After extensive beta testing CpuIdle available as Pro version for WinNT and
Win2000. CpuIdle Pro supports both single and multi-processor systems.
Although Microsoft claims that these systems take care of power management
themselves, measurements show that CpuIdle Pro can still achive a considerable
decrease of temperature and power consumption.
Even under Win2000 a Celeron 500 will run at 37°C with CpuIdle Pro instead of
56°C without.
What wonderful spam.. FYI Win2000 already uses the cpu halt command and on
my 450 it's never gotten over 32 degrees. In other words, cpuIdle is
completely useless on a NT based machine.
On some boards, especially BP6, Win2k does not support ACPI yeat (except for
the QQ bios), and without it, it does not implement the HLT instructions.
For me, CpuIdle made a 7 degree difference, from 32 to 25, this is two
C366@550
so, don't diss it till u try it
AFAIK the HLT instruction executed by a process running on a CPU has
nothing to do with the BIOS.
AFAIK the HLT instruction executed by a process running on a CPU has
nothing to do with ACPI.
Are you saying that Win2k does not include the HLT instruction natively as
it does in NT4?
What type of thermometer did you use to measure the temperature drop in the
system you are referring to?
Wayne Monteath
Twentyfingers <twenty...@mail.ru> wrote in article
<7r5o4.12357$W4.6...@news1.rdc1.on.wave.home.com>...
Wayne Monteath schrieb:
> AFAIK the HLT instruction executed by a process running on a CPU has
> nothing to do with ACPI.
>
Wrong. Does not work without ACPI in SMP-Systems with MS-HLT-Routine.
>
> Are you saying that Win2k does not include the HLT instruction natively as
> it does in NT4?
>
No, it DOES include the HLT natively, but does not work in Multiproc-Systems
without ACPI (Why? Ask MS!). (NT4.0 does not execut HLT-command in
Multiproc-sys either)
On a BP6 with with QQ-beta-Bios it semms that with ACPI enabled CPU-Idle is no
needed any more although it's HLT-routine seems a bit more effective (gives an
aditional drop of 2 to 3 deg C).
Jörg
>> No, it DOES include the HLT natively, but does not work in
Multiproc-Systems
>> without ACPI (Why? Ask MS!). (NT4.0 does not execut HLT-command in
>> Multiproc-sys either)
>> On a BP6 with with QQ-beta-Bios it semms that with ACPI enabled CPU-Idle
is no
>> needed any more although it's HLT-routine seems a bit more effective
(gives an
>> aditional drop of 2 to 3 deg C).
>> Jörg
It's probably more effective cause you processor spends more time in
halt... - sounds like an un-overclocker.
--Bob
Bob schrieb:
> It's probably more effective cause you processor spends more time in
> halt... - sounds like an un-overclocker.
>
> --Bob
Nop! The usual 2x366@550
CPU-Idle aparently has a more "aggressive" HLT-routine using even very short
idle periodes to execute HLT (causing minor performance drops). Can't verify
cause I don't get the source-code from MS ;-)
Jörg