Re: Improving performance of 2D convolution

From: prathamesh tagore (prathameshtagore_at_[hidden])
Date: 2021-06-05 17:43:11

Following is the list of tasks I intend to complete until Wednesday based
on my meeting with Olzhas :
1. Add a benchmark for the current implementation of 2D convolution.
2. Olzhas will probably add Opencv benchmark by Monday. I will then perform
a comparison between our implementation and Opencv's implementation and
post my results on the mailing list.
3. As suggested by Olzhas, I will add more test cases to cover additional
scenarios in #612 such as offsetting in a direction opposite to what I have
considered currently.
@lpranam, please share your opinion regarding this and intended
modifications (if any) in schedule. I will start working as soon as I get
your approval.
Best regards,
Prathamesh Tagore

On Thu, Jun 3, 2021 at 12:42 PM prathamesh tagore <
prathameshtagore_at_[hidden]> wrote:

> I am attaching a flowchart (
> https://github.com/meshtag/gil/blob/testing/Screenshot%20from%202021-06-03%2012-20-07.png)
> describing the first version related to improvement in 2D convolution. It
> is a rather simple version without simd modifications. Its purpose is to
> decrease cache reloads and misses by loading relevant image data in a 1D
> buffer container. Please share your opinion on whether it should be
> benchmarked and should I make any changes to it.
> Best regards,
> Prathamesh Tagore
>
> On Wed, Jun 2, 2021 at 9:49 PM Olzhas Zhumabek <
> anonymous.from.applecity_at_[hidden]> wrote:
>
>> This thread will probably pretty long and follow our findings and
>> direction
>> on how will improve the
>> performance of 2D convolution in GIL.
>>
>> During my meeting with Prathamesh, we discussed the nature of the 2D
>> convolution in OpenCV.
>> At the moment it uses some switches which allows it to change between SIMD
>> instructions,
>> high performance libraries and just a plain implementation with prefetch.
>> We decided to go the
>> simplest route for the first version, while continuously benchmarking past
>> implementations.
>>
>> Here is the initial plan:
>>
>>  - Prathamesh will send a flowchart for the basic implementation
>>  - I will sketch some benchmarking code with the cmake switches for
>> OpenCV,
>> so that we could pit
>> our versions against OpenCV (ETA until Monday next week)
>>  - Prathamesh will complete the benchmarking code with comparisons (ETA
>> sometime next week)
>>  - After discussing the flowchart, the implementation of the new version
>> will begin (next week too, probably)
>>
>> Please share your ideas and suggestions.
>>
>> Best,
>> Olzhas
>> --
>> Boost-gil mailing list
>> Boost-gil_at_[hidden]
>> https://lists.boost.org/mailman/listinfo.cgi/boost-gil
>>
>