Archived from groups: alt.comp.hardware.homebuilt (
More info?)
On Wed, 31 Aug 2005 19:50:13 +0200, Mxsmanic
<mxsmanic@gmail.com> wrote:
>kony writes:
>
>> yes it can, because "100%" is an arbitrary indicator of what
>> the CPU is actually doing. Consider that ACPI "idling" can
>> just as well be described as taking up what approaches 100%
>> of a CPU's time, it is "busy" idling in that case, not busy
>> performing operations.
>
>When a computer is not busy, most systems will halt the processor,
>period, and in modern processors, a halt brings power consumption to a
>very low level.
No, not "period". It is a specific setup option. Most
users will accept defaults, putting their ACPI power
management active and allowing a system idle process at
lowest priority.
>
>A loop, in contrast, pushes roughly the same instruction mix (which
>may contain tens or hundreds of thousands of instructions) through the
>processor over and over.
"May contain".
Test it.
Do stress tests known to maximize CPU temp and compare to
these "loops". A loop actively doing something "could"
approach maximal temps, but the fact that it's looping at
"100%" is most definitely not evidence of maximum thermal
output.
I have done such testing. Anyone not running ACPI power
management, and/or old Athlon boards that disabled it, are
running one of these hypothetical "loops" even when Task
Manager doesn't attribute it to any particular process. It
is quite clear that CPU temps further elevate depending on
the specific work consuming this "100%" time.
>Most processor generate maximum heat when
>put under this kind of load, with some slight variations depending on
>the instruction mix (but most real-world mixes are fairly random).
Absolutely not. None do.
>
>My PC is not a laptop and makes no use of power management. If it
>gets stuck in a loop, the system will run as fast as it can go until
>something takes it out of the loop.
"Fast" and heat generation are not same thing.
A system can be busy 100%, but that doesn't mean the CPU is
operating at maximum IPC potential.
>
>The same type of loop on the old AMD system that I replaced is what
>resulted in a meltdown of the processor (at temps in excess of 130° C
>at least).
No, the meltdown resulted from running a system with a
substandard fan.
>
>> Nope, CPU stress type programs show this difference. You
>> are presuming that a "100%" busy CPU is actively processing
>> as much as possible but in fact the core can be mostly idle
>> and still be "100% busy".
>
>The loop in which the CPU was running was a game (SimCity4). I don't
>think it was mostly idle.
You are arbitrarily defining "loop". Of course any program
that doesn't terminate is a loop, BUT it is very clear some
stress CPU more than others. Test it, this is easily
reproducible.
>
>> yes/no/maybe, it's what the whole subtopic has been about.
>> If the CPUs are operated in spec, it can be expected a
>> faster MHz CPU with all other parameters equal, will be
>> hotter.
>
>But it will still be in spec, so that won't matter.
Well that's the supposed and yet variable assumption being
addressed, it's a bit redundant to try to conclude it based
on the prior erroneous assumption.
>
>> System, not CPU. System MUST be set up to accomodate a true
>> (as near as possible with software available) full load, not
>> just some stuck loop, continuously.
>
>The system is not Intel's responsibility.
Never claimed it was.
>Only the processor is
>Intel's problem (unless you buy other components from Intel). The
>system as a whole is normally the gamer's responsibility, and if he
>doesn't know how to keep the machine cool, naturally it will overheat.
Obvious enough, but knowing is only half the story... then
there's "doing". Some even have no idea what is necessary
but just throw a bunch of fans at the problem- and do end up
with a cool enough system without knowning, albeit a
relatively loud system.
Facebook
Twitter
Instagram