[ltp] Celeron M: sentenced to fan for life?

Laurent Gilson linux-thinkpad@linux-thinkpad.org
Mon, 13 Feb 2006 00:40:07 +0100


Hello

> < there are people who believe that frequency modulation has no
> < measurable effects on Celeron M's. Just try it and you will find out.
>
> $ cpufreq-info
>   driver: p4-clockmod
>   hardware limits: 175 MHz - 1.40 GHz

Yes, but does it achive the 175 MHz with frequency modulation (switching  
on and off very quickly, also know as throttleing) or by really running at  
175MHz ? I think it's using freq-mod, as 175MHz is way lower than any  
pentiumM can go (600MHz is a hard limit).

The next problem is the voltage used. Does your celeronM reduce the  
voltage while running at 175MHz or does it keep it's normal voltage ? The  
papers from intel sugest it is not reducing its voltage (seems logical  
because freq-mod and voltage changes don't go along very well).

The pentiumM is running so cool because most of the time it's at 600MHz  
and 0.9xxV (< 2.9W in deep sleep mode). If it were running 600MHz and the  
standard-voltage it's fan would spin too (<11.5W in deep sleep).

> (P.S., saw "p4-clockmod: Warning: Pentium M detected..."!)

That's bad. The celeronM-interface is different from the pentiumM, as the  
celeronM is lacking some features.

I guess you can get a better fan off/on ratio by playing with the voltage  
and freq-mod. But after reading up on the celeronM i don't think you can  
stop the fan. Summary based on the intel papers:

- The celeronM is using <16W in deep sleep, <28W in working mode.
- The pentiumM is using <2.9W in deep sleep, <31W in working mode.

You have about 14W too much heat.

cu