gpu-dawn: linux aarch64 support #194
Labels
No labels
CI
all
basisu
blog
bug
build
contributor-friendly
core
correctness
deferred
dev
direct3d-headers
docs
driver-os-issue
duplicate
dxcompiler
editor
examples
experiment
feature-idea
feedback
flac
freetype
gamemode
gkurve
glfw
gpu
gpu-dawn
harfbuzz
help welcome
in-progress
infrastructure
invalid
libmach
linux-audio-headers
long-term
mach
mach.gfx
mach.math
mach.physics
mach.testing
model3d
needs-triage
object
opengl-headers
opus
os/linux
os/macos
os/wasm
os/windows
package-manager
priority
proposal
proposal-accepted
question
roadmap
slipped
stability
sysaudio
sysgpu
sysjs
validating-fix
vulkan-zig-generated
wayland-headers
website
wontfix
wrench
www
x11-headers
xcode-frameworks
zig-update
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
hexops/mach#194
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
While trying to run
zig build run-exampleon Debian ARM64 in a Chromebook Crostini VM I get:and adding -Ddawn-from-source=true to this command or trying to set from_source to true any other way doesn't fix it.
Update: when going directly in the libs directory version of mach-gpu-dawn, it does accept the dawn-from-source option as in
zig build -Ddawn-from-source=true, it's just not passed on from running the example. Now sleep, will try more later.Thanks for filing! I filed https://github.com/hexops/mach/issues/196 for the
dawn-from-source=truebug.Currently, aarch64-linux is not supported, I'd like to get it working soon and it shouldn't be too hard. What needs to be done:
5a231eebf1be120ba7d904675f08e1464f/glfw/system_sdk.zig#L47-L49)
gpu-dawn/build.zigto utilize those binary releasesgithub.com/hexops/mach@a2f65d5a23/gpu-dawn/build.zig (L194)For step one: https://github.com/mitchellh/mach-sdk-linux-aarch64 Feel free to grab that and pull it into the
hexopsorg.Fwiw, with the two steps already taken, you can now run
zig build run-dawn-example -Ddawn-from-source=truein thegpu-dawndirectory and this all just works on aarch64 Linux. I tested locally. Hopefully binary builds can come though :)Plan for binary builds: GitHub actions doesn't have ARM runners, but we can just cross-compile. Should be easy, I'll get this working within a few days.
I tried but couldn't build 'run-dawn-example' from source on a Raspberry PI 400 (Ubuntu 22). I have a working zig environment there and can cross-compile simple stuff. But the Dawn build stops with a bunch of errors complaining about undefined atomics symbols:
Other people have reported similar problems here: https://github.com/ziglang/zig/issues/10086#
There is an open PR which may solve the atomics issues with linux aarch64: https://github.com/ziglang/zig/pull/11828#. Hopefully it will be merged eventually.
In theory this probably all works now, but I don't think anyone has tested it / confirmed it does.
Getting an aarch64-linux machine to test on has been very annoying, I had bought a ROCK pi 5B for this but just last week it randomly died and I don't know why (nvme failure perhaps.)
If someone can test it works or not, that'd be helpful information in pushing support of this along.
The way to test:
Suspect that might be an issue with running on the nightly? Otherwise it doesn't show the previous issue anymore.
added to #811
with 0.10.1 I get
/home/ikojba/mach/libs/gpu/build.zig:7:21: error: root struct of file 'std' has no member named 'Build'mach tracks master so
0.10.1shouldn't work at all. all you have to do is changingstd.builtin.Versiontostd.SemanticVersionin this lineNow just some "Missing struct field: .patch" errors...
std.SemanticVersion{ .major = 12, .minor = 0, .patch = 0 };After fixing #811 this seems to work flawlessly :D
Works as in you ran on an arm device and saw graphics?
Yes, exactly that!
On June 21, 2023 at 4:58 GMT, Stephen Gutekanst @.***> wrote:
Works as in you ran on an arm device and saw graphics?
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you modified the open/close state.Message ID: @.***>