when astc_bc is on by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/377
- Add Switch 2 controllers over Bluetooth LE with the game's HD rumble (implementation ported from a different recomp listed below) by @zigzagulouszagger in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/413
- Start a new guest thread on its ideal core alone, which keeps a loader thread out of the main thread's job graph build (boot crash in #317) by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/414
- TOTK_MIRROR_STAGED (off by default): feed pools' VRAM mirrors through staging instead of importing them, and keep GPU writes in VRAM until the CPU reads them by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/410
New Contributors
- @zigzagulouszagger made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/413
Full Changelog: https://github.com/Lucas-blip-png/dkb-pc-recomp/compare/v0.4.1...v0.4.2
0.4.1
Full changelog
DKB Recomp 0.4.1: a static recompilation of Donkey Kong Bananza (Switch 2) to a native Windows exe with a Vulkan renderer. It is still experimental (0.x). This is a smaller release: what landed in the two days after 0.4.0, most of it from contributors.
The packages hold no game data, and no release ever will. You build the game on your own PC from your own dump of 1.0.0 or the 3.2.0 update, given as a .nup or .nx file or as ExeFS and RomFS folders. Other updates are not supported.
Not run in play by us: the Lanczos window filter at 1440p, AMD FSR 3.1 frame interpolation, AMD Anti-Lag 2 on AMD hardware, FSR 4 on AMD hardware and HD rumble on real controllers. The CRT and window filters and the frame interpolation come from community pull requests and have had little testing in play.
What is new since 0.4.0
- Crashes. A null edge target in the job graph builder is guarded on 1.0.0 and 3.2.0 (#317). A call that crashed after the Forest Layer snow tunnel gets a guard that skips it when its slot is null (#136, still open until the reporter retests). A GPU that finishes nothing for 30 seconds is reported as a device loss instead of a silent freeze (#329). A failed swapchain rebuild destroys the retired swapchain (#332). The frame generation display thread no longer takes its start lock under the present lock (#343).
- Picture and upscalers.
upscaler = fsr3is AMD FSR 3.1 temporal upscaling andupscaler = fsr4is AMD FSR 4 through your own AMD DLLs (#270, #333); both are off by default and fsr4 has not been run on AMD hardware by us. DLSS, FSR 3.1 and FSR 4 also run as native anti-aliasing atscene_res = fullon 3.2.0 (#345). Only draws that test or write depth are jittered (#321, #325), and the upscaler dispatch is found by its shape when its hash is not a listed one (#323). The upscaler row stays open on Bananza, and DLSS says in the log why it does not run (#152, #306). Window scaling filter:window_filter = fsr1,linear,catmull_romorlanczos, with Sharpness at any window size (#340);fsr1stays the default, and Lanczos at 1440p has not been run by us. CRT filters:crt_filter = off(default),lottes_grille,lottes_tv,hyllianorzfast, withcrt_lines(#342). ASTC textures stored as BC3 are copied instead of skipped, which is meant to bring back the Ingot Isle textures (#310, awaiting the reporter). Text above 1080p is sharper behindDKB_FONT_SCALE=1(#214, off by default). Dug terrain builds on RADV (#322). - Frame generation.
frame_genhas an Auto mode and 3x to 8x besides 2x, with the pacing they need (#334); it is off by default. At 2x,frame_gen_by = fsr3makes the frame with AMD FSR 3.1 frame interpolation instead of the built-in one (#346, off by default, not run by us), and it gets the frames without their HUD (#348). Frames are also made from the game's own HDR picture (#337). The Display tab has a Frame generation row and a Made by row (#349), and a set that waited is counted in the log (#352). - Performance. The uniform snapshot takes shortcuts (
TOTK_USNAP_FAST) and deferred guest reads are written back without waiting, on by default, about 1.5 ms a frame in Canyon Layer (#312).TOTK_MIRROR_OWNleaves the VRAM mirror's own GPU writes out of the sync (#295),TOTK_CALL_REFrecords some CallCommands as one call (#307), and a submission's GPU writes are copied back in one pass (#328).TOTK_DRAW_MERGEdraws runs of c4-only draws as instanced draws and is off by default (#300). The end of a sliced pass is held back without the recorder thread too (#292), and sliced passes are recorded by default on AMD's proprietary driver on the RX 9070 and 9070 XT, the chip measured (#298; RDNA 1 and 2 are left out, #84, #155). An upscaler that is no longer chosen releases its images after 120 frames (#351). On Linux the shader disk caches get 10 GiB of room unless you set a size (#330), and on Intel under Windows the pipeline cache data is fetched only while no pipeline is being built (#331). - Sound. The audio renderer is paced evenly instead of in bursts after each device period, which is meant to remove the crackling on a 1024-frame device (#133, #272, #347); #272 awaits the reporter.
- Languages and text. Polish (
pl,EUpl,pl-PL) is a language choice (#335). - Input.
pad:touchpadcan be bound, so the DualSense or DualShock 4 touchpad click can open the map (#291). HD rumble goes to a Switch Pro Controller or Joy-Con as amplitude and frequency (#285, not run on real controllers by us). The Controls tab builds against SDL older than 3.4.18 (#324). - Installer and Linux. The setup's host check builds against the release sysroot on Linux, so the AppImage no longer fails at step 1 (#293), and its last check accepts a CPU that has no AVX2 when you build for x86-64-v2 (#318).
vulkan_driverruns a Mesa build of your own on Linux without a wrapper script (#339). The--warm-shadersand--prepare-shaderslibrary builds run at below-normal priority on one core fewer than the machine has, from 4 cores up (#3). - Diagnostics. The performance overlay has a Breakdown mode that stacks each frame's parts in the graph (#341) and a Counter mode, a large FPS number with a small label (#350); both are values of
perf_overlay. AMD Anti-Lag 2 isantilag = 1, off by default, on an AMD GPU whose driver has VK_AMD_anti_lag (#338); not run on AMD hardware by us.TOTK_MEM_REPORT=Nlogs where the process's commit goes, andtools/lowend_bench.pyruns a low-end benchmark. The perf trace counts the waits on the generation-table lock. - Long sessions. The VRAM mirror sync asks Windows for written pages before it takes the mirror lock, so the replay and its slices wait less on it (#317, partial: the query itself gets slower as the process ages).
