[Numpy-discussion] How will Google DeepMind's AlphaTensor effect numpy

2022-10-26 Thread canhtart 
Hello!

I was curious on how AlphaTensor will effect NumPy and other similar 
applications considering it has found a way to perform 3x3 matrix 
multiplication efficiently. 
https://www.deepmind.com/blog/discovering-novel-algorithms-with-alphatensor. I 
am not even sure how NumPy does this under the hood, is it 2x2?

Is anyone working on implementing this 3x3 algorithm for NumPy? Is it too 
early, and if so why? Are there any concerns about this algorithm?


I am new to the mailing list, it is nice to be here! Long-time user of NumPy in 
my projects, thank you for your amazing work!
Can
___
NumPy-Discussion mailing list -- numpy-discussion@python.org
To unsubscribe send an email to numpy-discussion-le...@python.org
https://mail.python.org/mailman3/lists/numpy-discussion.python.org/
Member address: arch...@mail-archive.com

[Numpy-discussion] Re: How will Google DeepMind's AlphaTensor effect numpy

2022-10-26 Thread Robert Kern 
On Wed, Oct 26, 2022 at 9:31 AM  wrote:

> Hello!
>
> I was curious on how AlphaTensor will effect NumPy and other similar
> applications considering it has found a way to perform 3x3 matrix
> multiplication efficiently.
> https://www.deepmind.com/blog/discovering-novel-algorithms-with-alphatensor.
> I am not even sure how NumPy does this under the hood, is it 2x2?
>
> Is anyone working on implementing this 3x3 algorithm for NumPy? Is it too
> early, and if so why? Are there any concerns about this algorithm?
>

numpy links against accelerated linear algebra libraries like OpenBLAS and
Intel MKL to provide the matrix multiplication. If they find that the
AlphaTensor results are better than the options they currently have, then
numpy will get them. In general, I think it is unlikely that they will be
used. Even the older state of the art that they compare with, like
Strassen's algorithm, are not often used in practice. Concerns like memory
movement and the ability to use instruction-level parallelism on each kind
of CPU tend to dominate over a marginal change in the number of
multiplication operations. The answers to this StackOverflow question give
some more information:


https://stackoverflow.com/questions/1303182/how-does-blas-get-such-extreme-performance

-- 
Robert Kern
___
NumPy-Discussion mailing list -- numpy-discussion@python.org
To unsubscribe send an email to numpy-discussion-le...@python.org
https://mail.python.org/mailman3/lists/numpy-discussion.python.org/
Member address: arch...@mail-archive.com

[Numpy-discussion] Re: How will Google DeepMind's AlphaTensor effect numpy

2022-10-26 Thread Ilhan Polat 
Moreover, not to mention the result is mostly valid for mod2 aritmetic;
something the authors chose to mention it in the fine print causing this, a
bit overblown in my opinion, excitement.

So for now it seems like we don't need to take action for regular matmul.


On Wed, Oct 26, 2022, 16:03 Robert Kern  wrote:

> On Wed, Oct 26, 2022 at 9:31 AM  wrote:
>
>> Hello!
>>
>> I was curious on how AlphaTensor will effect NumPy and other similar
>> applications considering it has found a way to perform 3x3 matrix
>> multiplication efficiently.
>> https://www.deepmind.com/blog/discovering-novel-algorithms-with-alphatensor.
>> I am not even sure how NumPy does this under the hood, is it 2x2?
>>
>> Is anyone working on implementing this 3x3 algorithm for NumPy? Is it too
>> early, and if so why? Are there any concerns about this algorithm?
>>
>
> numpy links against accelerated linear algebra libraries like OpenBLAS and
> Intel MKL to provide the matrix multiplication. If they find that the
> AlphaTensor results are better than the options they currently have, then
> numpy will get them. In general, I think it is unlikely that they will be
> used. Even the older state of the art that they compare with, like
> Strassen's algorithm, are not often used in practice. Concerns like memory
> movement and the ability to use instruction-level parallelism on each kind
> of CPU tend to dominate over a marginal change in the number of
> multiplication operations. The answers to this StackOverflow question give
> some more information:
>
>
> https://stackoverflow.com/questions/1303182/how-does-blas-get-such-extreme-performance
>
> --
> Robert Kern
> ___
> NumPy-Discussion mailing list -- numpy-discussion@python.org
> To unsubscribe send an email to numpy-discussion-le...@python.org
> https://mail.python.org/mailman3/lists/numpy-discussion.python.org/
> Member address: ilhanpo...@gmail.com
>
___
NumPy-Discussion mailing list -- numpy-discussion@python.org
To unsubscribe send an email to numpy-discussion-le...@python.org
https://mail.python.org/mailman3/lists/numpy-discussion.python.org/
Member address: arch...@mail-archive.com

[Numpy-discussion] Re: Documentation Team meeting - Monday October 24 at 2pm UTC

2022-10-26 Thread Melissa Mendonça 
I am in favor of alternating times for the Docs team meeting - making it a
little bit more accessible for folks in other timezones :)

Please keep an eye out for announcements, and if you have strong opinions
on this let us know. I think we could start with alternating between 4pm
UTC and noon UTC, unless there are better suggestions.

