"David H. Burns" <dhb### [at] cherokeetelnet> wrote:
> I blamed those "access violations" that occasionally show up on Windows. So
> their really due to a "safety" feature in Intel's hardware. Is that why
> machines with
> AMD processor's (reportedly) crash and hangup less frequently?

No, not really: AMD processors use the very same features.

They do have a fierce and long-standing competition however, as to who can get
the most computing power out of a single chunk of silicon; naturally, this
involves pushing the CPUs quite close to their limits of stable operation; if
these limits are exceeded, once in a while a glitch may occur and just throw
the computer off-track; in such a case, anything might happen then, from a
computation error, to a messed up address pointer or misread programm command
that may have the CPU's / operating system's safety measures kick in as
response.

I guess it is worth noting here that not every such access violation message is
due to a glitch in the CPU; probably the vast majority of such messages in
everyday use is due to straightforward programming errors (often flawed pointer
handling in C/C++ code). Instead, true glitches are most frequently observed by
high-end gamers trying to push the system even closer to its limits than AMD or
Intel decided to, by operating them beyond the official operational parameters
(bus clock, clock multiplier, memory access timings, voltages and the like).

The allegations about processor stability tend to change over time, too: A few
years ago, it was Intel CPUs that were considered more stable.


So to wrap it up, those "access violations" are actually the applications' fault
- even though they seem to be a rather Windows-typical phenomenon, but I
attribute this to some good degree to the fact that Windows just happens to be
the most popular target for unexperienced or lazy program authors.

Post a reply to this message