Pull requests in this release
Thank you to everyone who sent code. Each one below is new since 0.4.0:
- #292 Hold back the end of a sliced pass without the recorder thread too (@GlebShekovtsov)
- #294 Read page stamps under gen_lock in TOTK_MEMORY=verify (@Zorkats)
- #295 Leave the VRAM mirror's own GPU writes out of mirror_sync (TOTK_MIRROR_OWN) (@Zorkats)
- #296 Commit the probe's heap before the freeze checks (exe-probe-gfx-abi with the lazy heap) (@thatbananaman08)
- #298 Record sliced passes by default on AMD's proprietary driver too (@GlebShekovtsov)
- #300 Draw runs of c4-only draws as instanced draws (TOTK_DRAW_MERGE, off by default) (@Zorkats)
- #307 Record CallCommands of a recording without deferred records as one call instead of a copy (TOTK_CALL_REF) (@thatbananaman08)
- #312 Take shortcuts in the uniform snapshot and write deferred reads back without waiting, about 1.5 ms a frame in Canyon Layer (@Zorkats)
- #321 Jitter only the draws that test or write depth, so the temporal upscalers are not handed a picture shifted twice (@roverdrubber)
- #322 Build the geometry stage of dug terrain on RADV, whose input locations past 15 the location check refused (@roverdrubber)
- #323 Take the game's upscaler dispatch for DLSS, FSR 3.1 and FSR 4 by its shape when its hash is not one of the listed ones (@roverdrubber, landed by hand)
- #324 Build the Controls tab against SDL older than 3.4.18, whose headers have no SDL_GAMEPAD_TYPE_STEAM (@roverdrubber)
- #325 Jitter the scene draws by their depth test alone, as the review of #321 noted, and say which draws that leaves out (@roverdrubber)
- #326 Say in geometry_validate that taking the vertex stage's location limit on RADV is deliberate, as the review of #322 asked (@roverdrubber)
- #328 Copy a submission's GPU writes back to guest memory in one pass (@roverdrubber, landed by hand)
- #329 Report a GPU that finishes nothing for 30 seconds as a device loss (@roverdrubber)
- #330 Give Mesa's and NVIDIA's shader disk caches 10 GiB of room on Linux unless the player set a size (@roverdrubber, landed by hand)
- #331 Fetch the pipeline cache's data only while no pipeline is being built on Intel under Windows (@roverdrubber, landed by hand)
- #332 Destroy the retired swapchain when creating its replacement fails (@roverdrubber, landed by hand)
- #333 Give FSR 4 an input pass, FSR 3.1's output pass, a Linux library and depth flags (@roverdrubber)
- #334 Add frame generation's Auto mode and the 3x to 8x multipliers, with the pacing they need (@roverdrubber, landed by hand)
- #337 Make frames from the game's own HDR picture too (@roverdrubber)
- #338 Add AMD Anti-Lag 2 as an option (antilag = 1) (@roverdrubber)
- #339 Add an optional vulkan_driver setting so a Mesa build of the player's own can run the game on Linux without a wrapper script (@roverdrubber)
- #340 Give the window a choice of scaling filter, with Sharpness at any window size and the game's own HDR picture for the filters chosen (@roverdrubber)
- #341 Add a Breakdown mode to the performance overlay (@roverdrubber)
- #342 Add Hookshot's CRT filters to the present (crt_filter) (@roverdrubber)
- #343 Keep the frame generation display thread from taking the start lock under the present lock (@roverdrubber)
- #345 Run DLSS, FSR 3.1 and FSR 4 as native anti-aliasing at scene_res = full (@roverdrubber)
- #346 Add AMD FSR 3.1 frame interpolation as a way to make frame generation's 2x frames (frame_gen_by = fsr3) (@roverdrubber)
- #347 Pace the audio renderer evenly instead of in bursts after each device period (@Gabinou22s)
- #356 Refill a behind audio queue at once, drop the dead device pacing path (@Gabinou22s)
- #348 Give AMD FSR 3.1 frame interpolation the frames without their HUD, so the HUD is not moved with the scene (@roverdrubber)
- #349 Add Frame generation and its Made by row to the Display tab (@roverdrubber)
- #350 Add a Counter value to the performance overlay: a large FPS number with a small label, after Hookshot's (@roverdrubber)
- #351 Let an upscaler no longer chosen release its images, feature or context after 120 frames (@roverdrubber)
- #353 Slice indirect draws and cut the slice threads' lock waits, Canyon Layer replay 11.8 to 10.0 ms a frame (TOTK_PAR_INDIRECT) (@Zorkats)
- #358 Say in inline_recording's comment that TOTK_CALL_REF records a call instead of a copy (@thatbananaman08)
- #352 Count frame generation's set waits, name Auto's flight thumbnails, and correct its pacing comment and docs (@roverdrubber)
Issues closed or answered by this release
Thank you to everyone who reported a problem. Open ones wait for the reporter to retest.
- #133 Linux audio static with a 1024-frame period (@xenovorx), closed; #272 the same on Windows (@q1209786034-netizen), open, retest asked (#347)
- #152 DLSS not working properly (@Gavury), closed, and #306 DLSS/DLAA is not working (@badman101011), open: the upscaler row stays open on Bananza and DLSS says in the log why it does not run
- #84 (@philneitz) and #155 (@bubblesnout), closed: AMD RDNA 1 and 2 are left out of the default replay threads
- #291 Touchpad button in the controls (@Dimiduh), closed
- #270 FSR 4 / FSR 3.1 temporal upscaler (@MatyMeatBoy), open: fsr3 and fsr4 are on
bananza - #285 HD rumble request (@vink2212), open: Switch Pro Controller and Joy-Con only, untested
- #293 AppImage error on step 1 (@Nikoroq), open
- #310 Ingot Isle cave-in missing textures (@shiwntr), open
- #317 FPS drops after a while (@li-beihai), open: one null edge in the job graph builder is guarded
- #318 No AVX2 (@Enzo1383), open
- #335 Polish language (@nario999), open
- #136 Forest Layer snow tunnel crash (@luis-7654), open
- #214 Blurry text at 1440p (@Felitendo), open:
DKB_FONT_SCALE=1
Thanks
Pull requests since 0.4.0: roverdrubber, Zorkats, GlebShekovtsov, thatbananaman08, Gabinou22s. Thank you.
Download
| Package | For | Status |
|---|---|---|
DKB-Setup--windows.zip |
Windows 10/11 x64 | Supported. Carries its own tools and the installer window: unpack, run DKB-Setup.exe, pick your dump. |
DKB-Setup--linux-x86_64.tar.gz |
Linux x86-64, kernel 6.7 or later | Preview. Source package with setup.sh: you install the system packages it lists (./setup.sh --check), then it runs the command-line installer. |
DKB-Setup--macos-arm64.zip |
macOS 14.4 or later, Apple silicon | Preview. Same idea with setup.command. Play is not verified and the destructible terrain is missing (Metal has no geometry stage). |
Requirements (Windows)
- NVIDIA GPU with a recent driver (tested on an RTX 5060 Ti and an RTX 3070 Ti Laptop). AMD and Intel are experimental: on AMD's Windows driver the device loss most players hit is fixed (recorder thread off), and RADV on Linux can still hang (#99, #122, #198, #204). Intel Arc: #3.
- 16 GB of RAM plus a large page file (the process commits about 9 GB), 32 GB is comfortable.
- A controller is recommended.
Known issues
- GPU hangs under RADV on Linux (#99, #122, #198, #204), and a driver fault on Intel Arc (#3).
- Crashes reported without a reproduction here: #126 (Lagoon to Hilltop tunnel, retest asked), #136 (Forest snow tunnel landing: needs a guard on a call, after this release), #148 (Linux, after the first boss: the next log will name the function).
- Black shards where salt or acid meets slime (#139, #266).
- Text is soft at 1440p and above: the game's fonts are drawn at a fixed size (#214;
DKB_FONT_SCALE=1helps small text). - A device loss on an RTX 5070 in the Landfill Layer that we cannot reproduce (#304), and frame rate falling after a long session (#317).
