Minutes_2018_05_17
Stan Seibert edited this page May 17, 2018
·
1 revision
Attendees: Ehsan, Todd, Siu, Stuart, Stan
- #2972: [Feature Request] Support array.conj()
- Easy fix
- #2971: Recursive function with optional types
- Siu will take a look
- #2965: CUDA target needs error message overhaul similar to CPU
- Makes it hard to interpret errors
- #2964: @{gu,}vectorize docs should mention promotion/dispatch behaviours.
- Need to do until we replace guvectorize
- Easy to add
- #2962: numba.errors.LoweringError: Failed at object (object mode frontend)
- A lot of unsupported features. Lambda with conditional in use.
- #2961: Generators within @jitclass that yield vectors produce segmentation faults bug
- Siu will take a look
- #2960: Feature request: optimize permutable nested prange
- Not possible to solve in general
- Polyhedral analysis in LLVM help?
- prange makes this extra hard
- We should document the nested prange behavior
- Suggest VTune as an option for advanced analysis
- #2958: Device function declaration order error message
- take out the type annotations to allow defered compilation
- Need FAQ on why not to use explicit type annotations
- Siu will reply
- #2956: Passing a jitted function reference to a jitted function makes cache=True fail
- Can this limitation be lifted?
- caching in general has issues with pointers to other things
- Siu will think about this
- #2955: "AssertionError: Failed at object (object mode frontend)" exception from simple code
- Need to put things back if compiler pipeline fails
-
#2968 WIP: Adding xnd kernels
- Siu giving author feedback on approach
- This might work for ragged arrays now??
-
#2963 Improve alias analysis to be more comprehensive
- Ehsan will update tests
-
#2959 np.unique return_counts support
- Doesn't work as is
- Provided feedback on how to fix
-
#2957 Add new 3.7 opcode support.
- Not being automatically tested yet since we need llvmlite for Python 3.7
- Stuart will add object mode test that produces these opcodes
-
#2953 percentile and nanpercentile
- reviewed, waiting on author fixes
-
#2950 Fix dispatcher to only consider contiguous-ness.
- Stuart to talk to Siu about read-only array case
-
#2942 Fix linkage nature (declspec(dllexport)) of some test functions
- Should maybe generalize this to cover other places
-
#2939 comprehensive SVML test
- Stuart will fix with autopep8 and merge it
-
#2930 Add configuration file and color schemes.
- Merged
-
#2929 Add Flag To Prevent Unneccessary D->H Copies
- Need to implement decorator to show user how to solve their problem without putting a global default copy strategy into code
-
#2928 [WIP] Making error about untyped list more informative.
- Check with Siu as to whether this is ready to review
- #2910: [WIP] More CUDA intrinsics
- Check if it works on GPU
- Make sure it goes thru the buildfarm before merging
- Blocked on how to simulate in CUDA simulator. See recent njwhite patches?
- #2905: Fix for #2862
- need to replicate old incorrect behavior in NumPy <= 1.11
- #2897: [WIP] Fix line position of delete statement in numba ir
- Stuart will review
- #2894: [WIP] Implement jitclass default constructor arguments.
- Need to give some feedback to contributor on approach
- #2869: WIP: use pairwise summation in
sum
- Producing wrong results in some cases
- Open new PR with Stuart's non-recursive implementation
- #2860:
__cuda_array_interface__
- Two small doc fixes, and then ready to merge
- Stan will merge
- #2840: Support list of refcounted types
- Problems found in review that need to be fixed
- Siu will take a look
- #2817: [WIP] Emit LLVM optimization remarks
- Want to refactor to not require temp file on disk
- #2793: Simplify and remove javascript from html_annotate templates.
- Need to push some refactoring onto the PR
===========================
-
Numba in Anaconda defaults channel status
- tag llvmlite 0.23.1
- ask distro team to build
-
Preparing for Python 3.7
- _
-
Stuart to add an issue full of things about Parfors that seem to be common questions with view of turning this into a FAQ.
-
Add PEP8 testing in CI for PR?
- Python 3.7 support
- ARM fixes
- Reference-counted items (arrays lists) in lists
- Python mode blocks in nopython mode
- HTML annotation improvements