-
Notifications
You must be signed in to change notification settings - Fork 190
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
[FEA]: Validate cuda.parallel type matching in build and execution #2429
Conversation
🟨 CI finished in 6h 01m: Pass: 97%/437 | Total: 2d 12h | Avg: 8m 18s | Max: 1h 27m | Hits: 99%/41645
|
Project | |
---|---|
+/- | CCCL Infrastructure |
libcu++ | |
CUB | |
Thrust | |
CUDA Experimental | |
pycuda | |
CUDA C Core Library |
Modifications in project or dependencies?
Project | |
---|---|
+/- | CCCL Infrastructure |
+/- | libcu++ |
+/- | CUB |
+/- | Thrust |
+/- | CUDA Experimental |
+/- | pycuda |
+/- | CUDA C Core Library |
🏃 Runner counts (total jobs: 437)
# | Runner |
---|---|
320 | linux-amd64-cpu16 |
66 | linux-amd64-gpu-v100-latest-1 |
28 | linux-arm64-cpu16 |
23 | windows-amd64-cpu16 |
That might conflict with the host standard library
…imental/__init__.py and fix bug in python/cuda_parallel/tests/test_reduce_api.py
🟩 CI finished in 1h 15m: Pass: 100%/368 | Total: 1d 11h | Avg: 5m 42s | Max: 1h 06m | Hits: 99%/25663
|
Project | |
---|---|
+/- | CCCL Infrastructure |
libcu++ | |
CUB | |
Thrust | |
CUDA Experimental | |
pycuda | |
CUDA C Core Library |
Modifications in project or dependencies?
Project | |
---|---|
+/- | CCCL Infrastructure |
+/- | libcu++ |
+/- | CUB |
+/- | Thrust |
+/- | CUDA Experimental |
+/- | pycuda |
+/- | CUDA C Core Library |
🏃 Runner counts (total jobs: 368)
# | Runner |
---|---|
297 | linux-amd64-cpu16 |
28 | linux-amd64-gpu-v100-latest-1 |
28 | linux-arm64-cpu16 |
15 | windows-amd64-cpu16 |
Hi @gevtushenko, I think I addressed what we discussed this morning, although I'm not sure: is there anything else you want me to add to this PR? |
🟩 CI finished in 1h 00m: Pass: 100%/368 | Total: 1d 21h | Avg: 7m 23s | Max: 42m 53s | Hits: 81%/25679
|
Project | |
---|---|
+/- | CCCL Infrastructure |
libcu++ | |
CUB | |
Thrust | |
CUDA Experimental | |
pycuda | |
CUDA C Core Library |
Modifications in project or dependencies?
Project | |
---|---|
+/- | CCCL Infrastructure |
+/- | libcu++ |
+/- | CUB |
+/- | Thrust |
+/- | CUDA Experimental |
+/- | pycuda |
+/- | CUDA C Core Library |
🏃 Runner counts (total jobs: 368)
# | Runner |
---|---|
297 | linux-amd64-cpu16 |
28 | linux-amd64-gpu-v100-latest-1 |
28 | linux-arm64-cpu16 |
15 | windows-amd64-cpu16 |
# example-begin reduce-min | ||
def min_op(a, b): | ||
return a if a < b else b | ||
# example-end reduce-min |
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.
important: the test_*_api
files are used for documentation purposes. They illustrate usage of API. Sphinx essentially copy-pastes code between the example-begin reduce-min
and example-begin reduce-min
labels. Here's an example how it looked like. This cange makes it so that our reference API documentation only contains min_op
definition. Let's revert this change.
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.
Done. Thanks for the explanation!
I only kept some minor changes.
def test_device_reduce_dtype_mismatch(): | ||
dtypes = [numpy.int32, numpy.int64] | ||
h_inits = [numpy.array([], dt) for dt in dtypes] | ||
h_inputs = [numpy.array([], dt) for dt in dtypes] | ||
d_outputs = [cuda.device_array(1, dt) for dt in dtypes] | ||
d_inputs = [cuda.to_device(h_inp) for h_inp in h_inputs] | ||
|
||
reduce_into = cudax.reduce_into(d_inputs[0], d_outputs[0], min_op, h_inits[0]) | ||
|
||
for ix in range(3): | ||
with pytest.raises(TypeError, match=r"dtype mismatch: __init__=int32, __call__=int64"): | ||
reduce_into(None, d_inputs[int(ix == 0)], d_outputs[int(ix == 1)], h_inits[int(ix == 2)]) |
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.
important: this file's purpose is testing code snippets in our docs. Let's put functionality tests into test_reduce.py
.
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.
Done.
Interesting, so this is akin to the builder/executor distinction in nvmath-python's stateful APIs, where a similar check also exists: |
…phinx documentation. Also fix existing typo.
I was looking for some way to assert that what's passed to cccl/python/cuda_parallel/cuda/parallel/experimental/__init__.py Lines 101 to 110 in 1cfe171
I figure I have to get into that code when I make |
🟩 CI finished in 1h 13m: Pass: 100%/370 | Total: 1d 15h | Avg: 6m 19s | Max: 35m 40s | Hits: 99%/25712
|
Project | |
---|---|
+/- | CCCL Infrastructure |
libcu++ | |
CUB | |
Thrust | |
CUDA Experimental | |
pycuda | |
CUDA C Core Library |
Modifications in project or dependencies?
Project | |
---|---|
+/- | CCCL Infrastructure |
+/- | libcu++ |
+/- | CUB |
+/- | Thrust |
+/- | CUDA Experimental |
+/- | pycuda |
+/- | CUDA C Core Library |
🏃 Runner counts (total jobs: 370)
# | Runner |
---|---|
297 | linux-amd64-cpu16 |
30 | linux-amd64-gpu-v100-latest-1 |
28 | linux-arm64-cpu16 |
15 | windows-amd64-cpu16 |
Description
Closes: #2416
Checklist