Thunk Change #10
Closed
Thunk Change #10
Conversation
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
Use `[sources]` section of `docs/Project.toml` instead of messing with `LOAD_PATH`.
* fix: inconsistent return dims * test: inconsistent indexing * fix: inconsistent dimensions inside gather getindex
* fix: define getindexing into sub reshaped array * test: failure case
Co-authored-by: enzyme-ci-bot[bot] <78882869+enzyme-ci-bot[bot]@users.noreply.github.com>
Co-authored-by: enzyme-ci-bot[bot] <78882869+enzyme-ci-bot[bot]@users.noreply.github.com>
…nzymeAD#560) * [CI] Format generated files twice to work around JuliaFormatter bug * [CI] Don't run benchmarks on PRs from forks
* CUDA: fix nv intrinsic errs * Update ext/ReactantCUDAExt.jl Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Restrict running different workflows based on the touched files. This should _slightly_ reduce CI workload.
* respect scopping rules in for * `@isdefined` * Update ReactantCore.jl * fix
Co-authored-by: enzyme-ci-bot[bot] <78882869+enzyme-ci-bot[bot]@users.noreply.github.com>
* feat: support arbitrary structures in while loop * test: add nested structure test * feat: if conditions can now handle arbitrary strucutres * fix: mutation support inside if condition * fix: restore while support * Update src/TracedUtils.jl
* profiler: Add option to generate perfetto url * remove sort * Update api.md
* More jll/cuda stuff * Apply suggestions from code review Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> * force xla * add set arg attr * Update WORKSPACE * fix support * Now with noalias * Update ext/ReactantCUDAExt.jl Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> * Update Project.toml * we are now sane * now with alignment * Update src/Profiler.jl Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: enzyme-ci-bot[bot] <78882869+enzyme-ci-bot[bot]@users.noreply.github.com>
* Allocator stats docs * Add API to create profiler annotations * profiling tutorial * sig cleanup * vitepress config * vitepress config 2 * note about allocatorstats
* PTX fma and other flags * we can keep cuda debug info now, that's cool * version bump * Update XLA.jl * Update src/XLA.jl Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* [CI] Move tests on aarch64 linux to GitHub Actions * Fix name of aarch64-linux runners Co-authored-by: Ian McInerney <mcianster@gmail.com> * Skip CUDA integration tests that could fail --------- Co-authored-by: Ian McInerney <mcianster@gmail.com>
* Fix condition to skip CUDA tests on aarch64 * Skip more tests
* feat: expose more XLA GPU options to the user * feat: check for bazel install * fix: load env vars before * Update deps/build_local.jl Co-authored-by: Mosè Giordano <765740+giordano@users.noreply.github.com> * Update Project.toml --------- Co-authored-by: Mosè Giordano <765740+giordano@users.noreply.github.com> Co-authored-by: William Moses <gh@wsmoses.com>
Co-authored-by: enzyme-ci-bot[bot] <78882869+enzyme-ci-bot[bot]@users.noreply.github.com>
…zymeAD#825) * feat: support implicit padding from XLA * feat: use XLA for shard-info if we need padding * test: padding for sharding * fix: return type
Co-authored-by: enzyme-ci-bot[bot] <78882869+enzyme-ci-bot[bot]@users.noreply.github.com>
* Fix LLVM dialect registration * Add functionality for parsing single operations * Bump Enzyme-JAX commit
macOS runners are occasionally hanging, having a shorter timeout helps by making the job fail earlier than waiting uselessly for 90 minutes, saving some time, especially important since there are only 5 concurrent macOS runners across the entire organisation.
* docs: housekeeping + memref dialect * docs: autoadd dialects
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
11 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.
stablehlo.sortOps (stablehlo.sortOps EnzymeAD/Reactant.jl#374)SpecialFunctionssimple functions (SpecialFunctionssimple functions EnzymeAD/Reactant.jl#384)no_nanmode and more passes (feat:no_nanmode and more passes EnzymeAD/Reactant.jl#458)isnanandisfinitefor TracedRNumber (Implementisnanandisfinitefor TracedRNumber EnzymeAD/Reactant.jl#525)should_rewrite_ftforcallandinvokeexpressions, and overlayBase._unique_dims(Splitshould_rewrite_ftforcallandinvokeexpressions, and overlayBase._unique_dimsEnzymeAD/Reactant.jl#505)pathssettings for workflow triggers ([GHA] Addpathssettings for workflow triggers EnzymeAD/Reactant.jl#563)Pkg.instantiate([CI] Remove useless call toPkg.instantiateEnzymeAD/Reactant.jl#576)Enzyme.autodiffTenet + Reactant + Enzyme.gradient broken on last releases EnzymeAD/Reactant.jl#593 (Fix dense elements attribute inEnzyme.autodiff#593 EnzymeAD/Reactant.jl#604)xla/stream_executor/cuda:cuda_compute_capability_proto_cc_implonly on non CUDA (Usexla/stream_executor/cuda:cuda_compute_capability_proto_cc_implonly on non CUDA EnzymeAD/Reactant.jl#646)trim->strip(Replacetrim->stripEnzymeAD/Reactant.jl#661)ClientCompileto pass CUDA data dir ([ReactantExtra] Add argument toClientCompileto pass CUDA data dir EnzymeAD/Reactant.jl#683)LLVMOpenMP_jllto call OpenMP functions (UseLLVMOpenMP_jllto call OpenMP functions EnzymeAD/Reactant.jl#673)similarreturn empty tensors. (makesimilarreturn empty tensors. EnzymeAD/Reactant.jl#632)@tracefunction calls (@tracefunction calls EnzymeAD/Reactant.jl#366)build_local.jlscript ([deps] Some improvements to thebuild_local.jlscript EnzymeAD/Reactant.jl#689)hardware_interference_size([ReactantExtra] Stop removing references tohardware_interference_sizeEnzymeAD/Reactant.jl#704)Type(value)instead oftype(value)(fixType(value)instead oftype(value)EnzymeAD/Reactant.jl#733)remwith only one operand being aConcreteRNumber(Support tracing ofremwith only one operand being aConcreteRNumberEnzymeAD/Reactant.jl#754)mod(Fix implementation ofmodEnzymeAD/Reactant.jl#758)--coloroption for Bazel ([build_local] Add argument to set--coloroption for Bazel EnzymeAD/Reactant.jl#776)Ops.sharding_constraint(feat: addOps.sharding_constraintEnzymeAD/Reactant.jl#798)raisea keyword argument ([Compiler] Makeraisea keyword argument EnzymeAD/Reactant.jl#797)make-bindings.jlto generate MemRef dialect