This article is more than 1 year old
Microsoft: Don't overclock Windows 8 unless you like our new BSOD
Redmond boffins burn up CPUs so you don't have to
Microsoft's Blue Screen of Death is the curse of Windows. Not just its appearance, mid way through some serious game play or spreadsheeting, but the messages themselves - digital monologues on the existence of a problem, its possible causes and how you can fix it.
Yeah, right. Just shut down and re-start like everybody else.
With Windows 8 Microsoft has re-engineered BSOD for the era of Twitter and texting - unhappy smilies and chirpy one-liners.
Based on some Microsoft research, though, it seems you could be seeing more of the Windows 8 BSOD if you buy a PC from an OEM who's fiddled with the chip to make it go that little bit faster. Also, beware the temptation to buy a PC from an unrecognized PC maker.
A Microsoft Research report, published in April 2011 but only just coughed to the top of the web ahead of the Windows 8 launch, has found that overclocked CPUs are substantially more likely to make a Windows PC crash than chips left untouched. The message: pick your PC wisely and resist home-brew fiddling.
Also, it seems that PCs from makers in the top 20 are less prone to crash than machines from white box manufacturers.
The Microsoft Research white paper based its findings on reported hardware crashes, taken from millions of PCs using Windows Error Reporting.
The report is based on something Microsoft called Total Accumulated CPU Time (TACT) - continuous uptime that excludes machines in sleep mode or turned off.
According to Microsoft, the longer the TACT - the longer a PC runs continuously, without shutting down - the more likely it is to experience its first crash thanks to a CPU failure. And once it's crashed once, chances are increased it will keep crashing.
The probability that a machine with five days TACT would crash was one in 330 compared to one in 190 for a PC with an uptime of 30 days.
Once a PC crashes, its crash probability rate goes up by a factor of 100 and for a second and third crash.
On machines that ran for five days, the probability of a second crash jumped to one in 3.3 and chances of a third failure were one in 1.8.
Buying a branded PC instead of a white box? Wise choice, as the probability of crash due to CPU failure is significantly reduced.
Microsoft found CPU subsystems of brand name PCs was one in 120 versus one in 93 for white boxes machines when tested for 30 days of continues running.
Microsoft identified brand name as one of the top 20 OEMs by worldwide sales volume - so companies such as Hewlett-Packard, Acer, Asus, Dell and Lenovo.
Overclocking of CPUs dramatically increased the probability of that first PC outage.
Microsoft Research compared chips from two different vendors - simply called vendor A and vendor B - and found overclocking hit them both right in the reliability bracket.
Overclocked chips from vendor A were 20 times as likely to crash during Microsoft's eight month test period after a period of five days of continuous running versus chips from vendor B, at four times as likely. Once overclocked, chances of the first crash for vendor A were one in 21 and one in 2.4 for a second; for vendor B, it was one in 86 and one in 3.5.
The relatively good news was the probably of follow-on crashes for overclocked systems were about the same as those for tamper-free chips - albeit incredibly high.
Amateur hacks aside, CPUs are often sold at a rated frequency by the PC maker that differs slightly from the frequency set by the chip's maker.
Laptops were less likely to crash than desktops: laptops were between 25 per cent and 60 per cent less likely than a desktop to suffer a hardware failure and crash during 30 days of uptime.
The full report also covers DRAM and disk subsystem failures.
You can read the full report here. ®