zmowera day ago
Already done here https://github.com/simonjwright/building-gcc-macos-native Simon's the man when it comes to GNAT on macOS.
kevlar700a day ago
True though from what he has said early on in it being more of a hobby and liking building tools then I am sure he would welcome any help or input.
zoom6628a day ago
Well that's amazing. And now I can (accidentally of course) install on my M2 work machine.
Pinusa day ago
Only vaguely related, but: Why is everything called a.*64 these days? I keep getting confused between amd64, arm64 and aarch64...
lloekia day ago
To nitpick there is not quite such a thing as "ARM64", instead there's:
- Aarch64, the execution state
- A64, the instruction set
- ARMvX, the vX of the architecture, both of the above were introduced in ARMv8
ARM64 is kind of a loose umbrella term that might or might not have been officially retconned.
Why is that? Because ARM specifications are modular, so for each vX there are mandatory and optional set of features.
unscaled21 hours ago
There's also no official x64, x86-64 or i686 "architecture", if we want to be pedantic.
There is no official name for the 64-bit instruction set for the x86 architecture, as far as I know. As far as I remember, AMD called this ISA "AMD64" when it was released and Intel just called it "Intel 64". Of course, this term only refers to the basic instruction set. If you see an x86-64 binary, it may be compiled with an extended instruction set like SEE4 or AVX that is not necessarily supported by every x86-64 CPU out there.
Even the umbrella term "x86" for the 32-bit ISA was retconned as far as I know: Intel did not use that term originally.
MarcusE1Wa day ago
This is often used in relation to the processor architecture. The 64 is added to indicate it's a 64 bit CPU architecture.
This is to distinguish it to the for a long time common 32 bit CPU architectures.
Looking at the current market you could think that there are mostly 64 bit CPUs sold nowadays, but many people still use (older) 32 bit CPUs.
tjra day ago
Bring on the quadruple-precision floats!
jiehong21 hours ago
Alternatively, I see an LLVM front-end for Ada [0], but I'm not a user, so the maturity of such a solution is unclear to me.