Fix dispatch for rdiv! with LU
#55764
Conversation
dkarrasch
added
performance
|
This will need to be backported manually... |
dkarrasch
removed
backport 1.10
|
For the reproducer in the original issue, I get julia> A = rand(1000,1000); @btime inv($A); @btime inv2($A); @btime inv3($A);
25.634 ms (10 allocations: 8.13 MiB)
38.917 ms (9 allocations: 15.27 MiB)
37.542 ms (9 allocations: 15.27 MiB)
julia> inv(A) ≈ inv2(A) ≈ inv3(A)
true |
|
@jishnub The failures are unrelated, but still concerning. Do you remember if those NaN-ignoring things were known and maybe fixed? Since these tests are random, they may become invisible in another CI run, but we need to handle them, right? |
|
I think these were fixed in #55737 and #55777. The best way to check would be to rebase and rerun the tests locally by specifying the same random seed that had caused the failures, as suggested in JuliaLang/LinearAlgebra.jl#1091 |
|
How do I extract that seed from "my" previous run? Is it printed somewhere in the test logs? Oh, or is it the "global RNG seed" printed after the big table? |
|
Yes, that's the seed |
|
Okay, unfortunately it failed only on Mac, and not on linux, so I can't test it locally. Anyway, let's merge this when tests pass, and assume this was nothing new, but the fixed issue. |
|
But the tests on Linux and Mac must have used different seeds, no? Tests for a specific seed would fall on all platforms if it's due to the issue above. In any case, I think it's fixed now. |
Backported PRs: - [x] #55886 <!-- irrationals: restrict assume effects annotations to known types --> - [x] #55867 <!-- update `hash` doc string: `widen` not required any more --> - [x] #56084 <!-- slightly improve inference in precompilation code --> - [x] #56088 <!-- make `Base.ANSIIterator` have a concrete field --> - [x] #54093 <!-- Fix `JULIA_CPU_TARGET` being propagated to workers precompiling stdlib pkgimages --> - [x] #56165 <!-- Fix markdown list in installation.md --> - [x] #56148 <!-- Make loading work when stdlib deps are missing in the manifest --> - [x] #56174 <!-- Fix implicit `convert(String, ...)` in several places --> - [x] #56159 <!-- Add invalidation barriers for `displaysize` and `implicit_typeinfo` --> - [x] #56089 <!-- Call `MulAddMul` instead of multiplication in _generic_matmatmul! --> - [x] #56195 <!-- Include default user depot when JULIA_DEPOT_PATH has leading empty entry --> - [x] #56215 <!-- [REPL] fix lock ordering mistake in load_pkg --> - [x] #56251 <!-- REPL: run repl hint generation for modeswitch chars when not switching --> - [x] #56092 <!-- stream: fix reading LibuvStream into array --> - [x] #55870 <!-- fix infinite recursion in `promote_type` for `Irrational` --> - [x] #56227 <!-- Do not call `rand` during sysimage precompilation --> - [x] #55741 <!-- Change annotations to use a NamedTuple --> - [x] #56149 <!-- Specialize adding/subtracting mixed Upper/LowerTriangular --> - [x] #56214 <!-- fix precompile process flags --> - [x] #54471 - [x] #55622 - [x] #55704 - [x] #55764
KristofferC
removed
the
backport 1.11
This method (added in #31853) was more specific than the one on line 493, which was added in #31285 at about the same time.
Fixes JuliaLang/LinearAlgebra.jl#1085.