Does the AMD compiler work on Gentoo?

Name ONE (2) thing wrong with AMD

>already compiles stock LLVM in the course of building mesa3d openswr
eh shouldn't be insanely different from a normal appleclang build from source

i assume they want people to use amd cpus in their servers so they provide their own compiler to squeeze out more out of those cpus.

Interesting. But what's the point of this? Don't GCC and clang already do as well of a job as current knowledge allows, given that you target a recent CPU architecture?

Not sure, my guess is AOCC takes advantage of proprietary AMD technology.
I wanted to know if compiling with it would break packages that, for some reason, are customized for GCC.

Now they should disable all optimizations if the compiler detects non amd or via cpu.

>patches amdcc to emit a completely neutral CPU dispatch function
heh

Why did they make their own patched version of clang instead of contributing to clang itself (and maybe gcc too)?

AMD stronk. Who needs Intel CPUs and Nvidia GPUs, when you have AMD APUs?

Attached: AMD-APU-diagram-1-640x300.png (640x300, 169K)

>cpus
>gpus
Where's my research money

Why would they

You said Apu?

Attached: apu.png (500x500, 39K)

Contributing to the compilers everyone uses would make sure all users who target their specific platform get the speed benefits, and not just the ones who have heard about AOCC and downloaded it. It would also be great for publicity, evil Intel with their overpriced, exploit-ridden CPUs that come with the botnet engine vs. good guy AMD who contributes to free software and makes affordable CPUs.
Obviously, there could be technical reasons behind their choice, like maybe they had to make changes to non-architecture-specific parts of the compiler that actually decrease performance on non-AMD systems and thus wouldn't have been accepted. Or corporate ones, like maybe they simply didn't want to spend money on contributing to a code base that they don't have total control over.

should be possible to build a cross-compiler at a minimum, though i'd think building a native compiler should also be possible (using the same LLVMBuild.py + CMake build system from upstream)

native compiler can then be welded to either mingw64 or Microsoft C/C++

Intel C compiler produces slightly faster executables than GCC and clang on average.
I assume it fucks up on AMD hardware, which is the point of this.