10 ms·

> can anyone reverse-engineer the mali-400 and write a shader compiler for it in their spare time? I don't believe this is the hardest part (though surely not

by wolfgke 10y ago

> can anyone reverse-engineer the mali-400 and write a shader compiler for it in their spare time? I don't believe this is the hardest part (though surely not easy). In my opinion the much harder part is to do it (and be able to show that you did it) in a way that is 100% legal (i.e. strong conditions on clean room reverse engineering etc.). So for example a few years ago some PowerVR driver source code was leaked (cf. for example http://libv.livejournal.com/26972.html http://libv.livejournal.com/26972.html). Of course you are not allowed to use it to build some "open" Linux driver for PowerVR GPUs. We could easily have some PowerVR GPU driver with source code available. The only reason why this does not exist yet is that people care about legality.

jancsika 10y ago
Well, when 1000+ hackers have domain expertise you typically end up with at least a handful of projects containing working code. That's even if the code itself lives in a legally grey area. (Remember the various Popcorn Time forks?) So I don't think legality is the fundamental issue. Rather, I think it's that there are probably only 2 people with the domain expertise necessary to lead a successful reverse-engineering project of this particular chip. As a general response to the grandparent: I think the reason we don't have an affordable, fully-free modern laptop is because it requires many components for which the free software world only has a handful of domain experts. In other words, such a laptop is in the category of problems for which hackers tend to vastly overestimate their potential resources. (I think secure messaging applications also fall into this same category.)