diff options
Diffstat (limited to '')
| -rw-r--r-- | results/scraper/box64/52 | 4 | ||||
| -rw-r--r-- | results/scraper/box64/520 | 14 | ||||
| -rw-r--r-- | results/scraper/box64/522 | 22 | ||||
| -rw-r--r-- | results/scraper/box64/525 | 146 | ||||
| -rw-r--r-- | results/scraper/box64/527 | 2 | ||||
| -rw-r--r-- | results/scraper/box64/529 | 10 |
6 files changed, 198 insertions, 0 deletions
diff --git a/results/scraper/box64/52 b/results/scraper/box64/52 new file mode 100644 index 000000000..02f68069c --- /dev/null +++ b/results/scraper/box64/52 @@ -0,0 +1,4 @@ +Failing to pull (unrelated histories) +I'm not sure if its just me but I've gotten a fatal error when pulling due to unrelated histories multiple times with this repo. + +Has there been any reason on your end for this to happen? \ No newline at end of file diff --git a/results/scraper/box64/520 b/results/scraper/box64/520 new file mode 100644 index 000000000..daeb7a12a --- /dev/null +++ b/results/scraper/box64/520 @@ -0,0 +1,14 @@ +80-bit floating-point math on ppc64le +From [this comment](https://github.com/ptitSeb/box64/issues/242#issuecomment-1052656727) by @ptitSeb : + +> SSE/SSE2 instruction are converted t the underling SIMD instruction. Same for x87 that are converted to Float and Double operation (ignoring 80bits operation for most part). + +[According to](https://github.com/FEX-Emu/FEX/issues/2409#issuecomment-1432585400) @awilfox : + +> There is quad-float support in hardware with ISA 3.0 (Power9). + +It might be interesting to support 80-bit x87 math on POWER9 via quad-precision floating-point math, for a speed-up when the default [BOX64_X87_NO80BITS=0](https://github.com/ptitSeb/box64/blob/main/docs/USAGE.md#box64_x87_no80bits-) is used. + +My understanding is that quad float is only hardware-accelerated in POWER9 and higher (so POWER8LE users would need to disable it), and is only aliased to `long double` on a subset of distros (e.g. recent Fedora, grep [Talospace](https://talospace.com/) for articles that mention "long double"). + +It looks to me like the GCC options you want are `-mfloat128 -mfloat128-hardware`, and then use the `__float128` data type instead of `long double` so that it works as expected on older distros. \ No newline at end of file diff --git a/results/scraper/box64/522 b/results/scraper/box64/522 new file mode 100644 index 000000000..7660f7694 --- /dev/null +++ b/results/scraper/box64/522 @@ -0,0 +1,22 @@ +getAlignedMutexWithInit is extremely slow +On a workload I'm testing, Box64 (with #521) is slow. Profiling, an inordinate amount of time is spent within getAlignedMutexWithInit, seemingly due to the application constantly creating and destroying pthread mutexes. That's a silly thing for the application to do, of course, but if I had the application source code to fix, I would just recompile for arm64 :wink: + +The following patch massively improves performance -- 10x fps in some cases -- taking the application from slideshow to a smooth 60fps.. + +``` + pthread_mutex_t* getAlignedMutex(pthread_mutex_t* m) + { +- return getAlignedMutexWithInit(m, 1); ++ return m; +``` + +Of course, this patch is probably wrong, since now we're potentially doing operations on unaligned mutexes, which I assume could blow up at any moment. Though it does seem to work fine for this application, somehow. + +I see potentially two issues here: + +1. box64 might be wrapping mutexes to ensure ABI compatibility in places where it might not actually be necessary (else, I would expect the above patch to blow up spectacularly, maybe I'm just getting lucky) +2. box64's wrapped mutex code itself is inefficient. I don't really understand why the code is doing what it is, so I can't suggest good fixes. I see that NewMutex open-codes an arena allocator based on a global unrolled linked list, I don't know why it does that, if that's supposed to be an optimization over a plain calloc or whether there's some subtle correctness issue here. If this really is the way to go, then `mutexes` should be in **thread-local storage** to eliminate the `mutex_mutexes` locking, which is the heaviest weight operation here, and `taken` should be replaced with a **bitset** to allocate mutexes within the block in constant-time instead of the linear-time search. Neither of these changes should be difficult but I don't want to attempt them before I understand the bigger picture of what this file is for. + +--- + +Bug report aside, it was pretty magical to see this old application come to life on my M1 Linux, with full GPU acceleration and everything. If this is where we are