manx
manx
> For architectures where unaligned access is not allowed you will see a SIGBUS abort of the process well before you see any data corruption. And wrong again. The first...
I have explained my arguments very thoroughly and in great detail. I honestly cannot think of what you might want to hear as any additional proof. The C standard is...
I am sorry that you feel attacked personally by my arguments. I certainly did not intend that. However, if you repeatedly claim facts that are simply not true, I am...
> I unfortunately can not take responsibility for what others say during a debate, only what I do with the combined information in the end. Agreed, fair enough. > So,...
> If our code would be as shit as what was claimed in this issue, it wouldn't have worked on 32-bit PowerPC as it has quite high restrictions on unaligned...
> We compile on all architectures that can be emulated on qemu or wine inside Ubuntu Linux (using GitHub Actions). Unfortunately, QEMU, at least for ARM, up until very recently...
I found a way to obtain a dsdmo.dll (64bit), however I have not yet found a way to extract it without using Windows itself. According to https://www.thecollectionbook.info/builds/windows/version/93 , Windows 8.1...
``` manx@quadratus:~/projects/openmpt/trunk-1$ time clang-format-14 -i src/mpt/arch/x86_amd64.hpp real 0m2.576s user 0m2.555s sys 0m0.020s ``` @codido So yes, changing `QualifierAlignment` to `Leave` makes it run in seconds and avoids the performance problem....
Are your performance numbers ASM vs `-march=i686` (or anything similarly generic) or ASM vs `-march=pentium4/bonnell/athlon-xp` for the particular CPU? Given that such old CPUs are nowadays kind of a special...
> Actually, this has nothing to do with march tunings. FLAC is by default configured to add -msse2 to the command line (and similar to Visual Studio project files). I...