Comment on Unreal Engine supervisor at ModelFarm blasts 50% failure rate with Intel chips — company switching to AMD's Ryzen 9 9950X, praises single-threaded performance

Laser@feddit.org ⁨1⁩ ⁨month⁩ ago

• Sweetpeaches69@lemmy.world ⁨1⁩ ⁨month⁩ ago

  Better off buying from Temu at that failure rate.

  source
• Blackmist@feddit.uk ⁨1⁩ ⁨month⁩ ago

  This is probably a worse rate than all those motherboards years ago with the fucked up capacitors.

  We had fucking stacks of them.

  source
• Monomate@lemm.ee ⁨1⁩ ⁨month⁩ ago

  Why are grey market CPU parts more prone to failure? For GPUs I can understand due to possible mining usage, but CPUs too?

  source

  • Laser@feddit.org ⁨1⁩ ⁨month⁩ ago

    Grey market chips usually include chips that failed quality assurance to prop up numbers.

    source

    • Monomate@lemm.ee ⁨1⁩ ⁨month⁩ ago

      Most CPUs I’ve found in these Chinese sites claim they’re used parts, probably from old servers from Chinese companies, which explains the amount of Xeons being offered. But if the part comes in the original box, why would Intel/AMD create an official package for these failed QA parts?

      source

      • nomous@lemmy.world ⁨1⁩ ⁨month⁩ ago

        Counterfeit packaging seems easy enough to produce.

        source

        • -> View More Comments
      • surewhynotlem@lemmy.world ⁨1⁩ ⁨month⁩ ago

        If they’re swiping failed QA chips, it’s easy to swipe a couple boxes at the same time.

        source
• werefreeatlast@lemmy.world ⁨1⁩ ⁨month⁩ ago

  I’m ready for the eBay batch purchase…batch of 10 please!

  source
• merc@sh.itjust.works ⁨1⁩ ⁨month⁩ ago

  I’m curious about the timeframe for failures though. Like, if these machines are being used for feature film VFX, I wouldn’t be surprised if the CPU is running at near 100%, 24/7 for months on end. If it fails after 6 months under those conditions, a typical home user might be able to go years without an issue. Of course, there would probably be unlucky people who have problems long before that.

  It’s also interesting that we’re not hearing anything from Amazon, Google or Microsoft. They Intel-based servers and they also push them hard. Are they not seeing these problems, or are they just not talking about them? If they’re not seeing them, is it safe to push the affected Intel CPUs hard as long as you avoid very specific code / algorithms?

  source

  • Laser@feddit.org ⁨1⁩ ⁨month⁩ ago

    The news is about Core i9 CPUs, which are the enthusiast / possibly workstation offering by Intel with high single core clock speeds. Amazon, Google and Microsoft use server CPUs which usually don’t feature such high speeds, but rather focus on more cores and more possible RAM.

    However, if a vendor sells a product with the main feature of high clock speeds and the product fails when I’m using that exact feature for prolonged periods of time, I’d say it’s faulty.

    source

    • merc@sh.itjust.works ⁨1⁩ ⁨month⁩ ago

      Amazon, Google and Microsoft probably do mostly use server CPUs, but I’d be surprised if there weren’t a few special projects that used specialized desktop CPUs. And, at their scale “a few special projects” probably means a few hundred machines.

      But yeah, clearly these CPUs are faulty.

      source