Underclocking..

What are you underclocking? How much? Most high-end microprocessors use some sort of dynamic logic so there is a lower limit to the clock speed. Purely static processors (most UCs) can be run down to DC, so there isn't any problem there. Depending on the I/O, there may be real limitations there, too, particularly if timing loops are used or the clock is used for other things.

The only other problem I can see is that, they won't heat the room. ;-)

If you underclock by too much there is a risk that any dynamic registers will forget their contents. Most modern CPUs can tweak their clock rate according to demand to some extent. Decreaseing the clock rate and CPU voltage equates to less power consumption so greener and longer battery life. It is common for battery powered kit to idle whenever it can.

Intel called it speedstep technology when they introduced it into its dedicated Laptop M series CPUs almost a decade ago. See for example:

Later generations added "turboboost" which does the opposite. It seems that Intel has started encouraging serious overclocking these days. A nominal i5-2500K 3.3GHz will comfortably overclock to 4.3GHz with a better than THE rubbish stock Intel cooling fan heatsink attached to it.

Regards, Martin Brown

If the question is simply, does it make sense to run a total amount of work (CPU cycles) at full speed at say 1/10 duty time or run something at 10 % speed at 100 % duty cycle, the end result is the same.

The only real advantage of the slower clock frequency is that you can run at a lower Vdd and hence save a lot of power, since in a typical situation, the power consumption is proportional to the square of Vdd.

2500K... is that the temperature the CPU will run at when overclocked?

Kidding. =3D)

Michael

modern fast processes can be rather leaky so the static power consumption can be rather high regardless of clock rate, so sometimes it migh make sense to run as fast as possible for a short time and then turn off as much as you can

-Lasse

If you're cutting the clock rate it's also possible to lower the voltage, thus the static/leakage power.

Specifically, a Pentium 2-266 clocked down to 180.

Why would you bother? They are a waste of space and better off clocked at 0Hz.

Regards, Martin Brown

A P-II will have dynamic logic so you can't go all the way to DC but I'd bet

180MHz would work, as long as the multipliers are there. Intel did quite a bit of clock-locking back then.

Indeed. There are far better low-power solutions today. The P-II was a pig.

I am not looking for low power; i need a computer with at least one ISA slot, and it would be nice to have the hardware speaker driver available. That last part is a dig to the idiots that designed the Compaq Presario 5185 and similar (NO speaker circuitry).

• The only reason for the underclocking is that the slot one CPU got hot as all heck without a fan at its rated 200MHz clock. So now a fan and clock at 180MHz which the motherboard allows.

Let me guess. You have an old ISA National Instruments GPIB card and you don't want to pay for a modern USB or ethernet version.

ssors

ny

too,

=A0;-)

d bet

uite a

was a pig.

e

There were cpu-fanless P2s too, they just had a much larger heatsink on the cpu. ISA slots were moderately common on P3s too, with just 1 or sometimes 2 such slots on the mobo. And a P3 is surely a better proposition, if you only need one ISA slot.

NT

Sort-of; it is a combo A/D and DIO board that i programmed in BASIC, compiled, and run in the most robust lowest overhead OS: DOS.

Yes, this CPU had the large heatsink - but got hot a i indicated. Check; only need one ISA slot, but this MB seemed less expensive in that it needed less "overhead": could use my old AT supply for one and so did not need major case surgery for the power switch (AT supply used a real switch and the ATX uses a fake power logic button).