-
Notifications
You must be signed in to change notification settings - Fork 433
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Couple of changes in WGSL layer #1231
Conversation
2. ONNX IR fix ReduceL* 3. ignoring ipynb checkpoints
# Conflicts: # burn-book/src/building-blocks/tensor.md
# Conflicts: # burn-autodiff/src/ops/tensor.rs # burn-candle/src/lib.rs # burn-candle/src/ops/tensor.rs # burn-fusion/src/ops/float.rs # burn-fusion/src/stream/operation.rs # burn-ndarray/src/ops/tensor.rs # burn-tch/src/ops/tensor.rs # burn-wgpu/src/ops/float_ops.rs
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The bug regarding num_data needs to be fixed!
@@ -28,7 +28,11 @@ fn main( | |||
|
|||
if id_local == 0u { | |||
var sum = {{ elem }}(0); | |||
for (var i: u32 = 0u; i < WORKGROUP_SIZE; i++) { | |||
var num_data = min(WORKGROUP_SIZE, arrayLength(&input) % WORKGROUP_SIZE); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
From myself on discord: "it has to depend on the invocation id, because now imagine you have workgroup_size=1000 ans input length is 2500. You will want 1000, 1000, 500 but the formula with min and modulo will give 500, 500, 500"
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Error in CI is likely to be caused by that
I assume this PR is blocked? |
Pull Request Template
Checklist
run-checks all
script has been executed.Related Issues/PRs
The changes are issues identified during #1173
Changes
Fixes a couple of edge cases with sum/reduction WGSL layer.
Testing
"run-checks all"
TBD