- Heavy scenes can still fall under 60 fps, and the first visit to an area can stutter.
- Checkerboard textures (#5), #112 and unrecorded NVN 2 calls (#6).
Reporting a bug
Open an issue with the log (dkb.log), the build line at its top (dkb 0.4.1+), your GPU and driver, and what you did before it failed. Never share generated files, keys or screenshots that show game content.
Roadmap: #10.
What's Changed
- Commit the probe's heap before the freeze checks (exe-probe-gfx-abi with the lazy heap) by @thatbananaman08 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/296
- Draw runs of c4-only draws as instanced draws (TOTK_DRAW_MERGE, off by default) by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/300
- Read page stamps under gen_lock in TOTK_MEMORY=verify by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/294
- Leave the VRAM mirror's own GPU writes out of mirror_sync (TOTK_MIRROR_OWN) by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/295
- Take shortcuts in the uniform snapshot and write deferred reads back without waiting, about 1.5 ms a frame in Canyon Layer by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/312
- Jitter only the draws that test or write depth, so the temporal upscalers are not handed a picture shifted twice by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/321
- Build the geometry stage of dug terrain on RADV, whose input locations past 15 the location check refused by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/322
- Build the Controls tab against SDL older than 3.4.18, whose headers have no SDL_GAMEPAD_TYPE_STEAM by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/324
- Jitter the scene draws by their depth test alone, as the review of #321 noted, and say which draws that leaves out by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/325
- Say in geometry_validate that taking the vertex stage's location limit on RADV is deliberate, as the review of #322 asked by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/326
- Report a GPU that finishes nothing for 30 seconds as a device loss by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/329
- Add an optional vulkan_driver setting so a Mesa build of the player's own can run the game on Linux without a wrapper script by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/339
- Hold back the end of a sliced pass without the recorder thread too by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/292
- Pace the audio renderer evenly instead of in bursts after each device period by @Gabinou22s in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/347
- Keep the frame generation display thread from taking the start lock under the present lock by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/343
- Add AMD FSR 3.1 frame interpolation as a way to make frame generation's 2x frames (frame_gen_by = fsr3) by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/346
- Run DLSS, FSR 3.1 and FSR 4 as native anti-aliasing at scene_res = full by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/345
- Give FSR 4 an input pass, FSR 3.1's output pass, a Linux library and depth flags by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/333
- Give the window a choice of scaling filter, with Sharpness at any window size and the game's own HDR picture for the filters chosen by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/340
- Add a Breakdown mode to the performance overlay by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/341
- Add Hookshot's CRT filters to the present (crt_filter) by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/342
- Record sliced passes by default on AMD's proprietary driver too by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/298
- Add Frame generation and its Made by row to the Display tab by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/349
- Let an upscaler no longer chosen release its images, feature or context after 120 frames by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/351
- Add a Counter value to the performance overlay: a large FPS number with a small label, after Hookshot's by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/350
- Make frames from the game's own HDR picture too by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/337
- Add AMD Anti-Lag 2 as an option (antilag = 1) by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/338
- Give AMD FSR 3.1 frame interpolation the frames without their HUD, so the HUD is not moved with the scene by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/348
- Count frame generation's set waits, name Auto's flight thumbnails, and correct its pacing comment and docs by @roverdrubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/352
- Record CallCommands of a recording without deferred records as one call instead of a copy (TOTK_CALL_REF) by @thatbananaman08 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/307
- Refill a behind audio queue at once, drop the dead device pacing path by @Gabinou22s in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/356
- Slice indirect draws and cut the slice threads' lock waits, Canyon Layer replay 11.8 to 10.0 ms a frame by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/353
- Say in inline_recording's comment that TOTK_CALL_REF records a call instead of a copy by @thatbananaman08 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/358
New Contributors
- @roverdrubber made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/321
- @Gabinou22s made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/347
Full Changelog: https://github.com/Lucas-blip-png/dkb-pc-recomp/compare/v0.4.0...v0.4.1
0.4.0
Full changelog
DKB Recomp 0.4.0: a static recompilation of Donkey Kong Bananza (Switch 2) to a native Windows exe with a Vulkan renderer. It is still experimental (0.x). On recent NVIDIA cards players get well past the first level (Glacier, Snow World, Tempest), and it can crash.
The packages hold no game data, and no release ever will. You build the game on your own PC from your own dump of 1.0.0 or the 3.2.0 update, given as a .nup file or as ExeFS and RomFS folders. Other updates are not supported.
What is new since 0.3.0
- Crashes fixed or guarded. The Snow World descent device loss on NVIDIA is gone: one light quad (vertex stage ba1dd1e7ae8abb33) is skipped (#196, #197, #264). A crash while loading a save or the title screen at main+0x82fee0 is guarded (#157, #206, #173). The 3.2.0 game now has the layer-change guards 1.0.0 had (#218, as #126), a crash at a layer change when an id cache runs out of nodes is guarded on both versions (#218), and so is a record loop that reads through an unset table index (#229). A Linux crash report now gets as far as the game's function and frames (#148). On AMD's Windows driver the Vulkan recorder thread is off by default, which stopped the device loss for several players (#84, #124, #142, #155, #167, #176).
- Picture. The #9 light quads are drawn again, so their particles are back (#62), and the Tower light quads are drawn with a fold of their selector instead of faulting (#220).
scene_res = fulldraws the 3D scene at the full output size instead of the 2/3 the game settles at (#210), on 1.0.0 and 3.2.0 (#223). Draws get depth twins like compute passes, so the validation layer is clean in the first level. - Performance. Big render passes are recorded on 4 threads by default from 8 logical processors (not on AMD), dynamic uniform buffers are on by default (not on Intel), and the guest heap is lazy by default on Windows (commit about 9.3 GiB instead of 14.5 GiB). In the heavy -101 scene on an RTX 5060 Ti: about 57.5 fps with a 1% low of about 29 fps (0.3.0: 51.8 and 20.6). The end of a sliced pass no longer waits for its slices (#265), and small VRAM mirror write-backs are copied by one compute dispatch on NVIDIA (#269), and the destructible terrain's geometry-stage storage stays in VRAM instead of being copied back every frame (#268; Hilltop Layer on an RTX 5090: GPU time 16.1 to 12.4 ms a frame), with its bookkeeping kept off the mirror lock (#288). Command recordings reuse cached blocks and hand their buffer over instead of being copied, which removes almost all page faults in the Hilltop Layer (about 470,000 a second before, #289). Many lock and copy cuts in the sliced replay come from contributors (see Thanks).
- DLC and mods. Setup installs add-on content from a
.nupintoaoc-folders (#183, #111, #258, #263). Loose RomFS mods load from amodsfolder besidedkb.exe(#252). - Sound. A Windows or macOS device that opens with a period that is not a multiple of the renderer's block is reopened with one block per period, which stopped the crackling on a 1024-frame device (#272).
- High resolutions. The game's GSys heap grows with the render-target pool, so
scene_res = fullat 3840x2160 and very large outputs no longer crash at start (#244, #279). A guide for 4 GB VRAM cards and laptops:docs/4gb-vram-optimization.md(#260). - Settings and input. PlayStation button prompts when a DualShock or DualSense is connected at start (
pad_prompts, #274). The vertical mouse camera no longer locks during automatic camera moves (#245). Big passes use 6 replay threads from 12 logical processors (#271). NVIDIA DLSS is listed in the upscaler row (#241), and its warm-up waits when another upscaler is chosen at start (#280). DLSS uses preset K except in Ultra Performance, which is cheaper than M on the GPU (dlss_preset, #287).console = 0hides the console window for frontends such as Playnite (#238). Controller connection handling, DK moving on walls and ceilings above 60 fps (#201), the 3.2.0 version on the title screen. - Installer. It refuses a RomFS folder that holds a RomFS image (#278), compiles only the sources a release ships (#203, #208), recompiles every runtime source when any of them changes (#275), and builds the ultrawide overlays. RecompLauncher can set up and start the Linux and macOS packages, and the release now has a self-contained Linux AppImage,
DKB-Setup-0.4.0-x86_64.AppImage, beside the.tar.gz(#242, #174)..nx(NXFormat) dumps are accepted next to.nup(#224). CPUs without AVX2 can build the game: setup compiles it for x86-64-v2, which is slower (#231). The shader library prebuild of--warm-shadersresumes where it stopped after a restart (#3). - Diagnostics. A crash outside the executable (a driver, a Vulkan layer) is reported with its module, and every crash writes a minidump to
logs/after the report (#219); keep it on your PC and send it only when asked, privately. - Project. Issue forms ask for the build, game version and OS; a security policy and an accessibility page were added.
Pull requests in this release
Thank you to everyone who sent code. Each one below is on bananza and in this release:
- #209 Slice passes under DLSS, and close the follow-ups of #195 (@GlebShekovtsov)
- #211 Take the texture and mirror locks out of the sliced draws, and reuse a draw's texture descriptors (@GlebShekovtsov)
- #213 Let RecompLauncher set up and start the Linux and macOS packages (@Felitendo)
- #217 Fail a store exclusive after another thread's change and change back (@thatbananaman08)
- #219 Report crashes outside the executable and write a minidump for every crash (@thatbananaman08)
- #220 Draw the faulting light quads with a fold of their selector instead of skipping them, and fix Tower's device loss (@thatbananaman08)
- #222 Run every test from one command, and fix three tests that no longer built or ran (@thatbananaman08)
- #223 Make scene_res draw the 3D scene at its ratio on 3.2.0 too (@wrothgaril)
- #225 Make the DLSS feature ahead of play and once for every input size (@wrothgaril)
- #226 Batch small GPU writebacks and reuse copy buffers (@oshoubber)
- #227 Keep the texture cache generation stable during replay (@oshoubber)
- #228 Batch terrain reset copy commands (@oshoubber)
- #234 Take the texture lock off a guest read's GPU wait, and sort only the new GPU write ranges of the uniform snapshot (@GlebShekovtsov)
- #235 docs: keep the 3.2.0 build status sentence in one tense (@lunar-me)
- #236 docs: fix the verb for the shader census step (@lunar-me)
- #237 docs: add the missing preposition in the cursor-release list (@lunar-me)
- #242 Add selfcontained AppImage build (@Osohow)
- #244 Grow Bananza's GSys heap with the render-target pool, so scene_res and large outputs no longer crash at start (@PorlocK-BR)
- #245 Fix intermittent vertical mouse camera lock in Bananza (@oshoubber)
- #246 Keep DK moving on walls and ceilings above 60 fps (@Zorkats)
- #250 Show the 3.2.0 version on the title screen (@Zorkats)
- #254 Improve controller connectivity, selection and hot-swapping (@shadowlink1174)
- #255 Retire a released pool's descriptors from an index instead of the whole table (@wrothgaril)
- #256 vk_exec_draw: add depth variant origin for mine voxel native GS on RDNA 1/2 (@MatyMeatBoy)
- #257 Take the texture and pipeline locks out of most sliced draws (@wrothgaril)
- #259 Wait for a texture write-back's GPU copy without the texture lock (@wrothgaril)
- #260 perf(docs): add 4GB VRAM optimization guide and config profile (@devadobe11-pixel)
- #261 Prepare Bananza ultrawide overlays during installation (@Marjaperse)
- #262 Find a resolved pool through a per-thread sorted list and a block map (@wrothgaril)
- #265 Let the end of a sliced pass go on without waiting for its slices, by default (@wrothgaril)
- #268 Keep the storage the geometry stage writes in VRAM (@wrothgaril)
- #269 Copy small VRAM mirror write-backs with the compute dispatch by default (@wrothgaril)
- #271 Record sliced passes on 6 threads by default from 12 logical processors (@wrothgaril)
- #273 Wait for an upload batch's fence without cmd_lock (@wrothgaril)
- #274 Show PlayStation button prompts in Bananza when a DualShock or DualSense pad is connected at start (@GlebShekovtsov)
- #275 Compile every runtime source of an install build again whenever its contents change (@GlebShekovtsov)
- #277 Document Steam Deck menu access when the fullscreen hold is enabled (@shadowlink1174)
- #280 Let the DLSS warm-up wait while another upscaler is chosen (@wrothgaril)
- #287 Ask DLSS for preset K outside Ultra Performance, with dlss_preset to choose (@wrothgaril)
- #288 Keep the kept-write bookkeeping off mirror_lock's hot path (@wrothgaril)
- #289 Keep large recording copies in cached blocks and hand the recording buffer over (@wrothgaril)
Issues resolved
Thank you to everyone who reported a problem, with logs, saves and recordings:
- Device loss on NVIDIA in the Snow World descent and the Glacier rock ride: #197 (@hotcoffeek), #264 (@dudesh34), #196 (@philneitz)
- Device loss or black screen on AMD's Windows driver, with the recorder thread now off by default: #84 (@philneitz), #123 (@BloodThirstyQ), #124 (@TeknaPT), #142 (@beijingpunk), #155 (@bubblesnout), #167 (@jesusen904), #176 (@BoySKAyt), #193 (@stuzukila82-bit), #215 (@thabans), #267 (@350Plays)
- DLC and the update from a
.nup: #111 (@infinitytom), #183 (@Blinkk000), #258 (@ElBusAlv), #263 (@evylmen) - Loose mods: #252 (@Caladbolg95)
- Render scale and 3D resolution: #149 (@formandfunctionlab), #210 (@dancaiman), #279 (@Dimiduh)
- DLSS: #152 (@Gavury), #241 (@thecharted135-sketch)
- Performance in heavy scenes: #154 (@patrickfuly), #156 (@supermahdi-enjoyer), #194 (@li-beihai), #207 (@solidaytug), #230 (@zachcristallguy-netizen), #284 (@wrothgaril)
- Above 60 fps, climbing walls: #201 (@XDM-Inc)
- Raw mouse camera: #177 (@pritomroyreal-cloud)
- Controllers: #253 (@shadowlink1174)
- Installer: #203 (@bika66), #208 (@Sapo2099), #278 (@sonic-un), #138 (@asluan287-gif)
Thanks
Pull requests: wrothgaril (#223, #225, #255, #257, #259, #262, #265, #269, #268, #271, #273, #280, #287, #288, #289), GlebShekovtsov (#209, #211, #234, #274, #275), oshoubber (#226, #227, #228, #245), PorlocK-BR (#244), devadobe11-pixel (#260), Osohow (#242), thatbananaman08 (#217, #219, #220, #222), shadowlink1174 (#254, #277), Zorkats (#246, #250), Marjaperse (#261), MatyMeatBoy (#256), Felitendo (#213), lunar-me (#235, #236, #237). Thank you.
