OpenBLAS VS Lunar

The Fortran implementation is just a reference implementation. The goal of reference BLAS [0] is to provide relatively simple and easy to understand implementations which demonstrate the interface and are intended to give correct results to test against. Perhaps an exceptional Fortran compiler which doesn't yet exist could generate code which rivals hand (or automatically) tuned optimized BLAS libraries like OpenBLAS [1], MKL [2], ATLAS [3], and those based on BLIS [4], but in practice this is not observed.

Justine observed that the threading model for LLaMA makes it impractical to integrate one of these optimized BLAS libraries, so she wrote her own hand-tuned implementations following the same principles they use.

[0] https://en.wikipedia.org/wiki/Basic_Linear_Algebra_Subprogra...

[1] https://github.com/OpenMathLib/OpenBLAS

[2] https://www.intel.com/content/www/us/en/developer/tools/onea...

[3] https://en.wikipedia.org/wiki/Automatically_Tuned_Linear_Alg...

[4]https://en.wikipedia.org/wiki/BLIS_(software)

Yeah. I'm going to be helping to work on expanding CI for OpenBlas and have been diving into this stuff lately. See the discussion in this closed OpenBlas issue gh-1968 [0] for instance. OpenBlas's Skylake kernels do rely on intrinsics [1] for compilers that support them, but there's a wide range of architectures to support, and when hand-tuned assembly kernels work better, that's what are used. For example, [2].

[0] https://github.com/xianyi/OpenBLAS/issues/1968

[1] https://github.com/xianyi/OpenBLAS/blob/develop/kernel/x86_6...

[2] https://github.com/xianyi/OpenBLAS/blob/23693f09a26ffd8b60eb...

We'll have to wait until part 2 to see what they are actually proposing, but they are trying to solve a real problem. To get a sense of things check out the handwritten assembly kernels in OpenBlas [0]. Note the level of granularity. There are micro-optimized implementations for specific chipsets.

If progress in ML will be aided by a proliferation of hyper-specialized hardware, then there really is a scalability issue around developing optimized matmul routines for each specialized chip. To be able to develop a custom ASIC for a particular application and then easily generate the necessary matrix libraries without having to write hand-crafted assembly for each specific case seems like it could be very powerful.

[0] https://github.com/xianyi/OpenBLAS/tree/develop/kernel

libraries mkl_rt not found in ['C:\python\lib', 'C:\', 'C:\python\libs'] ``` Install this and try again. Might need to reboot, never know with Windows https://www.openblas.net/

There isn't any fortran code in the repo there itself but numpy itself can be linked with several numeric libraries. If you look through the wheels for numpy available on pypi, all the latest ones are packaged with OpenBLAS which uses Fortran quite a bit: https://github.com/xianyi/OpenBLAS

Sure - write functions in another language, export C bindings, and then call those functions from Python. An example is NumPy - a lot of its linear algebra functions are implemented in C and Fortran.

Read the official docs yet?

It’s done in a similar way on macOS: a dylib is added to the bundle and an LC_LOAD command is added to the app binary. The dylib is the first thing that runs because of using the constructor attribute, like this: https://notes.alinpanaitiu.com/Injecting%20a%20DYLIB%20into%...

The nice thing is that a signed app will refuse to load a dylib that does not have the same signature. So crackers will be forced to change the whole app signature which can be easily detected in app code.

I have that kind of protection in Lunar (https://lunar.fyi/) and Clop (https://lowtechguys.com/clop) and it seems to be good enough as they have no recent cracks.

Pretty sure Lunar [0] can do this for you, and you can buy a lifetime license.

[0]: https://lunar.fyi/

I've had good luck with the Lunar app - it manages my Dell and LG monitors on an M2. (No affiliation) https://lunar.fyi

Wild! I am working on exactly the same thing now for Lunar (https://lunar.fyi), and I'm also calling it Night Mode ^_^ what a coincidence

I've been trying to make "white regions in dark backgrounds" less painful for months, but doing that at the system level on macOS is incredibly hard. I see you're doing it with CSS filters, which make sense in the limited scope of an article. But applying something like that on the whole macOS UI would cause confusion.

I already use something similar on the iPhone: I read on the Kindle app which has white text on black background, then I have a full red Color Tint filter on the Triple Back Tap shortcut which I use before reading. Very similar effect to your solution, although I don't have images in my books.

I was comparing anti-piracy measures with DRM, I don't have actual DRM in my app. I can't block users that really bought the app from using it (which is what DRM is notorious for).

But I do have a license verification for the Pro features (https://lunar.fyi/#pro), and that is what people are cracking in the app. I only added more protection around this verification.

Lunar

There's no Reduce White Point on Mac as far as I am aware. However, you can use the fantastic Lunar [0] app to achieve this, as it supports "Sub-Zero Dimming".

To use it, I think you just need to start Lunar, and then press the Reduce Brightness button on your keyboard until it goes below the minimum Mac allows.

[0] https://lunar.fyi

As the dev of a macOS app that breaks all the time because of external hardware, the tone of the article hits close to home. (I’m talking about https://lunar.fyi/ whose brightness control commands can be blocked by USB-C hubs, “smart” monitors, too long cables etc.)

I had to disable public GitHub issues on the app repo [1] because people seemed to fuel each other with spiteful comments and “why can’t you just!!” sentences.

The contact form still attracts many such “entitled” people and it hurts to wake up to such messages, but at least I can choose to ignore those if I can’t bring anything to the discussion. There’s no peer pressure.

These people are expecting too much from a handful of developers who are sharing a lot of free work and time that could have been spent better than hunting new IDs in URLs and updating regular expressions.

[1] https://github.com/alin23/Lunar

Last year I bought a second computer for my music studio. I wanted to use the same set of 2 monitors and wired keyboard + trackpad on both machines.

I wrote simple scripts to switch my monitor inputs with keyboard shortcuts (even simpler with Lunar, amazing new Mac app — https://lunar.fyi), which saved me from having to press annoying input-source buttons.

But I couldn't for the life of me find a simple, suitable software controllable KVM switch. That still requires the hardware button to be controlled, so frustrating.

Some people like the process of writing code, more than the end result. I had a few months of that feeling, but nowadays it’s rarely about writing for me.

Just the other day I used Copilot to explain the disassembly of macOS KeyboardBacklight code, so that I can turn off the keyboard lights when using Lunar’s Blackout (https://lunar.fyi/#blackout)

It even helped me generate the ObjC function signatures from assembly and use the right calling convention in Swift afterwards. It really feels like magic.

I would have no joy in writing that code, it’s mostly bridging and translation anyway. I just need it to do this thing so that people can take advantage of it.