forums.ps2dev.org

there are thousands of 8byte integer bit patterns that when loaded as fld get munged It's about the 80 bit full-precision fpu content which are left fully intact when loaded/stored from/to memory, so they're fine with doing 10 byte moves using the fpu. I don't remember the details, but it looked li...

The stripes in carmageddon are due to the way they write to the screen. They're using the fpu to do large block transfers/masked transfers, so an 80bit precise fpu emulation is needed (currently only x86 can do this by using the host fpu). But as already noted the psp might not play this game at acc...

Hehe. Congrats for the excessiveness of your questionmarks.

Comparing gouraud-shading of triangles with a full system emulator is pretty dumb. And telling people that they are not good programmers disqualifies you for the second time. Feel free to take some old PC (x86 to simplify the task) like a 200mhz one, and try to get some system emulation like bochs o...

if someone doesn't want to do sth

He already put up everything needed including instructions,
but for some reason you don't want to use it.

I quoted the GPL

Uhm yes and if you look one line above that part you quoted it says that this is one of the possibilities, not the only one.

What matters is what the GPL says

Right, and what you're referring to is the GPL faq which is how the FSF thinks that the GPL should be taken. Not the GPL itself.

If one distributes a binary, the code must have been compilable. Which part of the GPL says that? Please, go read the GPL FAQ that J.F. quoted Please, go read the GPL FAQ and have a look at WHY they explicitly mention that a diff is not enough in the general case: because this does not allow the us...

So if the project is under the GPL, you cannot supply just diffs or what files are different. You have to supply everything. We assure that the sources will be available from the official project page, so a diff against them is fine as it fulfils the GPL's requirement that a user has access to the ...

This is an unfortunate problem with the recompiler, it cannot know beforehand if a recompiled section will overrun it's bounds. All I can do is reduce the number of x86 ops recompiled per block. Try increasing the spare buffer (CACHE_MAXSIZE) which should also enable you to have a higher maximum fo...