How quickly do silicon chips degrade?

The fans will probably maybe fail. The silicon is fine because miners usually undervolt and underclock to keep temperatures and power bills down.

They have no moving parts which degrade, only heat can damage them, still the chips can take some damage and continue dunctioning by turning off the damaged parts which leads to a loss of performance. To keep a chip alive as long as possible either cool it very well or turn down its clock speed and voltage, there's no real way to tell whe your chip will fail during normal operation, could be the day you bought it, could be in 30 years.

https://www.youtube.com/watch?v=44JqNJq-PC0
TLDW: no.

>>61978927
Undervolt or not, as long it's being used it'll degrade especially current silicon designs.
>>61978970
I'd take linus sex tips before i take his tech tips.

>>61978893
Its called bake and dump,the cards will stop working in a week.

Probably the thing that degrades first is capacitors, not silicon chips, especially they've been held under higher than usual temperature

>>61978935
> still the chips can take some damage and continue dunctioning by turning off the damaged parts which leads to a loss of performance.

[citation needed]

>>61979332
>as long it's being used it'll degrade especially current silicon designs
I sure hope you have a source to back that up. I myself have experienced zero silicon degradation in anything I've been involved with.

>>61978935
>the chips can take some damage and continue dunctioning by turning off the damaged parts
That's just bullshit. You're confusing this with fusing off damaged parts during manufacturing.

>>61978893
How much is he selling them for. If it's cheap enough, it could be well worth picking one up just to see if there are problems with it.

>>61978893
why's the cryptonigger selling them?

>>61982257
It literally says in the OP.

>>61978893
If they are really only an year old, then they are still under the 3 year warranty most cards come with.

If it's a cheap card, go for it, if the fans fail or are failing, RMA it, and viola new fucking card.

>>61978935
Silicon degradation isn't just heat damage. They do fail as a function of being used (electrons damage the physical structures over time), and most chips are designed today with a 10-12 year MTBF at stock settings.

I run a 7 year old Xeon, it's stock VID table was 1.005vcore at stock speed. This thing was presumably used as a server chip under moderate to high loads for four and a half years until I got my hands on it, and then I overvolted it 280mv, overclocked it ~1.2Ghz, and ran that for 3 years. A few times a year I crank everything up to absolute maximum and bench new programs and games to dickwave having a 7 year old machine as fast as a 4790K.

Only two weeks ago when I checked has the VID gone up a single bin, to 1.012vcore at stock speed. yet the CPU doesn't require more vcore at it's 24/7 overclock. But also I was getting RAM POST issues until I raised the timings from 7-8-7 to 8-8-8, meaning the PHY is also degrading over time.

Yes it's a very slow effect, yes it does indeed happen.

>>61978893
Degradation is a meme imho. I ran a Core2Duo overclocked for 7 years and it still works.

I'm currently on a AMD FX 8320 running at 4.2 GHz (stock is 3.5) for about 4 years no problem.

>>61982498
Fantastic and informative post. Thanks.

>>61978893

I'm still using a pair of stock R9 290Xs I bought during the first huge bitcoin sell-off a few years ago. I've never had any issues with them.

I'm eagerly awaiting the newest bitcoin GPU market crash so I can snag a cheap as fuck 1070 or 1080

>>61982498
>But also I was getting RAM POST issues until I raised the timings from 7-8-7 to 8-8-8, meaning the PHY is also degrading over time.
To be fair, though, if you had to raise the timings, that wouldn't be a problem with the memory controller, but with the RAM itself. Raising the timing does nothing with the bus signalling, but only determines how many cycles to wait for certain operations to complete. If you had a bad PHY, you would've had to lower the frequency instead.

>>61982585
>Degradation is a meme imho. I ran a Core2Duo overclocked for 7 years and it still works.
Just to reinforce this, I ran an Athlon 64 X2 for 9 years with no problem. It still worked perfectly well, I just replaced it because I wanted a new motherboard.

>>61982466
/thread

>>61982498
>Only two weeks ago when I checked has the VID gone up a single bin, to 1.012vcore at stock speed
I highly doubt that the chip has the capability to continually monitor its performance and update its announced VID over time. Unless you have significant evidence to back this up, I'm going to say you're seeing some other effect, depending on how you're reading it out.

File: 1498052970456.gif (127KB, 600x600px) Image search: [Google]

127KB, 600x600px

>>61982620
>Athlon

Those were the times. I remember all those buttblasted Intel fanboys on the schoolyard.
I hope Ryzen does the same for my Nephew's generation.

>>61982609
Being x58, the platform is known to act up when using tri-channel. I CAN get 7-8-7-22 to boot and recognize the full 12GB but it's a crapshoot, 85% of the time the system will only see 4 or 8GB

>>61982718
I was under the impression that even older CPUs are able to self-test boot conditions to select an appropriate VID from the firmware table, but am having trouble finding the same info.
I'm not talking about a sensor network responding to live conditions.

>>61982833
>Being x58, the platform is known to act up when using tri-channel.
Nonetheless, timings are a higher-level protocol setting. They don't affect the PHY operation at all.

>>61982620
same here, got Athlon X2 5600 for 8 years. Motherboard always fail but the chip is fine.

Shouldn't 1 year old cards still have warranty on them? Buy three, rma them and buy a liquid cooled Vega 64 from the money.

>>61982938
>Motherboard always fail
In my case, the motherboard hasn't gotten worse, it was just a bit weird from the outset. The nForce chipset seems to have a few bugs in the S-ATA implementation that would sometimes cause it to lock up in weird ways as a result of hotplugging drives.

>>61982938
Tbh I've never had a motherboard fail on me in the 20+ years I've building my own PCs.

Most of the time it was the RAM that died.

>>61982982
In my case, I've never had trouble with any solid-state component at all. I've had caps, fans and hard drives die multiple times, but never a chip.

>>61982466
>If it's a cheap card, go for it, if the fans fail or are failing, RMA it, and viola new fucking card.
Mining cards might be not eligible for RMA

>>61982466
>RMA it, and viola new fucking card.
you can RMA it if you're the first owner only.

>>61983177
Holy shit you guys are cucked. No once cares about it here, until you have the papers.

Silicon lasts forever

>>61978927
>>61978893
This.

Still running a 7870Ghz from the first BTC craze.
1 of 2 fans have failed (about 2 months ago, after ~4 years) but it's otherwise fine.

Replacement fans for most GPUs are like $20-30 off ebay (and usually have all the fans, so if it's a 3 fans card, you get 3 fans, since they're usually wired together)