Testing and reports: everyone who sent logs, saves and input recordings, in particular luis-7654, philneitz, Felitendo and Dimiduh.
Download
| Package | For | Status |
|---|---|---|
DKB-Setup--windows.zip |
Windows 10/11 x64 | Supported. Carries its own tools and the installer window: unpack, run DKB-Setup.exe, pick your dump. |
DKB-Setup--linux-x86_64.tar.gz |
Linux x86-64, kernel 6.7 or later | Preview. Source package with setup.sh: you install the system packages it lists (./setup.sh --check), then it runs the command-line installer. |
DKB-Setup--macos-arm64.zip |
macOS 14.4 or later, Apple silicon | Preview. Same idea with setup.command. Play is not verified and the destructible terrain is missing (Metal has no geometry stage). |
Requirements (Windows)
- NVIDIA GPU with a recent driver (tested on an RTX 5060 Ti and an RTX 3070 Ti Laptop). AMD and Intel are experimental: on AMD's Windows driver the device loss most players hit is fixed (recorder thread off), and RADV on Linux can still hang (#99, #122, #198, #204). Intel Arc: #3.
- 16 GB of RAM plus a large page file (the process commits about 9 GB), 32 GB is comfortable.
- A controller is recommended.
Known issues
- GPU hangs under RADV on Linux (#99, #122, #198, #204), and a driver fault on Intel Arc (#3).
- Crashes reported without a reproduction here: #126 (Lagoon to Hilltop tunnel, retest asked), #136 (Forest snow tunnel landing: needs a guard on a call, after this release), #148 (Linux, after the first boss: the next log will name the function).
- Black shards where salt or acid meets slime (#139, #266).
- Text is soft at 1440p and above: the game's fonts are drawn at a fixed size (#214).
- Heavy scenes can still fall under 60 fps, and the first visit to an area can stutter.
- Checkerboard textures (#5), #112 and unrecorded NVN 2 calls (#6).
Reporting a bug
Open an issue with the log (dkb.log), the build line at its top (dkb 0.4.0+), your GPU and driver, and what you did before it failed. Never share generated files, keys or screenshots that show game content.
Roadmap: #10.
What's Changed
- Slice passes under DLSS, and close the follow-ups of #195 by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/209
- Batch terrain reset copy commands by @oshoubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/228
- Let RecompLauncher set up and start the Linux and macOS packages by @Felitendo in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/213
- Run every test from one command, and fix three tests that no longer built or ran by @thatbananaman08 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/222
- Fail a store exclusive after another thread's change and change back by @thatbananaman08 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/217
- Batch small GPU writebacks and reuse copy buffers by @oshoubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/226
- Keep the texture cache generation stable during replay by @oshoubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/227
- Take the texture and mirror locks out of the sliced draws, and reuse a draw's texture descriptors by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/211
- docs: keep the 3.2.0 build status sentence in one tense by @lunar-me in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/235
- docs: fix the verb for the shader census step by @lunar-me in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/236
- docs: add the missing preposition in the cursor-release list by @lunar-me in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/237
- Show the 3.2.0 version on the title screen by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/250
- vk_exec_draw: add depth variant origin for mine voxel native GS on RDNA 1/2 by @MatyMeatBoy in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/256
- Make scene_res draw the 3D scene at its ratio on 3.2.0 too by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/223
- Keep DK moving on walls and ceilings above 60 fps by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/246
- Retire a released pool's descriptors from an index instead of the whole table by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/255
- Take the texture and pipeline locks out of most sliced draws by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/257
- Find a resolved pool through a per-thread sorted list and a block map by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/262
- Wait for a texture write-back's GPU copy without the texture lock by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/259
- Prepare Bananza ultrawide overlays during installation by @Marjaperse in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/261
- Make the DLSS feature ahead of play and once for every input size by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/225
- Improve controller connectivity, selection and hot-swapping by @shadowlink1174 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/254
- Document Steam Deck menu access when the fullscreen hold is enabled by @shadowlink1174 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/277
- Wait for an upload batch's fence without cmd_lock by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/273
- Take the texture lock off a guest read's GPU wait, and sort only the new GPU write ranges of the uniform snapshot by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/234
- Grow Bananza's GSys heap with the render-target pool, so scene_res and large outputs no longer crash at start by @PorlocK-BR in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/244
- Fix intermittent vertical mouse camera lock in Bananza by @oshoubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/245
- Show PlayStation button prompts in Bananza when a DualShock or DualSense pad is connected at start by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/274
- Compile every runtime source of an install build again whenever its contents change by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/275
- perf(docs): add 4GB VRAM optimization guide and config profile by @devadobe11-pixel in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/260
- Record sliced passes on 6 threads by default from 12 logical processors by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/271
- Copy small VRAM mirror write-backs with the compute dispatch by default by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/269
- Add selfcontained AppImage build by @Osohow in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/242
- Let the end of a sliced pass go on without waiting for its slices, by default by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/265
- Report crashes outside the executable and write a minidump for every crash by @thatbananaman08 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/219
- Draw the faulting light quads with a fold of their selector instead of skipping them, and fix Tower's device loss by @thatbananaman08 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/220
- Let the DLSS warm-up wait while another upscaler is chosen by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/280
- Ask DLSS for preset K outside Ultra Performance, with dlss_preset to choose by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/287
- Keep the kept-write bookkeeping off mirror_lock's hot path by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/288
- Keep large recording copies in cached blocks and hand the recording buffer over by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/289
New Contributors
- @Felitendo made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/213
- @thatbananaman08 made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/222
- @lunar-me made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/235
- @MatyMeatBoy made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/256
- @Marjaperse made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/261
- @shadowlink1174 made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/254
- @PorlocK-BR made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/244
- @devadobe11-pixel made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/260
- @Osohow made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/242
Full Changelog: https://github.com/Lucas-blip-png/dkb-pc-recomp/compare/v0.3.0...v0.4.0
0.3.0
Full changelog
DKB Recomp 0.3.0: a static recompilation of Donkey Kong Bananza (Switch 2) to a native Windows exe with a Vulkan renderer. It is still experimental (0.x): it plays the tutorial, its cutscene and the first stage on recent NVIDIA cards, and it can crash.
The packages hold no game data, and no release ever will. You build the game on your own PC from your own dump of 1.0.0 or the 3.2.0 update, given as a .nup file or as ExeFS and RomFS folders. Other updates are not supported.
