ox-arrays/src/Data/Array/Mixed/Internal/Arith, branch repro-9.14-branchNested, compositional struct-of-arrays orthotope arrays
https://git.tomsmeding.com/ox-arrays/atom?h=repro-9.14-branch2025-03-20T12:01:24ZSeparate arith routines into a library2025-03-20T12:01:24ZTom Smedingtom@tomsmeding.com2025-03-20T12:01:24Zurn:sha1:55036a5ea4a6e590d0404638b2823c6a4aec3fba
The point is that this separate library does not depend on orthotope.
arith: Don't FFI-import unused dotprod_*_strided ops2025-03-18T20:32:49ZTom Smedingtom@tomsmeding.com2025-03-18T20:32:49Zurn:sha1:27c2823387b21e8ed801e4d8eeb0b3e5588a2920Optimise reductions and dotprod with more vectorisation2025-03-14T20:58:51ZTom Smedingtom@tomsmeding.com2025-03-14T20:57:56Zurn:sha1:6276ed3c7bcd20c8b860e1275386ecd068671bcc
Turns out that if you don't supply -ffast-math, the C compiler will
faithfully reproduce your linear reduction order, which is rather
disastrous for parallelisation with vector units.
This changes the summation order, so numerical results might differ
slightly. To wit: the test suite needed adjustment.
Implement quot/rem2025-03-13T08:27:51ZTom Smedingtom@tomsmeding.com2025-03-12T22:20:13Zurn:sha1:ed6acbe5f409aba2fb222693da567ce04b7c4e01Binary ops without normalisation2025-03-12T21:25:35ZTom Smedingtom@tomsmeding.com2025-03-05T23:08:40Zurn:sha1:766a925698a97cac03e972bdaa2500085be17c65
Before:
> sum(*) Double [1e6] stride 1; -1: OK
> 68.9 ms ± 4.7 ms
After:
> sum(*) Double [1e6] stride 1; -1: OK
> 1.44 ms ± 50 μs
arith: Unary float ops on strided arrays without normalisation2025-03-05T21:09:50ZTom Smedingtom@tomsmeding.com2025-03-05T21:09:50Zurn:sha1:984e5315768dd190a97069167daf970c17c3c867arith: Only strided unary int ops2025-02-16T22:50:07ZTom Smedingtom@tomsmeding.com2025-02-16T22:49:56Zurn:sha1:71908c23307952fac26a4e24066e064d9cbb71c0
This should have negligible overhead and will save a whole bunch of C
code duplication when the FUnops are also converted to strided form.
arith: Unary int ops on strided arrays without normalisation2025-02-15T23:30:25ZTom Smedingtom@tomsmeding.com2025-02-15T23:30:25Zurn:sha1:c14017f4bc28951be7e298d01769b5b49384a7c3Add {m,r,s}dot1Inner2024-06-19T13:57:43ZTom Smedingt.j.smeding@uu.nl2024-06-19T13:57:43Zurn:sha1:aafe5f6b5fa772d0e2e9f9b4f91bc3e7cf696840Clean up Foreign.hs2024-06-18T19:55:35ZTom Smedingtom@tomsmeding.com2024-06-18T19:55:35Zurn:sha1:97ab8502b9cd3f7d908160d13c7d85d23c99e203