I recommend downloading the program SumatraPDF, it may come in handy if you encounter any PDF files in the future that you may open them.
Here's a link:
https://www.sumatrapdfreader.org/free-pdf-reader.html9.
ConclusionThis paper presents the first large-scale study of hardware-
induced operating system failures on consumer machines.
Using post-hoc analysis of machine status reports and OS
crash logs, we find that:
•
Failure rates are non-trivial. Hardware crash rates are
up to 1 in 190 over an 8 month observation period.
•
Recurrent failures are common. Hardware crashes in-
crease in likelihood by up to two orders of magnitude af-
ter a first such crash occurs, and 20% to 40% of machines
have faults that are intermittent rather than transient.
•
Recurrent failures happen quickly. As many as 97% of
recurring failures occur within 10 days of the first failure
on a machine.
•
CPU speed matters. Even minor overclocking signifi-
cantly degrades reliability, and minor underclocking im-
proves reliability. Even absent overclocking, faster CPUs
become faulty more rapidly than slower CPUs.
•
DRAM faults have spatial locality. Almost 80% of ma-
chines that crashed more than once from a 1-bit DRAM
failure had a recurrence at the same physical address.
•
Configuration matters. Brand name desktop machines
are more reliable than white box desktops, and brand
name laptops are more reliable than brand name desk-
tops. Machines with more DRAM suffer more one-bit
and CPU errors, but fewer disk failures.
Figure 15 shows a terse summary of our findings, with
comparisons against previous results where available.
Fyi every overclocker knows system needs to be tweaked and tweaked throughout time