I recall hearing CPU Tmax is designed for a 10 year lifespan.

I have never used a laptop for a decade, nor have I had the CPU fail. So perhaps faster performance and shorter lifespan is okay.

* * *

I have a i7 laptop of the same generation that I'm using to type this. I've replaced the screen, bezel, keyboard, touchpad, WiFi board, charger (3 or 4 times, thanks a lot Dell), battery. Upgraded the memory and added an SSD. Reapplied CPU paste once or twice. Never had a single problem with the CPU. Actually I don't know that I've ever had a computer with a CPU failure.

In all I've spent about 25% of the cost of the laptop over the years in repairs, but the result is that I'm able to keep using a device that's almost 12 years old instead of junking it. Saved quite a bit of money vs a new device, too.

Incidentally, it's interesting in regards to this issue that these old bulky laptops with beefy chips are actually quite good at avoiding thermal throttling. Despite having a traditional loud fan, I rarely even hear it come on.

This reminds me of Trigger's broom.

For non-brits: Triggers broom is a joke revolving around the ideas of the ship of theseus.

In the show, trigger is a street sweep who has "kept the same broom" for 10 years, only periodically replacing the brush and occasionally replacing the handle.

The joke is of course, that it's no longer the same broom.

https://www.youtube.com/watch?v=56yN2zHtofM

In parts of the US it's "grampa's axe": I've replaced the handle 3 times and the head twice!

I'm impressed that you only spent 25% of the laptop's purchase price on so many repairs. Where did you get the parts?

He cannibalized the flesh of others, so his companion may live a damned existence!!

(Response crafted by anti-repair lobby gpt)

Even if I am of the same type: boosted macbook pro with ssd and memory beyond what is officially supported, I think integration of almost everything on one board skipping connectors, sockets, is more cost effective and more "green" overall.

Dies of silicons are packaged, socketed, etc instead every part of silicon will get closer in a package and will handle everything.

Faster, cheaper, more energy efficient.

Did you ever overclock, though? My 2500k went from 5 to 4ghz over the course of its life.

No. Ironically I bought the unlocked version then never took it past stock speed. That's why I didn't bother getting the unlocked new one.

> ...was getting flaky powering back on when turned off, which I assumed was the motherboard.

This is my PC right now. Sometimes it does a boot loop, where it can get to BIOS but not GRUB and I'm not sure why, as there are no error messages. Pulling the plug and plugging it back in helps.

I'm not sure whether it's the motherboard, or maybe a disk not wanting to read data or something. Guess I'll have to wait and see, eventually swapping out parts one by one once it fails entirely.

There's a lot of 10+ year old CPUs in use, with heatsinks so clogged as to be at their thermal limit all the time, that I think 10 years is a highly conservative estimate which also assumes absolute max Vcore (which if you've seen the datasheets, are far higher than normal values, and something that only extreme --- as in, liquid nitrogen --- overclockers will exceed); it's more likely other things like motherboard capacitors will fail first.

Fwiw, I have 3 ThinkPad t420 and t420s running around the household, used daily, they're at about a decade.

Just replaced my main desktop which was fx8350, but that computer is now on its 2nd life with a friend and shows no signs of stopping.

There was a time when 10 year old computer was absolute rubbish. These days, if you upgrade from spinning drive to ssd, 10 year old well made computers are just fine :)

Basically, in 3 decades of personal computing, I never had a CPU failure other than that one time I crushed atlon xp with the heatsink :O

IIRC Intel consumer CPU target 5 years (or is it 3? Not sure and the info is not public) at 30% duty cycle.

Of course lots of CPU make it for longer, but 10 years is certainly not a design target.

How relevant is that figure when Intel CPUs universally outperform that by leaps and bounds?

If you actually use consummer CPUs a lot (maybe e.g. you are a kernel maintainer and compile Linux all day) some of them may not last how long as you think.

Or if you put consummer CPUs in embedded equipements that are always (or at least often) under load, Intel will not care much if you have an high failure rate after a few years, even if you buy an high volume. They have dedicated SKUs with better expected durability.

My T430s is still going strong.

I still use a T420 (can't stand chicklets) and a repaste does the trick.

My 2005 vintage T42p wants to have a word with you. In all its years it has seen the following parts replaced: the (PATA) 2.5" hard drive, now using an SSD through a PATA-SATA adapter.

That's it.

I still regularly use this machine even though it not longer is my main portable device. The only thing keeping me from doing so is its 32-bit CPU which now that 32-bit support has mostly gone the way of the dodo. I much prefer its 4:3 form factor 1600x1200 screen, keyboard and construction over the P50 I'm using in its stead.

I don't understand why this is a problem for Linux and not fit Windows.

The computers don't break on Windows, why Linux uses should take this overly conservative approach that limits the performance of their computer?

Intel ship DPTF drivers for Windows, so the appropriate thermal policy is configured at boot time. If you install thermald then Linux will do the same.

In the case of T480 the limit is specified by whether the laptop is in your lap or on the table. In the latter case it allows a much higher limit.

This check doesn’t work on Linux, so for safety reasons the former limit is enforced.

How does that check work? Do you have details?