Another fascinating article by Matt #Godbolt in his Advent of Compiler Optimization series. This one is about induction variables and loops:
https://xania.org/202512/09-induction-variables
Here, Matt uses the llvm-mca tool to visualize the x86 Haswell CPU pipeline to show loop-carried dependencies.
If you want to see this for RISC-V (SiFive U74) for one of the given examples, try this link (llvm-mca -march=riscv64 -mcpu=sifive-u74 -timeline):
I can't use godbolt.org to confirm if the bug is still present in newer versions of GCC because I can't figure out how to see the `.ci` output file on godbolt.org. (Also, there's no "none-eabi" on godbolt.org, just "unknown-eabi" and "linux-gnueabihf"; the code won't compile with the GNU eabi, but I suspect that unknown-eabi will show the same behavior as none-eabi.)
You don't realize just how good C++23 is until you take some old C++ code, paste it into #Godbolt, set the compiler flags to C++23, add a few concept-checking asserts, read the beautifully clear error messages from concepts, and fix your problem.
#ModernCPP #CPP #Concepts #Legacy #Evolution #Happyiness #Programming
[Перевод] Ускоряем работу видеодекодера на 1%
Какое-то время назад memorysafety.org объявил о конкурсе по повышению производительности rav1d — порта AV1-декодера dav1d на Rust. Моя фамилия Равид, совсем как название декодера, поэтому я решил, что будет забавно попробовать (хоть я и, вероятно , не смогу участвовать в конкурсе). Эта статья посвящена двум найденным мной небольшим улучшениям производительности ( первый PR , второй PR ) и рассказу о том, как я их нашёл.
https://habr.com/ru/articles/912136/
#оптимизация_кода #обработка_видео #декодер #dav1d #rav1d #godbolt
Very interesting (and informing) video about x86_64 versus ARM architectures:
Recorded live on twitch, GET IN ### GuestCasey Muratori | https://x.com/cmuratori?ref_src=twsrc%5Egoogle%7Ctwcamp%5Eserp%7Ctwgr%5Eauthor### My Streamhttps://...
Need a clue pls. What's the trick to get #godbolt to target RISC-V for Rust? Just adding the `--target riscv32i-unknown-none-elf` compiler flag didn't work:
https://godbolt.org/z/1df9GqEPW
Clearly there's a way as https://rust.godbolt.org/noscript/z/Y8s3Kf (with the appropriated corrections) works.
#UICouldBeBetter
// Type your code here, or load an example. // As of Rust 1.75, small functions are automatically // marked as `#[inline]` so they will not show up in // the output when compiling with optimisations. Use // `#[no_mangle]` or `#[inline(never)]` to work around // this issue. // See https://github.com/compiler-explorer/compiler-explorer/issues/5939 #[no_mangle] pub fn square(num: i32) -> i32 { num * num } // If you use `main()`, declare it as `pub` to see it in the output: // pub fn main() { ... }
Does anybody know how to prevent #clang from issuing calls to libc functions like `memset` in this case when compiling C programs? I have tried I don't know how many command line arguments to try to disable it but none worked at all.
Here you have an example program in #godbolt: https://godbolt.org/z/jheYoPWzj
These are the command line arguments I've tried to disable it:
-ffreestanding -disable-simplify-libcalls -fno-builtin -nostdinc -nostdlib -fno-builtin-memset -nostdlib++ -nostdinc++
Any idea?
typedef unsigned long long size_t; extern void do_stuff(unsigned char *, size_t); int main() { { const size_t stuff_len = 65; unsigned char stuff[stuff_len] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF }; do_stuff(stuff, stuff_len); } }
Michael Engel
Luke T. Shumaker
mittorn
Mathias Hasselmann
N-gated Hacker News
Habr
gprimola$ 
Tommy Þ
Joxean Koret (@matalaz)