Thanks for bringing this up, we'd love to make it better for other folks to
participate as well.

Cheers,

Melissa

On Wed, Oct 26, 2022 at 2:44 AM 腾刘 <27rabbi...@gmail.com> wrote:

> Not suitable for Asian area but I could attend the meeting next semester,
> hopefully I will be in Europe for further education 😋.
>
> Melissa Mendonça  于2022年10月25日周二 23:19写道：
>
>> So sorry about this - I forgot to update this link! The hackmd link is
>> the correct one. Will make sure to keep you all updated for next time. But
>> the plan is to move it to 4pm UTC starting next time, November 7.
>>
>> Best,
>>
>> Melissa
>>
>> On Mon, Oct 24, 2022 at 12:35 PM 腾刘 <27rabbi...@gmail.com> wrote:
>>
>>> Seems that the link is broken. Because it is different from the one in
>>> the markdown ~
>>>
>>> --
>>>
>>> If you wish to join on Zoom, use this link:
>>>
>>> https://zoom.us/j/96219574921?pwd=VTRNeGwwOUlrYVNYSENpVVBRRjlkZz09
>>>
>>> ---
>>>
>>>
>>>- Time: 2 pm UTC
>>>- Join via Zoom:
>>>
>>> https://us06web.zoom.us/j/85016474448?pwd=TWEvaWJ1SklyVEpwNXUrcHV1YmFJQT09 
>>> (To
>>>dial in, find your local number:
>>>
>>>
>>> Inessa Pawson  于2022年10月24日周一 23:23写道：
>>>
 Hi, Teng!
 We didn’t have a lot of topics on the meeting agenda and finished 20
 min earlier.
 If you have any questions or suggestions related to contributing or
 improving the official NumPy documentation, feel free to ask them here, on
 the mailing list, or in the #docs channel on the NumPy workspace.

 On Mon, Oct 24, 2022 at 10:22 AM 腾刘 <27rabbi...@gmail.com> wrote:

> Hiii~
>
> Have the meeting already re-scheduled to 4pm UTC? I found nobody there.
>
> Melissa Mendonça  于2022年10月24日周一 08:05写道：
>
>>
>> Hi all!
>>
>> Our next Documentation Team meeting will happen on *Monday, October
>> 24* at ***2PM UTC***. (We are also considering moving this meeting
>> to its original time in the future - 4PM UTC. If you have opinions on 
>> this
>> time change, let us know!)
>>
>> All are welcome - you don't need to already be a contributor to join.
>> If you have questions or are curious about what we're doing, we'll be 
>> happy
>> to meet you!
>>
>> If you wish to join on Zoom, use this link:
>>
>> https://zoom.us/j/96219574921?pwd=VTRNeGwwOUlrYVNYSENpVVBRRjlkZz09
>>
>> Here's the permanent hackmd document with the meeting notes (still
>> being updated):
>>
>> https://hackmd.io/oB_boakvRqKR-_2jRV-Qjg
>> 
>>
>> Hope to see you around!
>>
>> * You can also visit https://scientific-python.org/calendars to add
>> the NumPy community calendar as an .ics file to your preferred calendar
>> manager
>>
>> - Melissa
>> ___
>> NumPy-Discussion mailing list -- numpy-discussion@python.org
>> To unsubscribe send an email to numpy-discussion-le...@python.org
>> https://mail.python.org/mailman3/lists/numpy-discussion.python.org/
>> Member address: 27rabbi...@gmail.com
>>
> ___
> NumPy-Discussion mailing list -- numpy-discussion@python.org
> To unsubscribe send an email to numpy-discussion-le...@python.org
> https://mail.python.org/mailman3/lists/numpy-discussion.python.org/
> Member address: albusc...@gmail.com
>


 --
 Cheers,
 Inessa

 Inessa Pawson
 Contributor Experience Lead | NumPy
 https://numpy.org/
 GitHub: inessapawson
 ___
 NumPy-Discussion mailing list -- numpy-discussion@python.org
 To unsubscribe send an email to numpy-discussion-le...@python.org
 https://mail.python.org/mailman3/lists/numpy-discussion.python.org/
 Member address: 27rabbi...@gmail.com

>>> ___
>>> NumPy-Discussion mailing list -- numpy-discussion@python.org
>>> To unsubscribe send an email to numpy-discussion-le...@python.org
>>> https://mail.python.org/mailman3/lists/numpy-discussion.python.org/
>>> Member address: meliss...@gmail.com
>>>
>> ___
>> NumPy-Discussion mailing list -- numpy-discussion@python.org
>> To unsubscribe send an email to numpy-discussion-le...@python.org
>> https://mail.python.org/mailman3/lists/numpy-discussion.python.org/
>> Member address: 27rabbi...@gmail.com
>>
> ___