View Full Version : CPU lifespan
I'm looking for some hard data on the expected lifespan or MTBF for modern CPUs, like the P3/4 and Athlon.
I've heard over and over again that overvolting or overclocking my CPU will decrease it's overall operational life- but this is meaningless until I know how long it's supposed to last in the first place. Once I know that, I can take a shot at calculating how messing with it will affect it.
I've downloaded just about all the spec sheets availible on the Intel site, and no luck there.
NASA's Pathfinder survived 3 years of hard radiation on the way to Mars, then worked perfectly for months until the battery died. And it was powered by a 486!:) NASA don't do stuff and just cross their fingers that it works, so the data I need has to be out there somewhere!
Any ideas? I have a feeling that it is in the 10s of years, but I'd like to be sure.
All I know is that by the time your o/ced cpu dies, you won't care, and it will be going for $5 on ebay.
lol sfa- yeah, that's what I thought.
But- what if it's already a year and a half old, was OC'd to hell and back with a massive vcore increase and a peltier, and now I want to put it in a system and use it at stock speeds to run a server? It'll need to run 24/7 in a closet for the forseeable future, as in years.
See what I mean? I need to figure as closely as possible how much time it has left in it.
Shadow рс
12-28-01, 01:36 AM
just a logical guess, retail chips come with a 3 year warranty. I've seen many more chips last longer than 3 years, however I believe Intel/AMD would offer longer warranties, if nothing else but for bragging rights, if they designed their chips to last longer.
So, if you expected a chip to last 3 years...and oc'd it to the point you cut 1/3 it's life off, 2 years would be it's expected fail time. In 2 years, if I'm still using the same chip, shame on me. If I were to have the same chip in 2 years, as said above, I could probably buy it on ebay for $5 by then.
iggybaseball
12-28-01, 07:55 AM
yes do u mean lifetime of a chip physically, or until it's so old u need to upgrade?
I wouldn't use a chip that was volted to hell and had a TEC on it in a server that has to have 100% uptime, but for regualar, casual use it should be fine.
The Coolest
12-28-01, 12:41 PM
I think that the Athlons supposed to live about 10years. if u overclock it to hell, it will live lets say 5 years. would u care for it after 5 yearS? I know that I wont...
It'll propably will be back to stock speed and running in some1 else's house :D
iggy- I'm talking about physical life, like how long before it wears out.
I'm not wealthy. I have a mortgage, car payments, etc. and I can't go out and get the latest CPU just because Intel says my current one is obsolete. Just because it ain't cutting edge anymore doesn't mean I should toss it. I am the poster child for ghetto computing.
I use computer equipment until it either can't get the job done, or it breaks. Guess what- my 486 still does everything I need it to (Word, e-mail, and X-COM:)), so why should I replace it? I'll use it till it fails- if you think I'm lame cuz I don't have only the latest equipment, no skin off my nose- old PCs are still fun to me!:)
Besides, even a P Pro will have enough power for the application I have in mind. Longevity is more important for this project.
KeyboardCowboy
12-28-01, 01:45 PM
Originally posted by sfa ok
All I know is that by the time your o/ced cpu dies, you won't care, and it will be going for $5 on ebay.
i would say that this isin't far from the truth
i have a 133Mhz machine that still works fine
Bmxpunk86pl
12-28-01, 01:52 PM
i have like a 15 year old 25 mhz laptop that still works, 2 megs of ram, 80 meg harddrive. I have win 95 running on it and i just use it for instant messeging, what else on 14.4 modem.
8088 at an amazing 10Mhz with DOS 5 (got turbo enabled on startup.. closest thing to an OC for that thing :D ). Dosen't do much for me anymore but then again, that is an ancient computer. My Pentium 133 still works for everything casual like surfing (If I got a 56K modem...), e-mail, old games, and of course crunching SETI. ;)
While we're on the topic of CPU life spans, how true do people believe the 10C rule is? You know... 10C decrese in temps doubles life, while a 10C increase will do the opposite.
JigPu
Sir-Epix
12-29-01, 01:47 PM
Originally posted by Monster of Rock
I'm looking for some hard data on the expected lifespan or MTBF for modern CPUs, like the P3/4 and Athlon.