What is new since 0.2.0
- Settings menu. Settings are split into Display and Graphics tabs, controller and mouse-and-keyboard settings are separate rows, dropdowns and steppers scroll and behave better, and the binding help is corrected. Two wider resolution presets for ultrawide screens were added.
- HDR (experimental, off by default). An HDR10 output with
hdr,hdr_peak_nitsandhdr_game_nitssettings, and an experiment that lets the game render its own HDR10 picture (DKB_HDR_NATIVE=1, 3.2.0 withhdr = 1). Native HDR highlights are taken up to the display's peak, because the game's own tonemap stops at about 1.36 times paper white. - DLSS (experimental, RTX only). DLSS is now temporal: sub-pixel jitter, depth and camera-only motion vectors reach NGX. DLSS output is the game's own 3D resolution, so
render_scaledoes nothing on Bananza and the HUD shows the upscaler's real sizes. - High resolutions. 3840x2160 and 5120x1440 no longer overflow the game's fixed render-target pool, which now grows with the output size (#152). 3840x2160 ran 90 s at 60 fps on 3.2.0 without a crash.
- Update 3.2.0. DLSS works there with the same shaders as on 1.0.0, and a radial blur that stayed on screen above 60 fps after a smash is released (#158).
- Frame generation (experimental, off by default).
FRAME_GEN=2xshows an interpolated frame half an interval before each real one, only in mailbox or immediate present mode; it skips a made frame on a hitch or a crop change. Seedocs/dkb-framegen.md. - Performance. Descriptor memos each sit behind their own switch (those that did not pay were removed), the draw state is one context per thread, uniform buffers can be dynamic (
TOTK_DYN_UBO, off by default until validated on AMD and Intel), range lists use an inlined merge sort, and about 1 ms was cut from the heavy-scene replay. Helper threads that translate long runs of draws (TOTK_REPLAY_PAR) and sliced big passes (DKB_REPLAY_THREADS) exist but are off by default. An unlimited frame limit (TOTK_FPS_LIMIT=unlimited) was added (#113). - Memory.
TOTK_HEAP_LAZY=1(Windows, off by default) commits the guest heap on first touch: the commit of the first level drops from about 14.5 GiB to about 9.3 GiB. It relies on spin-lock patches that are now in the 1.0.0 and 3.2.0 patch tables; the same patches fix a spin-lock livelock (#126). - Low-end bench.
tools/lowend_bench.pyruns the game on the first N logical processors and prints one table. On the first level it held 60, 60, 58.9 and 55.6 fps on 16, 8, 4 and 2 logical processors (docs/dkb-lowend.md). - Installer. A work folder path with shell-special characters is refused before the build starts (#189), and the work folder is named without them. The installer also tells you when a newer release exists (
--no-update-checkturns it off), and shows a clear message when Windows asks for elevation. - macOS and Metal. The macOS build is fixed (a function defined twice), and the host side of the graphics layer now goes through
GfxBackendto prepare a Metal back end. Play on macOS is still not verified. - Diagnostics.
TOTK_SVC_SLOW=1logs guest service calls that block for 100 ms or more, slow RomFS reads are logged,TOTK_MEM_REPORT=Nreports where the commit goes,DKB_DRAW_PROFILEgives replay time by section, and two AMD device-lost aids exist (TOTK_FIX_VS_FAULT=all,TOTK_CRUMB_BIND=1). - Scripted input. Wait markers and a replay clock that lands at the same game time on every run.
Download
| Package | For | Status |
|---|---|---|
DKB-Setup--windows.zip |
Windows 10/11 x64 | Supported. Carries its own tools and the installer window: unpack, run DKB-Setup.exe, pick your dump. |
DKB-Setup--linux-x86_64.tar.gz |
Linux x86-64, kernel 6.7 or later | Preview. Source package with setup.sh: you install the system packages it lists (./setup.sh --check), then it runs the command-line installer. |
DKB-Setup--macos-arm64.zip |
macOS 14.4 or later, Apple silicon | Preview. Same idea with setup.command. Play is not verified and the destructible terrain is missing (Metal has no geometry stage). |
Requirements (Windows)
- NVIDIA GPU with a recent driver (tested on an RTX 5060 Ti and an RTX 3070 Ti Laptop). AMD and Intel are experimental: device loss has been reported (#3, #84, #99, #124, #142, #193).
- 16 GB of RAM plus a large page file (the process commits about 14 GB, or about 9 GB with
TOTK_HEAP_LAZY=1), 32 GB is comfortable. - A controller is recommended.
Known issues
- Device lost on several AMD cards without a fault address (#3, #84, #99, #124, #142, #193), and a GPU hang on an RX 6600 under RADV (#122).
- A crash on Linux when walking into the tunnel from the Lagoon Layer to the Hilltop Layer (#126).
- A frame limit below 60 on 1.0.0 only holds with "Game speed follows the clock" on; without it the game keeps its own timing (#118).
- A reproducible device lost in the Glacier layer rock ride on an RTX 5090 Laptop (#196).
- The validation layer reports 10 errors from a depth-compare sampler on an R32_SFLOAT view; no visible effect has been reported.
- The first visit to an area can stutter. Open-area performance is 31 to 35 fps on an RTX 3070 Ti Laptop; a heavy scene on an RTX 5060 Ti measured 51.8 fps with a 1% low of 20.6 fps, so heavy scenes can still fall under 60 fps.
- The cutscene at the end of the tutorial lost the device on NVIDIA before a workaround (#75); it is on by default and the cause is not known.
- The NVIDIA workaround for #9 skips light quads, which hides some particles (#62).
- Checkerboard textures (#5), one-sided sound in the first 20 seconds (#63), #76, #83, #112 and unrecorded NVN 2 calls (#6).
Reporting a bug
Open an issue with the log (dkb.log), the build line at its top (dkb 0.3.0+), your GPU and driver, and what you did before it failed. Never share generated files, keys or screenshots that show game content.
Roadmap: #10. Testers: #3 (other GPUs) and #7 (play further).
What's Changed
- Guard the layer-change pool loop main+0x410680 against stale entries and cells, through a new guard patch kind by @luis-7654 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/135
- Release the radial blur a smash starts under the experimental dynamic step by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/159
- Use Bananza's own HDR10 output and take its highlights to the display's peak by @wrothgaril in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/150
- Measure the replay per draw in the perf log, and give descriptor pools 1024 sets by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/153
- Dynamic uniform buffers in the fragment set (+16% fps in dense areas); optional async write-back; per-frame A/B totals by @patrickfuly in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/161
- Add 3840×1600 and 5120×2160 ultrawide presets by @oshoubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/168
- Fix the macOS build: vk_device.c defined instance_ext_present twice by @LiamTheCreat0r in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/178
- Diagnostics for slow service calls, slow RomFS reads and the replay thread by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/182
- Metal back end, step 1: route the host side through GfxBackend and move the SPIR-V rewrites and c0 fill out of the Vulkan files by @LiamTheCreat0r in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/179
- Keep texture descriptor memos across submissions by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/181
- Improve settings dropdowns, arrow controls, and scrolling by @oshoubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/169
- Sort the per-submission range lists with an inlined merge sort instead of the CRT's qsort by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/184
- Replay helper thread by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/164
- Start three helper threads from 8 logical processors and two from 4 by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/185
- Organize settings into Display and Graphics tabs by @oshoubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/170
- Correct the Mouse & Keyboard binding help by @oshoubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/192
- Record a helper part's render into a texture released before its join, and read npend atomically by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/191
- Keep the draw state in one context per thread and compile the draw translator twice instead of six times by @Zorkats in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/190
- Record big passes in slices on worker threads, each with a draw context of its own (DKB_REPLAY_THREADS) by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/195
New Contributors
- @luis-7654 made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/135
- @wrothgaril made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/159
- @patrickfuly made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/161
- @Zorkats made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/182
Full Changelog: https://github.com/Lucas-blip-png/dkb-pc-recomp/compare/v0.2.0...v0.3.0
0.2.0
Full changelog
DKB Recomp 0.2.0: a static recompilation of Donkey Kong Bananza (Switch 2) to a native Windows exe with a Vulkan renderer. It is still experimental (0.x): it plays the tutorial, its cutscene and the first stage on recent NVIDIA cards, and it can crash.
