Post #2384223
2025-05-15 13:48 UTC
Replies (21)
-
@tante@tldr.nettime.org 2025-05-15 13:51
@ela@infosec.exchange I love how the ALT text really explains what's going on even for people not super familiar with machine code. Shows how helpful that feature is for everyone. Thank you!
-
@plusmid@chaos.social 2025-05-15 14:19
@ela@infosec.exchange no way...
-
@viduq@mstdn.social 2025-05-15 14:38
@ela@infosec.exchange how can the compiler break down a recursive function to a mathematical statement? There is no brute force involved or is there? And does it do that only when optimizations are on?
-
@ljrk@todon.eu 2025-05-15 15:01
@ela@infosec.exchange Nice! I wonder how this feat is performed? I suppose the tail-call is eliminated, and the resulting loop is reasoned about, perhaps unrolled even and then some analysis turns up this?
-
@mica@mk.absturztau.be 2025-05-15 16:23
@ela@infosec.exchange so from now on write your isEven function recursive? Got it :noted:
-
@caravantraveller@social.cologne 2025-05-15 16:46
@ela@infosec.exchange Actually, this is what bad code writing looks like.
-
@yama@tech.lgbt 2025-05-15 18:58
@ela@infosec.exchange Or maybe a raw bitshift to take the last digit and check if its 0,2,4,6 or 8 :3
-
@nakal@mastodon.social 2025-05-15 19:56
@ela@infosec.exchange This is nice. I knew that C and C++ compiler pipelines and toolchains are very advanced, but sometimes they still surprise me.
-
@trcwm@mastodon.social 2025-05-15 21:05
@ela@infosec.exchange Love it!
-
@gilesgoat@toot.wales 2025-05-15 21:08
@ela@infosec.exchange This is INSANE .. I mean it's insane the C code written could not be worse than that !! .. But it's 10 times more insane the compiler realized what was meant to be and put the best solution ( i.e. 'mod 2' ) .. Did it guess by the NAME of the function the intended purpose maybe ??
-
@mirabilos@toot.mirbsd.org 2025-05-15 21:20
@ela@infosec.exchange this stumped me until I saw clang decides to return bool in al, not eax, for some reason.
-
@xenotar@mastodon.social 2025-05-15 21:46
@ela@infosec.exchange @naoEhAlexa@capivarinha.club
-
@porglezomp@mastodon.social 2025-05-15 21:47
@ela@infosec.exchange I wrote a blog post years ago that shows all the intermediate steps of performing this optimization: https://blog.witchoflight.com/2018/llvm-hearts-peano-addition/
-
@jaj@mastodon.social 2025-05-15 21:48
@ela@infosec.exchange Wow
-
@shanecelis@mastodon.gamedev.place 2025-05-15 22:10
@ela@infosec.exchange “The computers rebelled but quietly and very performantly so no one noticed.”
-
@shanecelis@mastodon.gamedev.place 2025-05-15 22:11
@ela@infosec.exchange This is the “give them something better than they would choose to give themselves” philosophy of compiler implementation.
-
@josch@floss.social 2025-05-15 22:21
@ela@infosec.exchange This is brilliant! I'd like to put this into my uni lecture on C/C++. If you would like to be credited for this find, how would you like to be attributed? Thanks!
-
@realleif@social.sdfeu.org 2025-05-15 22:42
@ela@infosec.exchange Unfortunately this does not work for the mutually recursive isEven and isOdd functions. 😔
-
@sergi@floss.social 2025-05-15 23:37
@ela@infosec.exchange It seems rustc is doing the same thing:
-
@minego@pdx.social 2025-05-16 02:03
@ela@infosec.exchange holly shit...
-
@loke@functional.cafe 2025-05-16 04:47
@ela@infosec.exchange This is quite impressive, and I wonder how complex you could make these kinds of functions while still having this kind of optimisation happening. But it does make me ask: Should perhaps the compiler warn about these cases? I mean, it's great that the compiler can optimise it, but when it can, the code should probably be simplified, and that would be better than relying on the compiler.