I've heard over and over again that overvolting or overclocking my CPU will decrease it's overall operational life- but this is meaningless until I know how long it's supposed to last in the first place. Once I know that, I can take a shot at calculating how messing with it will affect it.
I've downloaded just about all the spec sheets availible on the Intel site, and no luck there.
NASA's Pathfinder survived 3 years of hard radiation on the way to Mars, then worked perfectly for months until the battery died. And it was powered by a 486!:) NASA don't do stuff and just cross their fingers that it works, so the data I need has to be out there somewhere!
Any ideas? I have a feeling that it is in the 10s of years, but I'd like to be sure.
First I believe the lifespan is supposed to be 5 years or so...I read that in Maximum PC or PC Mag. Not sure which one...and it was about 3 years ago I think. Now NASA doesn't use a regular CPU for there satillites. They have special ones made to block the radiation. Hope that helps ;)
JigPu- well, that could very well be true- logically, it makes sense that a CPU will last longer at lower temperatures. I believe it!:)
Unfortunately, since nobody seems to have any hard data on the actual lifespan of a CPU, it's not too useful.
Double the lifespan would be expressed as 2(x) but since we don't know what (x) is in the first place, I'm still up Creek No. 2 without a paddle.
Looks like I'm just going to have to keep my fingers crossed on this one. Thanks for the input, all.
Now NASA doesn't use a regular CPU for there satillites. They have special ones made to block the radiation. Hope that helps
Actually, Sojourner/Pathfinder was constructed entirely of off-the-shelf materials. Including the CPU. That's why it was "Better, Cheaper, Faster" as their new motto goes. Other spacebound equip may be different, but I doubt it- Intel is not going to fab a whole new type of gamma\x resistant CPU just for NASA. Especially when you consider that every major change at a chip fab is measured in the Billions of dollars. NASA spends a lot of dough, but not THAT much.:) Easier to encase a standard chip in a rad shield- it already has free super cooling thanks to outer space.
Anyway, my point was just that NASA would never use a part that had no MTBF data associated with it. The data I seek HAS to be out there somewhere...I think!:)
WELL! Here's some interesting stuff!
Done some searching through Intel's support site, and found this lovely sheet full of info! According to this, the MTBF for Intel's CPUs are quite high. So high that the lowest MTBF is 456 YEARS!! :eek:
Check out this link... Quite interesting... Shows the MTBF for the CPU processes...
http://support.intel.com/support/fittable.htm
Also check out these...
http://support.intel.com/support/processors/pentiumiii/sb/108065652564218.htm
http://support.intel.com/support/controllers/mcs96/components/fittable.htm
Monster of Rock, MTFB and expected life span are two quite different things, and nobody even picked that up yet, hmm...
The difference is - life span is the average and MTFB is the mean, thats what the M in MTFB stands for.
Ex. MTFB is the amount of time before 50% of all tested units will be dead, which can be ways off from the average expected life span
Shadow рс
12-30-01, 09:30 PM
actually it's MTBF
Mean Time Before Failure
Originally posted by LJ5L
Monster of Rock, MTFB and expected life span are two quite different things, and nobody even picked that up yet, hmm...
The difference is - life span is the average and MTFB is the mean, thats what the M in MTFB stands for.
Ex. MTFB is the amount of time before 50% of all tested units will be dead, which can be ways off from the average expected life span
Mean and average are the same thing... Or am I going insane again? :rolleyes:
JigPu
Shadow рс
12-31-01, 07:53 AM
since it's been questioned, I'll elaborate.
A processor is designed to last X amount of time...however when it's off the "down time" adds to the life of the product.
say you have a processor that is 3 years old (life expectancy), but has only been used 1.5 years...and has been off the other 1.5. Since your processor is now 3 years old, but only used for 1.5, it should essentially last another 1.5 years, which makes the life expectancy 4.5 years.
MTBF is 1.5 years here.
I know what MTBF means. Since I didn't have either MTBF OR lifespan data, I was willing to settle for either. Hence the 'or '. :)
Both sets of data tell much the same thing, however. The manufacturer uses an extrapolate of MTBF data to arrive at an expected lifespan number. In my (decidedly) unscientific usage, any bit of data at all was helpful. Thanks all who replied!
Extra Special Magic Happy thanks to Jigpu, those links look to be some serious first-order data. I'm gonna sink my teeth in and see what I can see.
w3rd!!!
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.