-
Notifications
You must be signed in to change notification settings - Fork 37
symm op #1777
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
symm op #1777
Conversation
src/Ops.jl
Outdated
| location, | ||
| ), | ||
| ) | ||
| return res |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
construct a TracedRArray before returning (see the dot_general case below)
src/Ops.jl
Outdated
| C.mlir_data, | ||
| alpha.mlir_data, | ||
| beta.mlir_data; | ||
| output=mlir_type(TracedRArray{eltype(C),length(ressize)}, ressize), |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
eltype(c) make this into unwrapped_eltype
src/stdlibs/LinearAlgebra.jl
Outdated
| β::Number=true, | ||
| ) | ||
| # Promote to traced arrays | ||
| A = call_with_reactant(Reactant.promote_to, TracedRArray, A.data) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
| A = call_with_reactant(Reactant.promote_to, TracedRArray, A.data) | |
| A = call_with_reactant(Reactant.promote_to, TracedRArray, parent(A)) |
| @nospecialize(C::TracedRArray{T,2} where {T}), | ||
| @nospecialize(A::Symmetric), | ||
| @nospecialize(B::AbstractMatrix), |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Also add the version where B is Symmetric
* feat: julia api to access device properties [skip ci] * fix: apply suggestion from @avik-pal [skp ci] * chore: bump reactant_jll * fix: remove deleted fields [skip ci] * chore: bump jll
* actually run code for symlinking ptxas, fatbinary, libdevice (the condition `build_backend == "cuda"` was now always false) * check the source of the symlinks exists, instead of creating broken symlinks, if that part of the code ever runs (which at the moment it doesn't because of point above) * rework backend handling, to use more structure instead of string comparisons (which lead to issues like the first point)
* Regenerate MLIR Bindings * fix: kernel call --------- Co-authored-by: enzyme-ci-bot[bot] <78882869+enzyme-ci-bot[bot]@users.noreply.github.com> Co-authored-by: Avik Pal <avikpal@mit.edu>
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>
* feat: new jll version + new compiler passes * fix: use newer libtpu builds * fix: try always downloading libtpu on ci * feat: version check for libtpu
* test: version 1.12 Comment out older Go versions in CI matrix * test: v1.12 * ci: disable downgrade * fix: get 1.12 KAExt loading to work * Mildly functioning on 1.12 * ci * fix 1.10 * Change Julia version from 1.12 to 1.10 * chore: run formatter --------- Co-authored-by: William S. Moses <gh@wsmoses.com> Co-authored-by: William Moses <wmoses@google.com>
|
closing in favor of #1798 |
No description provided.