Li, so they're not always

that small, considering that you run them lak once per picl yes, so that's certainly true. They can be very complex. And then the other point is, it sounds expensive to be running them everywhere. And the restrictions that raman has talked about, where you only know, like, where am i on the screen, and you have to use that to figure out color that you're going to output, that's really because the model that gpus use of restricting that information lets them run huge numbers of these invocations in parallel. So that's really where the gpu shines, is that the model of, you know, you get very limited information. You can't look at your neighbors, you can't do these big loops that lets the gpu process huge numbers of iles at the same time.

Inn addition to that, and that's how we'll tie back to your original question, chat about the language. The language used to write those programmes, is we more limited than what you're used to on the sep that you don't have strings, you con't create your own threads? Tees, not tr te ao, so how dyou write one of thes shaders? The next

release of android, what we've got coming is something called a runtime shader, which is exactly of what we've described here. It's part of the andrographic family of shaders, like gradients and bit maps. But this allows you to interject this new andrographic shading language code into the entire graphic process. And the basic io and the basic function that you'll be providing is really just a simple function that takes the coordinate the space that you're in on the screen, and you output of color. And then we can that information into the gpu, do all these operations in parallel, so that you can achieve the same kind of gpu level effects while still operating in the andrographic environment. Yes.