The packages hold no game data, and no release ever will. You build the game on your own PC from your own dump of 1.0.0 or the 3.2.0 update, given as a .nup file or as ExeFS and RomFS folders. Other updates are not supported.
What is new since 0.1.0
- Update 3.2.0 (title version 458752). The installer takes a
.nupfile, picks the package and builds the game from it. A 3.2.0 build installs, boots and held 60 fps for 150 s at the start on an RTX 5060 Ti; it has not been played further. It carries no code patches, so the output resolution, ultrawide and real-time frame settings do nothing there (the log says so). - Add-on content (DLC) reader. Checked on synthetic packs and on a real one; not tried in the game yet.
- AMD. The map's long cyan wedges are gone on an RX 9070 XT (reported by the contributor who fixed it, not tested elsewhere). A NaN guard on fragment outputs stops a NaN texel from being spread by the bloom into glowing bands (
DKB_NAN_GUARD, on by default). - Sound. The left channel was silent at the start; fixed.
- Installer. It warns when the Windows commit limit is below 24 GiB, and keeps the newer driver when the loader lists the same GPU twice.
- Performance and shaders. Fewer texture re-uploads, more translated shader forms (1D image atomics, packed halves), and the NVN 2 mode calls the game uses are handled instead of ignored.
- Linux (preview, untested guesses). The shader translator is made executable if an archive lost the bit, and the sound asks for a 240-frame period by default. Under RADV the feedback loops default to the pass-restart path.
Download
| Package | For | Status |
|---|---|---|
DKB-Setup--windows.zip |
Windows 10/11 x64 | Supported. Carries its own tools and the installer window: unpack, run DKB-Setup.exe, pick your dump. |
DKB-Setup--linux-x86_64.tar.gz |
Linux x86-64, kernel 6.7 or later | Preview. Source package with setup.sh: you install the system packages it lists (./setup.sh --check), then it runs the command-line installer. |
DKB-Setup--macos-arm64.zip |
macOS 14.4 or later, Apple silicon | Preview. Same idea with setup.command. Play is not verified and the destructible terrain is missing (Metal has no geometry stage). |
Requirements (Windows)
- NVIDIA GPU with a recent driver (tested on an RTX 5060 Ti and an RTX 3070 Ti Laptop). AMD and Intel are experimental: device loss has been reported (#3, #84, #99, #124).
- 16 GB of RAM plus a large page file (the process commits about 14 GB), 32 GB is comfortable.
- A controller is recommended.
Known issues
- Device lost on several AMD cards without a fault address (#3, #84, #99, #124), and a GPU hang on an RX 6600 under RADV (#122).
- A crash on Linux when walking into the tunnel from the Lagoon Layer to the Hilltop Layer (#126).
- A frame limit below 60 on 1.0.0 only holds with "Game speed follows the clock" on; without it the game keeps its own timing (#118).
- The first visit to an area can stutter. Open-area performance is 31 to 35 fps on an RTX 3070 Ti Laptop; a heavy scene on an RTX 5060 Ti measured 51.8 fps with a 1% low of 20.6 fps.
- The cutscene at the end of the tutorial lost the device on NVIDIA before a workaround (#75); it is on by default and the cause is not known.
- The NVIDIA workaround for #9 skips light quads, which hides some particles (#62).
- Checkerboard textures (#5), one-sided sound in the first 20 seconds (#63), #76, #83, #112 and unrecorded NVN 2 calls (#6).
Reporting a bug
Open an issue with the log (dkb.log), the build line at its top (dkb 0.2.0+), your GPU and driver, and what you did before it failed. Never share generated files, keys or screenshots that show game content.
Roadmap: #10. Testers: #3 (other GPUs) and #7 (play further).
What's Changed
- Keep the MoltenVK upscaler option out of the performance section and fix the rt_fpcr_sync comment by @LiamTheCreat0r in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/114
- Add the packed-half reduction and the halfword store to storage candidates by @dwk001 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/115
- Hash a shader's SPIR-V once and keep it with the shader, so the MoltenVK upscaler check no longer hashes nearly every dispatch by @LiamTheCreat0r in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/119
- Skip the re-upload of an untracked texture whose storage bytes are unchanged, found by a hash taken before each upload by @LiamTheCreat0r in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/120
- Cover 1D stores in the storage-image 1D fix and tighten the translator tests by @dwk001 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/121
- Fix the map's cyan wedges on AMD, and a few installer and dump fixes by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/125
- Keep NaN out of fragment colour outputs, so the bloom cannot spread it into glowing bands by @GlebShekovtsov in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/132
New Contributors
- @GlebShekovtsov made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/125
Full Changelog: https://github.com/Lucas-blip-png/dkb-pc-recomp/compare/v0.1.0...v0.2.0
0.1.0
Full changelog
First release of DKB Recomp: a static recompilation of Donkey Kong Bananza (Switch 2) to a native Windows exe with a Vulkan renderer. It is experimental (0.x): it plays the tutorial, its cutscene and the first stage on recent NVIDIA cards, and it is not finished.
The packages hold no game data, and no release ever will. You build the game on your own PC from your own dump of the base 1.0.0 game. The 458752 update and the DLC are not supported.
Download
| Package | For | Status |
|---|---|---|
DKB-Setup--windows.zip |
Windows 10/11 x64 | Supported. Carries its own tools and the installer window: unpack, run DKB-Setup.exe, pick your dump. |
DKB-Setup--linux-x86_64.tar.gz |
Linux x86-64, kernel 6.7 or later | Preview. Source package with setup.sh: you install the system packages it lists (./setup.sh --check), then it runs the command-line installer. |
DKB-Setup--macos-arm64.zip |
macOS 14.4 or later, Apple silicon | Preview. Same idea with setup.command. Play is not verified and the destructible terrain is missing (Metal has no geometry stage). |
Requirements (Windows)
- NVIDIA GPU with a recent driver (tested on an RTX 5060 Ti and an RTX 3070 Ti Laptop). AMD and Intel are experimental: device loss has been reported (#84, #99).
