Conversation

Notices

1. Embed this notice
   Gabriele Svelto (gabrielesvelto@fosstodon.org)'s status on Thursday, 13-Mar-2025 23:54:26 JST Gabriele Svelto

   I've been filing bugs to improve Mozilla's crash reporting infrastructure so it's easier to see at a glance if a particular crash affects only a specific CPU but also if only some microcode versions are affected.

   I said it before but I'll say it again. I would have never thought that CPU bugs would have become an everyday occurrence, but here we are.

   • Embed this notice
     翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Thursday, 13-Mar-2025 23:54:23 JST 翠星石

     in reply to

     • Ben Hearsum, with elbows up
     • Greg Tatum
     @gregtatum @bhearsum @gabrielesvelto CEMENT
   • Embed this notice
     Ben Hearsum, with elbows up (bhearsum@fosstodon.org)'s status on Thursday, 13-Mar-2025 23:54:25 JST Ben Hearsum, with elbows up

     in reply to

     @gabrielesvelto this is the most depressing part to me. We seem to have given up on understanding and rigour, and decided we'll just accept half baked shit. But this is, perhaps, simply the way of the world at this point.

   • Embed this notice
     Greg Tatum (gregtatum@fosstodon.org)'s status on Thursday, 13-Mar-2025 23:54:25 JST Greg Tatum

     in reply to

     • Ben Hearsum, with elbows up

     @bhearsum @gabrielesvelto Remember Windows ME? https://en.wikipedia.org/wiki/Windows_Me#Reception

   • Embed this notice
     Gabriele Svelto (gabrielesvelto@fosstodon.org)'s status on Thursday, 13-Mar-2025 23:54:26 JST Gabriele Svelto

     in reply to

     But by all means, keep pouring capitals into "artificial intelligence" while the very foundation of computing is falling apart and we can't trust a machine to execute instructions reliably.

   • Embed this notice
     翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Friday, 14-Mar-2025 00:02:14 JST 翠星石

     in reply to

     • Eniko Fox
     • wlach
     @gabrielesvelto @wlach @eniko CPUs have recently become more and more buggy.
     
     Intel in the core 2 Duo days intentionally released untested buggy garbage and later fixed some of the most severe bugs via proprietary software microcode updates (that definitely doesn't also contain malware), but even then, those are still plenty stable for 20 hour Church of Emacs sessions and 20 hour compiling session, but good luck trying to watch slop videos for >20 minutes.
     
     Intel has only gotten worse, with CPUs that fry themselves with voltage.
     
     AMD did previously actually deliver decent products from the factory - for example the Opteron 62XX series are very stable without any proprietary software updates, but later CPUs like Ryzen are buggy garbage.
   • Embed this notice
     wlach (wlach@mastodon.social)'s status on Friday, 14-Mar-2025 00:02:15 JST wlach

     in reply to

     @gabrielesvelto Are you saying these defects are becoming more frequent in newer CPUs? Or is this a problem that has already existed, and we're just getting better at detecting them? This would make a great hacks post (if that site still exists?)

   • Embed this notice
     Gabriele Svelto (gabrielesvelto@fosstodon.org)'s status on Friday, 14-Mar-2025 00:02:15 JST Gabriele Svelto

     in reply to

     • wlach

     @wlach it wasn't a thing which is why we have "famous" CPU bugs such as the Pentium FDIV issue. In the last few years it has become more and more common, with certain CPU families being particulary affected such as AMD's Jaguar and first Zen generation, as well as Intel's early Atom processors and lately Raptor Lake and its derivatives.