Conversation
* Supports random number generation for all relevant types. Includes uniform, normal and exponential random numbers. --------- Co-authored-by: David Krasowska <krasowska89@gmail.com>
* Add Diagonal / UniformScaling support for NDArray * NDArray <: AbstractArray now
* support more unary/binary ops --------- Co-authored-by: krasow <krasow@u.northwestern.edu>
* support var, std, argmin, argmax --------- Co-authored-by: David Krasowska <krasowska89@gmail.com>
* Add cholesky and eigen, split out batched linalg
…unching patches (#174)
* Add GPU fft/ifft/fft!/ifft! and batched FFT on NDArray
* Add pairwise tensor contract with mode labels. Wrap cupynumeric's binary contract kernel and the helpers it needs so a later TensorOperations backend can map Index2Tuple onto contract! without einsum strings.
* sort, sort!, searchsorted --------- Co-authored-by: krasow <krasow@u.northwestern.edu> Co-authored-by: David Krasowska <krasowska89@gmail.com>
Match Julia array equality (including mixed dtypes and shape mismatch) without falling back to cupynumeric's length-1 1-d array_equal result.
Codecov Report❌ Patch coverage is
📢 Thoughts on this report? Let us know! |
* tensor oeprations extension * force tensors which reduce to scalars to hit scalar indexing
* revamp benchmark suite * automate selection of N * fix bug in 0D benchmark correctness, add more logging * int overflow bug * add execution fences on certain benchmarks
#193) * Make Julia array conversion owning
* Enable distributed cuSolverMp solve and factorization paths
[container] [skip dev] [skip jll]
* mapreduce
* ci: update matrix to include Julia 1.13 * fix deprecated enum CxxWrap syntax in wrapper * bump wrapper ver --------- Co-authored-by: Ethan Meitz <54505069+ejmeitz@users.noreply.github.com> Co-authored-by: ejmeitz <emeitz@andrew.cmu.edu>
* benchmark harness changes [skip ci] * format: run all files * instantiate project script with model smoketest * smoke test for montecarlo now checks correctness for all models if correctness is enabled * reduce verbose output and provide progress meter for all models * update benchmarks_smoke.toml [skip ci] * montecarlo: default accelerate macro * montecarlo: Dagger.jl * montecarlo: map reduce cuda.jl * benchmarks: add gemm support for Dagger and JACC * benchmark: local options need to override global * benchmarks/cleanup: remove dead code * benchmarks: JACC and dagger grayscott * benchmarks: provide seperate .toml to run different forms of accelerate on grayscott * grayscott: add correctness checks * grayscott: default is accelerate * grayscott: simplfy materialization * fix rebase conflicts [container] [benchmark-container] * CG + kwarg support (#198) * CG + kwarg support * add accelerated variant * benchmarks: cg port to Dagger and Python. Change default behaviour of CG on cuNumeric * benchmarks: update Dagger CG to reduce runtime overheads --------- Co-authored-by: krasow <krasow@u.northwestern.edu> * fake change to trigger [container] * benchmarks: JACC needs to set backend for CUDA [skip ci] * benchmarks: fix auto-size conservativeness * Benchmark Container CI (#199) [container] [benchmark container] * benchmarks: JACC CG updates. reduce copies * benchmarks: Dagger grayscott init fixes * benchmarks: Dagger CG update to support multi partitions for multi-gpu scaling. [benchmark container] * Trigger benchmark image [benchmark-container] * benchmarks: NAS FT [benchmark-container] * benchmarks: NAS/EP ported * benchmarks: montecarlo uses map_reduce; original impl is left as a comparison to see effect of map_reduce * benchmarks: NAS/FT patches * benchmarks: NAS/MG init port * benchmarks: codex audit on fairness * move benchmarking harness to seperate repo --------- Co-authored-by: Ethan Meitz <54505069+ejmeitz@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Features:
LinearAlgebra.DiagonalNDArray <: AbstractArrayargmin,argmax,var,std,mean<<)eigen)@acceleratemacro which does escape analysis and multi-line broadcast fusionsort,sort!,searchsortedfirst,searcsortedlast@tensormacromapreduceTODO:
lib/cunumeric_wrapper_jll