[FRONTEND] Minor accumulated style and warning fixes (#388)

- Fix some whitespace.
- Make an undeclared dependency on `pytest` explicit.
- Fix deprecated `description-file` use.
- `#ifdef` out a deprecated `PyEval_InitThreads` call.
- Use a slightly different numpy invocation in `test_random.py` to quiet down overflow warnings in tests.
- Fix a deprecated cast in `test_core.py`.
- Suppress a warning about `visit_Constant` in Python 3.9+; we can't migrate yet because it'd break Python 3.6 and 3.7.
- Use chained exceptions for `CompilationError` rather than rolling our own; it makes the error messages nicer.
- Add a `__str__` for `tl.dtype` to make debugging kernels easier; it lets you `print` a dtype to see what type was inferred.
- Fix a few bad escapes.
This commit is contained in:
Madeleine Thompson
2021-12-10 15:19:20 -08:00
committed by GitHub
parent 9def2424ab
commit e575ae3443
10 changed files with 28 additions and 19 deletions

View File

@@ -339,7 +339,8 @@ def test_atomic_rmw(op, dtype_x, mode, device='cuda'):
('float32', 'int32', True)
])
def test_cast(dtype_x, dtype_z, bitcast, device='cuda'):
x = torch.tensor([43.5], dtype=cvt[dtype_x], device=device)
x0 = 43 if dtype_x.startswith('int') else 43.5
x = torch.tensor([x0], dtype=cvt[dtype_x], device=device)
# triton kernel
@triton.jit
@@ -665,4 +666,4 @@ def test_noop(device='cuda'):
def kernel(x):
pass
x = triton.testing.random((1,), dtype=torch.int32, device=device)
kernel[(1, )](x)
kernel[(1, )](x)