Hacker News Reader

Comment by ahelwer | Hacker News Reader

CPUs are immortal, it's the motherboards that die. If you want to run a certain chipset, after some amount of time it's motherboard availability that will keep you from doing that!

View on HN

CPUs are not immortal. One source of failure is electromigration: https://en.m.wikipedia.org/wiki/Electromigration

With RAM it’s a more common problem. There’s even software to check for it (memtest86)

I read that the reason Intel switched to LGA pins from standard sockets is the chip is more expensive and less likely to die than the motherboard, so it makes more sense to put the fragile part of the interface on the motherboard.

> CPUs are immortal, it's the motherboards that die

Can you elaborate?

There are more separate elements of lower quality testing than CPUs (which go through serious qa and binning). For example capacitors on motherboards break/explode, PCB traces degrade/crack more often than a CPU will break.

CPUs have a lower failure rate than motherboards, which makes sense if you think about it. (One is an IC, one is a complex assembly of many ICs and passive components.)

Anecdotally, I'm still running a computer I mostly bought in 2011, i5-2500K still going strong, but the motherboard has a few issues: one of the PCI-E slots works sometimes, one of the RAM slots is completely broken, and the RTC seems to lose a few minutes a week.

CPUs should go pretty much forever, provided they're run on clean power. They're a little square, entirely encased. Motherboards are bigger and with physical connectors you need to use force on.

I'm also on an i5-2500K from 2012, I only upgraded the GPU so far. I recently ordered a new pc for roughly the same price as the one i built in 2012, curious to see how much of a difference I will notice day-to-day.

Back in the day the story was that a knockoff manufacturer of capacitors used an incomplete recipe—the caps started out fine but didn’t last. Wikipedia says the story is unverified but still includes it under the “Investigation” section:

https://en.wikipedia.org/wiki/Capacitor_plague

The plague is true... I was repairing motherboards by replacing vrm capacitors around 2008.

Anecdotally, I built 4 computer labs at a college with Dell optiplex desktops, and a year later, 30 of the 104 had their motherboards replaced due to blown caps. And at the time, the nearest certified tech was 70 miles away, over a mountain pass.

So litterally up hill both ways

No moving parts, small and stable voltages, relatively stable temperatures, no capacitive sections to experience electromechanical strain. There's essentially nothing that can "wear out" in a CPU. Basically the only things that can degrade a CPU are if you run it on a noisy voltage source (unlikely, behind a power supply), or if you regularly turn it on and heat it up a lot and then allow it to cool. But if you keep a CPU well cooled or just run it constantly it will last essentially forever.

This isn't true. Aging is now more of a concern at 16/7nm to the point there's extra margin specially for it. But all that margin does is delay the impact beyond the realistic lifetime of the chip (normally approx 10 years for consumer stuff). Sometimes you can't afford the extra timing margin and you have to use extra anti-aging circuitry instead.

https://semiengineering.com/chip-aging-becomes-design-proble...

> There's essentially nothing that can "wear out" in a CPU

There's electromigration which is harder and harder to mitigate with smaller process nodes

I assume RAM is the same in this regard. If it passes memtest86, it's golden.

Reply on news.ycombinator.com