- 16 GB of RAM plus a large page file (the process commits about 14 GB), 32 GB is comfortable.
- A controller is recommended.
Known issues
- The first visit to an area can stutter. The game imports memory on first use now, which removed most of the long stalls, but shader and pipeline work on a new area can still take a second or more.
- The cutscene at the end of the tutorial lost the device on NVIDIA before a workaround (#75). The workaround is on by default and passed on an RTX 5060 Ti; the cause is not known and other cards are untested.
- The NVIDIA workaround for #9 skips light quads, which hides some particles (#62).
- Checkerboard textures (#5), one-sided sound in the first 20 seconds (#63), a darker opening wall, #76, #83 and unrecorded NVN 2 calls (#6).
- Open-area performance is 31 to 35 fps on an RTX 3070 Ti Laptop (replay cost).
Reporting a bug
Open an issue with the log (dkb.log), the build line at its top (dkb 0.1.0+), your GPU and driver, and what you did before it failed. Never share generated files, keys or screenshots that show game content.
Roadmap: #10. What this release needed: #85. Testers: #3 (other GPUs) and #7 (play further).
What's Changed
- Check staged Git blobs in the content guard by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/13
- Test memory limits against the configured pool by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/14
- Record deferred NVN 2 push constants at submission by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/15
- Include the Bananza patch table in the build instructions by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/16
- Drain the replay queue up to the items queued before the call by @NotShinySource in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/23
- Preserve serialized replay for compute overlap mode by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/18
- Count guest ticks at 31.25 MHz for Switch 2 games by @NotShinySource in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/25
- Give Switch 2 timestamp reports the time of their submission by @NotShinySource in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/26
- Reflect every location consumed by shader interface arrays by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/21
- Retain IPC sessions until their requests finish by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/17
- Wait for sync completion before preparing consumer submissions by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/28
- Install Bananza from verified ExeFS and RomFS folders by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/19
- Protect retired GPU resources during command recording by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/29
- Add optional defined shader min/max and saturation by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/27
- Publish submission timestamp tags and ticks as coherent pairs by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/30
- Keep failed shader requests for local offline replay by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/32
- Use SEH-aware setjmp on Windows by @zswxp59 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/33
- Pack compute root tables using each NVN 2 program layout by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/20
- Restore failed shader replay tests after compute c0 changes by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/35
- Replay partial stencil clears while preserving depth and other bits by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/36
- Keep submission timestamps alive through replay completion by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/37
- Guard Windows guest exits against unframed CRT jumps by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/34
- Do not wait for the GPU on small Switch 2 reads of rendered targets by @NotShinySource in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/24
- Decode ASTC textures during buffer-to-image copies on the GPU compute path by @Kvinkxerli in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/38
- Record ASTC replay breadcrumbs on the recorder thread by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/40
- Restore the indirect dispatch fixture after profiling was added by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/41
- Replay compatible decoded ASTC texture copies by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/43
- Ignore the per-vertex array dimension in geometry input reflection by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/47
- Resolve descriptors through their active texture and sampler pools by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/46
- Resolve compute storage bases with guarded runtime candidates by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/42
- Reset Bananza indirect records in replay order by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/48
- Require an explicit ordinary geometry shader translation contract by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/49
- Cache ordinary geometry shaders with an explicit topology contract by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/50
- Read shader contracts only from unique root fields by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/51
- Keep capturing after conditionally predicated native exits by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/53
- Report conservative geometry attribute accesses from Maxwell source by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/57
- Translate portable geometry passthrough from explicit skip masks by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/54
- Log bounded geometry draw context without marking buffer use by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/58
- Replay ordinary geometry shaders behind an experimental flag by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/52
- Accept typed geometry subgroup system inputs by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/60
- Add tools/extract_nup.py: write the ExeFS and RomFS folders from a NUP container by @dwk001 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/59
- Exercise geometry writes and indirect reads in one submission by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/56
- Replay portable geometry with source-checked cache entries and owned controls by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/55
- Capture bounded NVN 2 command mode lifecycles by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/61
- Test recorder payloads when the fragment descriptor set is empty by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/64
- Show DKB Recomp in the settings menu's version line by @dwk001 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/65
- Order splitter-fed submixes before their destinations by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/66
- Saturate Ampere 8-bit F2I to signed or unsigned bytes in the translator by @zswxp59 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/72
- Restore the SPIR-V interface test fixture fields by @monsterlag in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/71
- Apply NVN's alpha test in the fragment shader by @philneitz in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/68
- Note the two alpha test cases that follow Eglantine by @philneitz in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/73
- Look up written ranges through a hash table by @philneitz in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/74
- Replay native passthrough geometry stages behind TOTK_NATIVE_GS by @philneitz in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/67
- Let code after an unpredicated call run, and accept RGB-mask TXD in the Ampere lowering by @dwk001 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/78
- Allow passthrough geometry shaders to read unforwarded inputs by @dwk001 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/81
- Index pending GPU writes during shadow memory sync by @dwk001 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/82
- Reorder display settings, clarify resolution labels, and disable window size in fullscreen by @oshoubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/86
- Key the alpha test reference to the nearest 1/255 by @philneitz in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/90
- building.md: add the Fedora 44 package list and the NativeAOT notes by @martingrahn-cmd in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/91
- Add opt-in dynamic frame timing for Bananza by @oshoubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/89
- Add Bananza 21:9 and 32:9 support with HUD and title-screen fixes by @oshoubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/87
- Add dkb.exe --version and log the version; set VERSION to 0.1.0 by @Lucas-blip-png in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/94
- make_release.py: add --game dkb for the Bananza release folder and DKB-Setup zip by @Lucas-blip-png in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/95
- Translate every shader stage of the game's shader files up front with --prepare-shaders by @dwk001 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/79
- Add opt-in direct third-person mouse camera controls by @oshoubber in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/88
- Release zip: ship the translator identity inputs and the shader census so the install runs by @Lucas-blip-png in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/96
- Preserve Bananza voxel varyings on AMD Windows by @MamamaNanana-dot in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/104
- Lower the Ampere atomic, texture and load forms that the NVN 2 path still rejected by @dwk001 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/102
- Add a controlled performance harness that writes an evidence bundle for each dkb.exe run by @dwk001 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/103
- Build and boot on Apple silicon macOS with MoltenVK by @LiamTheCreat0r in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/98
- Follow up the reviews of #102 and #103 by @dwk001 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/105
- Lower a 2D texel fetch with an address offset to TLDS target 4 by @dwk001 in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/108
New Contributors
- @monsterlag made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/13
- @NotShinySource made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/23
- @zswxp59 made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/33
- @Kvinkxerli made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/38
- @dwk001 made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/59
- @philneitz made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/68
- @oshoubber made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/86
- @martingrahn-cmd made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/91
- @MamamaNanana-dot made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/104
- @LiamTheCreat0r made their first contribution in https://github.com/Lucas-blip-png/dkb-pc-recomp/pull/98
Full Changelog: https://github.com/Lucas-blip-png/dkb-pc-recomp/commits/v0.1.0
Reply on the forum
A question or a hands-on report? Open the discussion on the forum.