-
Notifications
You must be signed in to change notification settings - Fork 79
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
Towards 2.3 #490
Comments
Hi Marco,
I'm not sure where you got "3.0" from. That is a big management decision.
We had discussed calling the next release 2.3, since no interface/API are
going to change (please).
Thanks for gathering the list of things currently on the plate - the list
seems to include things already in master (#459, #471, #440, etc)
We have to make some decisions about what other nearly-ready things go into
2.3.
I propose *not* to include the larger changes that are only just begun, or
not even begun at all (de-macroizing, templating, type 3 on GPU, dispatch).
Could we make 2.3 the goal for now, and finish up for it *only*:
#429
#461 #477
#479
Everything else should wait! (even #488 needs work on 3D,etc that means it
should be decoupled from the rest).
Decoupling is crucial to maintain sanity here.
Thanks! Alex
…On Wed, Jul 17, 2024 at 11:41 AM Marco Barbone ***@***.***> wrote:
I am creating this milestone to discuss what goes in 3.0 and what is left
to do. This can be extended once the switching to template tasks are
formalized.
If we aim to 3.0 we could also consider changing the public API if there
are good reasons for it. The main changes are:
- Spreader performance improvements with manual vectorization #459
<#459>
- Interp perfomance improvements with manual vectorization and compile
flags #471 <#471>
- Kernel vectorization #459
<#459>
- Migrating to modern pyproject.toml from deprecated setup.py #429
<#429>
- Kernel use Estrin and exploit coefficient symmetry to reduce the
number of operations #461
<#461> #477
<#477>
- Optimizing foldrescale, removes the input range limitations #440
<#440>
- Switchable FFT, support for DUCC FFT #463
<#463>
- DUCC FFT documentation #479
<#479>
- De-Macroize #482
<#482> #483
<#483>
- constexpr everything #487
<#487>
- Type 3 on GPU #489
<#489>
- GPU performance improvements #488
<#488>
- Architecture dispatch #486
<#486>
—
Reply to this email directly, view it on GitHub
<#490>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACNZRSQXNPW5PTBKFESA4V3ZM2GALAVCNFSM6AAAAABLA4ZVD2VHI2DSMVQWIX3LMV43ASLTON2WKOZSGQYTGOJUHE2DIOI>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
--
*-------------------------------------------------------------------~^`^~._.~'
|\ Alex Barnett Center for Computational Mathematics, Flatiron Institute
| \ http://users.flatironinstitute.org/~ahb 646-876-5942
|
I also have to add for 2.3 the kernel coeff regeneration I'm currently doing for upsampfac=2.0. I think upsampfac=1.25 kernel xsimd eval should be left for 2.3.1. |
Sounds good. The main point of this issue/milestone is to discuss and prioritize the various tasks. |
I applaud the organization effort - just to scale back the number of things a lot here. Ok. If you can redo your milestone tagging with 2.3 according to my above scheme, that would make everything feel a little more in control - thanks !:) |
updated acknowledgements, and CHANGELOG for all that's merged right now. |
I am creating this milestone to discuss what goes in 2.3 and what is left to do. This can be extended once the switching to template tasks are formalized.
The main changes are:
The text was updated successfully, but these errors were encountered: