kernel_tuner VS soundfingerprinting

I've created Kernel Tuner (https://github.com/KernelTuner/kernel_tuner) as a small software development tool, because I was writing a lot of CUDA and OpenCL kernels at the time. I didn't want to manually figure out what best thread block dimensions and work division among threads were on every GPU over and over again.

The tool evolved quite a bit since the first versions. I'm also using it for testing GPU code, teaching, and it has become one of the main drivers behind a lot of the research that I do.

We have an open science policy, so anyone can use our framework yourself to optimize stuff, if you want! The original paper is linked at the bottom of the GitHub page.

This is a great post for people who are new to optimizing GPU code.

It is interesting to see that the author got this far without interchanging the innermost loop over k to the outermost loop, as is done in CUTLASS (https://github.com/NVIDIA/cutlass).

As you can see in this blog post the code ends up with a lot of compile-time constants (e.g. BLOCKSIZE, BM, BN, BK, TM, TN) one way to optimize this code further is to use an auto-tuner to find the optimal value for all of these parameters for your GPU and problem size, for example Kernel Tuner (https://github.com/KernelTuner/kernel_tuner)

Started 10 years ago as an open-source project, building an algorithm for audio fingerprinting. Added a commercial offering, selling storage built specifically for audio fingerprints, targeting enterprise customers. Since the offering was too technical (it's hard to sell solutions to problems that are too narrow and domain-specific), pivoted to more "business-oriented problems". This last year's pivot is a chance to finally grow. Running a business in single-player mode is, at times, too stressful. Aside from the technical part, which I very much enjoy, I need to wear marketing, sales, and customer support hats.

[1] - https://emysound.com

The OP said elsewhere they are using this[1] library, which allows you to specify minimum seconds to match, so you'd presumably set it to match 20 seconds or whatever minimum length podcast commercials usually are.

Most other audio fingerprinting libraries I've seen allow you to specify min/max time, as well.

HTH.

1. https://github.com/AddictedCS/soundfingerprinting

My goal for the next year is just to work fewer hours. Covid pushed my work habit to the extreme, and I need to rebalance.

1. Read more books, less social media/news.

2. Spend more time with my friends. I haven't seen some of them IRL for more than a year.

3. Exercise more, play tennis with my daughter, spend quality time with my kids.

4. Spend more time with my parents. They've become visibly older in the last years, an observation that scares me.

5. Focus more on marketing for the businesses that I've bootstrapped. All the shiny new features that are developed are not as important as getting more people to use your product.

6. Promote open-source project to 1k GitHub stars[1]. I know it isn't very meaningful, but it's just nice to receive a bit of recognition from the community.

7. Enjoy life, don't stress about all the little things that happen along the way.

Happy new year!

[1]: https://github.com/AddictedCS/soundfingerprinting