Computer temperatures

[Allan]

Does anyone have a rule of thumb about heat limits for various components inside a computer. I imagine it varies widely depending on the component, the hardware etc, but some idea of what others view as acceptable (or not) would be useful. I suppose the processor is the main one, but no doubt hard drive and graphics card are important too. I read a review of Defraggler (which gives HD temps) which said "anything over 50 degrees suggests that your PC needs better cooling". Speccy gives temperature readings on several components (where available).

Speccy and Defraggler are both from

Allan (no connection with Piriform other than as a user of their software)

[Adrian]

I've got hardware sensor apps running on all the laptops. 50deg is really nothing to get excited about. When they're working hard, 70deg is not exceptional. 80-90 is unusual, and normally only seen when they're working very hard and sat on something that's blocking the cooling vents. IIRC they start to slow down and cut off around 100deg.

[Martin Brown]

It is sort of true, but nothing really to obsess about. Mine runs warm when all four cores are run flat out with full 3D textured graphics.

Higher temperatures shorten the lifespan but silicon isn't too bothered about being run fairly hot so long as you can't touch it. I'd probably be worried if the CPU was showing much above 70C for extended periods.

Run much above 100C will measurably shorten their life. To put it into perspective the CPU core has an energy density roughly the same as a soldering iron and so needs active cooling to keep working.

If you are a dedicated gamer with a super hot (in both senses) graphics card or two running flat out then you need to pay more attention to the cooling but for ordinary users it is not usually a problem unless a fan gets gummed up with hair and dust and stops turning altogether.

[Tim Watts]

On Tuesday 26 November 2013 10:23 Allan wrote in uk.d-i-y:

CPU and a *bridge chips can get pretty hot as in 60-80C, sometimes more.

The rest of the system should be < 40C or so ideally as it's then the electrolytic caps and spinning disks that start to not like it so much.

[Martin Brown]

It is sort of true, but nothing really to obsess about. Mine runs warm when all four cores are run flat out with full 3D textured graphics.

Higher temperatures shorten the lifespan but silicon isn't too bothered about being run fairly hot so long as you can't touch it. I'd probably be worried if the CPU was showing much above 70C for extended periods.

Run much above 100C will measurably shorten their life. To put it into perspective the CPU core has an energy density roughly the same as a soldering iron and so needs active cooling to keep working.

If you are a dedicated gamer with a super hot (in both senses) graphics card or two running flat out then you need to pay more attention to the cooling but for ordinary users it is not usually a problem unless a fan gets gummed up with hair and dust and stops turning altogether.

[Andrew Gabriel]

I joined a startup nearly 20 years ago, and took on some of the server support tasks as no one was doing them, and there were problems with reliability of the main server and the disks. I quickly learned that hot disks die, and built disk/server enclosures which kept the disks cool with a howling gale blowing across them. Never lost another disk.

I still do that to this day in my own servers, although I put the server somewhere out of ear-shot so the noise doesn't matter.

General rule for electronics is the life halves for every 10C temperature rise. In most cases the life is going to be way longer than the system is going to remain viable due to obsolescence, but if you run a system hot for a long time, you might find it gives up on you before you give up on it.

[alan]

Look at the specification for a hard disk - usually around 60C maximum.

[Mark]

I'm also a firm believer that keeping HDDs cool extends their life. I make sure there is a fan blowing directly onto the drives in my desktop PCs.

[Theo Markettos]

Curiously, Google think otherwise:

Though it may depend on your definition of 'hot', and the graph suggests hot old discs die but young ones are more likely to die cold.

Theo

[meow2222]

/disk_failures.pdf

AIUI hot means around 60c and up, and the pdf only graphs temps upto 45C - the above 45C category is rather ambiguous. What little data they show arou nd the 45C region does demonstrate that failure rate climbs with increasing temp there. Unfortunaetly it tells us nothing about higher temps, but the graph shape suggests that its likely failure rates climb significantly as t emp goes up to 50,60C etc. Its a shame they failed to present key data usef ully there.

NT

[whisky-dave]

I think mine run about 50C in my iMac.

But one thing missed is where you measure the temerature.

I did some experiments using a 5V regulator 7805, it was OK to about 120C case temp so internally it was hotter. after about 125C the noise on the 5V tripled and a few degrees more the regulator started to shutdown and dropped to below around 4.5V. Squarting some freezer spare on it helped and a bigger heatsink also help keep temeratures down. The regulator didn't appear to be damaged at all and recoved quite quickly. Processors tend to run hotter typically 70C to 100C

[RJH]

Indeed, the figures would be useful. Fig.5 is for me most telling - colder temps associated with a higher failure rate for drives less than

3 months old. But then things even out for older drives. And then there's the use - these are server drives.

Of interest to me as I have a NAS in the cellar, which is 8C +/- 5C year round.

[docholliday93]

Which will mean that some of the components within the hard disk will be running significantly hotter than that...

[Rick Hughes]

I use SpeedFan to monitor my PCtemp ...

Shows the Idle temp (email browsing etc.) and when Processor running at close to 100% (Video rendering)

CPU(i5) 30 60 Core 0 25 42 Core 1 30 51 GPU 32 35 System 26 26