loopy
loopy copied to clipboard
inducer
A code generator for array-based code on CPUs and GPUs
This has not yet made the jump past https://github.com/inducer/loopy/pull/222/: https://github.com/inducer/loopy/blob/ee36f4c5333a60aeb8280c6cab90f5c4801720dc/loopy/statistics.py#L75-L76
Some issues were found in #222 pre-merge that we need to get to. Here are the places where they're listed: - [ ] #222 - [ ] https://gitlab.tiker.net/inducer/loopy/-/merge_requests/426 cc @kaushikcfd
#162 has made loopy quite strict in its bounds checking. (cf. https://github.com/inducer/meshmode/issues/81) That's probably a good direction, but we need to - [ ] improve what `assume` can do in...
@inducer, taking a look here I'm realizing there's a bit of tidying that could be done (before diving into the actual implementation). Aside from the dependency on [gl-284](https://gitlab.tiker.net/inducer/loopy/-/merge_requests/284) (whose changes...
I would like to merge TJs work on introduction of a C vector target in order to enable vectorisation. PyOP2 and firedrake pass all tests with this branch. If possible...
This PR intends to assign a more accurate set of inames for the assignment instructions added for CSE node.
- git-blame shows was deprecated in 2016: https://github.com/inducer/loopy/commit/94361fd75f742fd5138b922f6f884e42a68e1f4e. - I will wait doing this until #222 lands in as none of our other routines use this API.
(Applies to #222) cc @kaushikcfd
To be able to call `assignment_to_subst` on ``` a[i, 0] = f(i) a[i, 1] = g(i) ``` cc @jdsteve2
Consider: ```python knl = lp.make_kernel( "{[i,j, l]: 0