now, I can't wait to see where we'll be once we have 4K pages + TSO sorted in the kernel :smile: \ No newline at end of file diff --git a/results/scraper/box64/525 b/results/scraper/box64/525 new file mode 100644 index 000000000..f199388dd --- /dev/null +++ b/results/scraper/box64/525 @@ -0,0 +1,146 @@ +BalenaOS Jetson Nano | X server Docker Container & Box64 Unity3D Docker Container +Hey. I might be in a special use case here. + +We're trying to deploy graphical interfaces OTA with BalenaOS on ARM architecture. The license from Unity was quite expensive to compile to ARM so before we take that route I thought I'd try with Box64. + +I've got one container managing the X server through this https://github.com/balena-labs-projects/xserver +And another container running the exported unity project. + +But nothing shows up on the display other than a mouse cursor that i can move around once plugged in.. + + +Heres the docker container for the Unity project: + +``` + +FROM balenalib/jetson-nano-ubuntu:latest + +RUN apt update && apt install wget -y + +RUN wget https://ryanfortner.github.io/box64-debs/box64.list -O /etc/apt/sources.list.d/box64.list +RUN wget -O- https://ryanfortner.github.io/box64-debs/KEY.gpg | gpg --dearmor -o /usr/share/keyrings/box64-debs-archive-keyring.gpg +RUN apt update && apt install box64 -y + + +WORKDIR /usr/src/app +COPY JetsonNanoTest . +COPY entry.sh . +RUN chmod +x hmi.x86_64 +RUN chmod +x entry.sh + +RUN apt install libxss1 libwayland-egl1 libwayland-cursor0 libxkbcommon0 libxkbcommon-x11-0 libxcursor1 libxinerama1 libxi6 libxrandr2 libxxf86vm1 -y + + +ENV BOX64_LOG =2 +#ENTRYPOINT ["/bin/bash", "/usr/src/app/entry.sh"] + +CMD [ "box64" ,"/usr/src/app/hmi.x86_64"] +``` + +Docker-compose file: + +``` +version: '2.1' + +volumes: + x11: + + +services: + + xserver: + build: ./Containers/XServer + restart: always + privileged: true + network_mode: host + volumes: + - 'x11:/tmp/.X11-unix' + + emulator: + build: ./Containers/emulator + restart: always + privileged: true + network_mode: host + devices: + - /dev/dri + group_add: + - video + volumes: + - 'x11:/tmp/.X11-unix' +``` + + +The display is default :0 , is there any specific command that Box64 needs to target a display somehow? + + +Logs from startup + + +``` + +[Logs] [2/23/2023, 16:42:13] [emulator] Debug level is 1 +[Logs] [2/23/2023, 16:42:13] [emulator] Dynarec for ARM64, with extension: ASIMD AES CRC32 PMULL PageSize:4096 Cores:4 +[Logs] [2/23/2023, 16:42:13] [emulator] Params database has 15 entries +[Logs] [2/23/2023, 16:42:13] [emulator] Box64 with Dynarec v0.2.1 ce61d27d built on Feb 22 2023 07:15:03 +[Logs] [2/23/2023, 16:42:13] [emulator] Debug level is 1 +[Logs] [2/23/2023, 16:42:13] [emulator] Dynarec for ARM64, with extension: ASIMD AES CRC32 PMULL PageSize:4096 Cores:4 +[Logs] [2/23/2023, 16:42:13] [emulator] Params database has 15 entries +[Logs] [2/23/2023, 16:42:13] [emulator] Box64 with Dynarec v0.2.1 ce61d27d built on Feb 22 2023 07:15:03 +[Logs] [2/23/2023, 16:42:13] [emulator] Using default BOX64_LD_LIBRARY_PATH: ./:lib/:lib64/:x86_64/:bin64/:libs64/ +[Logs] [2/23/2023, 16:42:13] [emulator] Using default BOX64_PATH: ./:bin/ +[Logs] [2/23/2023, 16:42:13] [emulator] Counted 33 Env var +[Logs] [2/23/2023, 16:42:13] [emulator] Looking for /usr/src/app/hmi.x86_64 +[Logs] [2/23/2023, 16:42:13] [emulator] Rename process to "hmi.x86_64" +[Logs] [2/23/2023, 16:42:13] [emulator] Using emulated UnityPlayer.so +[Logs] [2/23/2023, 16:42:13] [emulator] Using native(wrapped) libm.so.6 +[Logs] [2/23/2023, 16:42:13] [emulator] Using emulated /lib/x86_64-linux-gnu/libgcc_s.so.1 +[Logs] [2/23/2023, 16:42:13] [emulator] Using native(wrapped) libpthread.so.0 +[Logs] [2/23/2023, 16:42:13] [emulator] Using native(wrapped) libc.so.6 +[Logs] [2/23/2023, 16:42:13] [emulator] Using native(wrapped) ld-linux-x86-64.so.2 +[Logs] [2/23/2023, 16:42:13] [emulator] Using native(wrapped) libutil.so.1 +[Logs] [2/23/2023, 16:42:13] [emulator] Using native(wrapped) librt.so.1 +[Logs] [2/23/2023, 16:42:13] [emulator] Using native(wrapped) libdl.so.2 +[Logs] [2/23/2023, 16:42:14] [emulator] [UnityMemory] Configuration Parameters - Can be set up in boot.config +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-bucket-allocator-granularity=16" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-bucket-allocator-bucket-count=8" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-bucket-allocator-block-size=4194304" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-bucket-allocator-block-count=1" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-main-allocator-block-size=16777216" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-thread-allocator-block-size=16777216" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-gfx-main-allocator-block-size=16777216" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-gfx-thread-allocator-block-size=16777216" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-cache-allocator-block-size=4194304" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-typetree-allocator-block-size=2097152" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-profiler-bucket-allocator-granularity=16" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-profiler-bucket-allocator-bucket-count=8" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-profiler-bucket-allocator-block-size=4194304" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-profiler-bucket-allocator-block-count=1" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-profiler-allocator-block-size=16777216" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-profiler-editor-allocator-block-size=1048576" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-job-temp-allocator-block-size-background=1048576" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-job-temp-allocator-reduction-small-platforms=262144" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-temp-allocator-size-background-worker=32768" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-temp-allocator-size-job-worker=262144" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-temp-allocator-size-preload-manager=262144" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-temp-allocator-size-nav-mesh-worker=65536" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-temp-allocator-size-audio-worker=65536" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-temp-allocator-size-cloud-worker=32768" +[Logs] [2/23/2023, 16:42:14] [emulator] "memorysetup-temp-allocator-size-gfx=262144" +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libX11.so.6 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libXext.so.6 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libxcb.so.1 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libXau.so.6 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libXdmcp.so.6 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libXcursor.so.1 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libXinerama.so.1 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libXi.so.6 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libXrandr.so.2 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libXrender.so.1 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libXss.so.1 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libXxf86vm.so.1 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libwayland-client.so.0 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libwayland-egl.so.1 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libwayland-cursor.so.0 +[Logs] [2/23/2023, 16:42:14] [emulator] Using native(wrapped) libxkbcommon.so.0 +[Live] Device state settled +``` diff --git a/results/scraper/box64/527 b/results/scraper/box64/527 new file mode 100644 index 000000000..c93866b11 --- /dev/null +++ b/results/scraper/box64/527 @@ -0,0 +1,2 @@ +Gamepads work incorrectly in Unity games +I tried to run Streets of Rogue and Tangledeep. Both games written in Unity and works fine, except gamepad buttons work incorrectly. I tried two gamepads: Xbox Wireless Controller (works properly using https://github.com/atar-axis/xpadneo dkms driver) and GameSir X2 Pro (configured it with xboxdrv). Both work fine in other native games and in Iconoclasts (not a Unity game). And Xbox Wireless Controller works fine on PC with the mentioned two games (and configured the same way via xpadneo). So I suspect that it could be an issue with box64. I use version 0.2. \ No newline at end of file diff --git a/results/scraper/box64/529 b/results/scraper/box64/529 new file mode 100644 index 000000000..6c38ab672 --- /dev/null +++ b/results/scraper/box64/529 @@ -0,0 +1,10 @@ +Steam login menu not working +Hey! + +I used the linux arm gaming chroot guide to run steam, and when I did, I ended up with this: + +I tried looking at the console output, which just complained about cef which I talked about in this issue https://github.com/Raezroth/Linux-ARM-Gaming-Chroot/issues/16 +I tried -nocefsandbox and -oldbigpicture as well and no dice +here's the command I use to launch steam: +MESA_GL_VERSION_OVERRIDE=3.2 PAN_MESA_DEBUG=gl3 steam +open steam://open/minigameslist -noreactlogin +any tips? |