You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
wgpu error: Validation Error
Caused by:
In Device::create_shader_module
Shader validation error:
┌─ :62:11
│
62 │ let _24 = arrays_0[_23];
│ ^^^^^^^^^^^^^ naga::Expression [84]
Entry point foo_f32 at Compute is invalid
Expression [84] is invalid
The expression [83] may only be indexed by a constant
on WGPU + SPIRV:
thread 'main' panicked at $PATH/cubecl/crates/cubecl-spirv/src/variable.rs:399:49:
index out of bounds: the len is 0 but the index is 0
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
A "quick and dirty" update to v23 locally fixed the WGPU + WGSL issue for me.
For a proper update a few things must be cleared up:
wgpu::Id isn't a thing anymore, how should WgpuDevice::Existing work now?
Currently if you try to use
comptime! { .. }
to buildArray
s in cube kernels accessing them works fine for CUDA.Using WGPU + WGSL doesn't work. This was recently fixed, the v23 changelog shows:
(See https://github.com/gfx-rs/wgpu/blob/trunk/CHANGELOG.md#contributor-spotlight-sagudev)
On
main
this currently happens on WGPU + WGSL:on WGPU + SPIRV:
A "quick and dirty" update to v23 locally fixed the WGPU + WGSL issue for me.
For a proper update a few things must be cleared up:
wgpu::Id
isn't a thing anymore, how shouldWgpuDevice::Existing
work now?None
be set?The text was updated successfully, but these errors were encountered: