GCC 14 vs. LLVM Clang 18 Compiler Performance On Fedora 40

[ArneBab]

Why is SMHasher sha3-256 twice in the detailed results?

Results (note smhasher sha3-256 and farmhash128 repeated): https://openbenchmarking.org/result/2404242-NE-GCC14AMDR04&sgm=1&swl=1&sgm=1&swl=1&ppt=D&sgm=1&sw l=1&ppt=D&sgm=1&swl=1&ppt=D&sgm=1&swl=1&ppt=D

There seems to be a very wide spread between two same named results — do these use different parameters or different data? https://openbenchmarking.org/result/2404242-NE-GCC14AMDR04&sgm=1&swl=1&sgm=1&swl=1&ppt=D&sgm=1&sw l=1&ppt=D&sgm=1&swl=1&ppt=D&sgm=1&swl=1&ppt=D&sgm= 1&swl=1&ppt=D&oss=smhasher

[topolinik]

Apart from open source games, nevertheless it would be relevant running a commercial heavy game on wine, when it's compiled using both gcc and clang.
Then you will see how much the compiler affects the wine layer performances.

[Anux]

ugv266 is something I mentioned myself

And? Is there some law that no one can say anything that you have mentioned?

and indeed nobody uses it

You for sure can provide a prove to that comical statement, what you can't? ... oh

[*]Every Phoronix user has shat on VVC

What a lazy lie and even if, a few phoronix users are not the whole world population. Try to use some logic ...

For you the only relevant codecs are VP8, VP9 and AV1.

What are you babbling? Where did I ever say such a thing? I may find AV1 to be a fine codec but I mainly use x264, it just has the best hardware support.

[*]vvenc is both a ton faster (several times) and produces higher quality output, there's no point in using an all-around inferior tool for the job.[*]It's dead, abandoned and unsupported.

Those are both no prove that no one uses it. It might be a good reason not to include it in tests. But of course one always has to take your statements with a grain of salt.

Senseless statements, zero argumentation from you.

damn, sometimes you are so funny

[kasper]

Easy enough to build on your own.

easiest on debian/ubuntu is their official script to install nightly binaries:

wget -qO-​ https://apt.llvm.org/llvm.sh | bash -s - 19 all     # or 'sudo bash' for non-root

and in ubuntu docker container:

 apt update && apt install -y wget lsb-release software-properties-common gnup

before running llvm.sh.

[avis]

Exactly the places that you avoid looking because the results are an inconvenience for your strange argumentation. Kvazaar ugv266

How do you know that no one uses it? Are you working for the NSA?

ugv266 is something I mentioned myself, can you even read?, and indeed nobody uses it, and indeed its performance doesn't apply to any workloads.

How do I know no one uses it?

• Every Phoronix user has shat on VVC and all other patented codecs non-stop. For you the only relevant codecs are VP8, VP9 and AV1.
• vvenc is both a ton faster (several times) and produces higher quality output, there's no point in using an all-around inferior tool for the job.
• It's dead, abandoned and unsupported.

Michael could have tested vvenc instead but it's heavy on hand-written assembly, so Clang and GCC will be very similar again. Like I said video codecs are not worth testing to compare compilers.

Senseless statements, zero argumentation from you. Goodbye.

[brad0]

Gentoo?

Easy enough to build on your own.

[imaami]

Gentoo?

Just Debian, tracking the unstable repo + specific parts of experimental (e.g. the clang/llvm stuff).

[Anux]

Um, where?

Exactly the places that you avoid looking because the results are an inconvenience for your strange argumentation. Kvazaar ugv266

What's the practical use of seeing the results of something that absolutely nobody uses?

How do you know that no one uses it? Are you working for the NSA?

[marios]

Although those benchmarks are a nice start, we need to see the effects of LTO, graphite vs poly etc. PGO and bolt, albeit useful, are too much of a hassle, since they are not just a flag.

[avis]

Good we have 2 OS compilers sort of competing with each other, more benefits and choices for us.

Since we see up to 5% difference in video codecs it certainly is valuable to compare them and assembly clearly isn't the only relevant factor.

Um, where?

SVT-AV1: ~2%
x265: ~2%

ugv266 is a dead encoder without too much assembly where yeah, up to 5%, but no one cares. VVC is hated and destroyed here on Phoronix, why would you use it? vvenc is tons better, why would you use ugv266?

What's the practical use of seeing the results of something that absolutely nobody uses? It's like the dude earlier in the thread who requested ages old Quake and Doom games to be tested. Why? This will extremely unlikely to translate to any modern workloads.