RMClock guide for T60?

[mattbiernat]

did anyone get RMClock working for their T60? if so let me know which paremeters did you change and to what values. are the temperatures considerably lower?
edit blah this should be probably posted in the utility work area. sorry.

[brentpresley]

did anyone get RMClock working for their T60? if so let me know which paremeters did you change and to what values. are the temperatures considerably lower?
edit blah this should be probably posted in the utility work area. sorry.

Not possible to the best of my knowledge (same result w/ Clockgen as well). IBM/Lenovo has always used VERY restrictive clock generators on their machines.


I spent the better part of a week on this, and the conclusion I came to is that just stock frequencies (667FSB, 533FSB, etc.) are programmed onto the clock generator chip. Nothing else.

[aaa]

What does the clock generator have to do with voltage?


Anyways, the voltages on the Core chips are kinda restricted. The idle temp probably won't change at all when you do it, however I think it would affect the load temp.

And look into what you can do for the ATi chip as well, assuming it isn't already on the lowest setting.

[mattbiernat]

too bad more people didn't try it. according to this guy it can be done

effectively

http://forum.thinkpads.com/viewtopic.ph ... ht=rmclock
it's just i don't know how to...

[brentpresley]

If all you want to do is LOWER the voltages, then NHC works well (except under Vista).

I was under the impression you were trying to overclock the chip, not undervolt it (not clear to me from the first post).

[mattbiernat]

If all you want to do is LOWER the voltages, then NHC works well (except under Vista).

I was under the impression you were trying to overclock the chip, not undervolt it (not clear to me from the first post).

last time with NHC the voltages were grayed out and i gave up...

[jeepcoma]

I've been using RMClock with my T60p T7200 since I got it and it works great. Currently using the latest 2.30.1 beta release. Lowering the voltage has a significant impact on heat put out.

As far as the ATI chip is concerned, the only thing you can use (that I know of) is PowerPlay. ATI tray tools for example lets you adjust mem and core speeds, but doesn't allow as drastic throttling as PP does, and also ATT doesn't underclock. If only there was an RMClock utility for the GPU, that would be perfect (performance on demand for GPUs makes a lot of sense). One nice feature of PP however is that you can set up profiles which activate with keystrokes, and since RMClock can be used with command line instructions, I've set it up so that ctrl-shift-1 sets PP to maximum battery and RMClock to my performance on demand profile, and ctrl-shift-2 sets PP to maximum performance and RMC to my max performance profile as well (tpfancontrol taking commands like this would be excellent, so I could set it to my auto settings and fixed at level 7 with the same keystrokes). I'm pretty happy with this configuration, as I can quickly set everything to max performance before playing a game or doing anything intensive, then back to power saving modes with a simple keystroke.

Since a picture is a thousand words, here are some links to screen shots of my configuration.

Mod Edit No inline pics allowed without a warning in the thread subject.
Harryc

http://i120.photobucket.com/albums/o175 ... gement.jpg
http://i120.photobucket.com/albums/o175 ... ofiles.jpg
http://i120.photobucket.com/albums/o175 ... ma/max.jpg
http://i120.photobucket.com/albums/o175 ... ma/pod.jpg
http://i120.photobucket.com/albums/o175 ... advcpu.jpg

And for fun I was playing around with some stress testing. You can see that with the fan forced off, with the voltage at .975 after 135 seconds the temperature has reached only 71*C. It eventually stabilized at about 95*C after ~600 seconds, remember this is with no fan, and none of the thermal protection features were activated by the CPU. I then forced it to clock down to 1 ghz and .950V and you can see the immediate drop in temp to hold stead at 85*C.

http://i120.photobucket.com/albums/o175 ... nofan1.jpg
http://i120.photobucket.com/albums/o175 ... nofan2.jpg

Here's the same test at stock voltage.

http://i120.photobucket.com/albums/o175 ... oltage.jpg

[eyestrain]

jeepcoma,

Have you compared your "performance on demand" vs. "maximum performance" when just idling?

I found that when idling (no programs running) speedstep seemed to offer very little benefit when I undervolted CPU down to 1.0125v (from stock 1.2v). Disabling speeds 6-11 caused only a tiny temperature increase.

Of course at load, running at speed 12 will make much more heat than at speed 6, but when you are at load, "performance on demand" really uses speed 12 anyway.

Instead of having separate CPU settings for quiet and performance, I just forced CPU to stay at speed 12 and undervolted. Maybe this is a bad choice, but my simple tests seemed to show it's ok, even at 1.0125.

I used NHC, since I noticed no real benefit from RMClock's throttling.

[mattbiernat]

wow thanks a lot jeepcome! finally somebody posted a guide in undervolting. i will be checking this out soon!

[jeepcoma]

A word about undervolting, every processor is different and may or may not undervolt to the same level. Mine works 100% at .975 at full power but that's as low as I can go for the 12x multiplier. It works with .950 for smaller multipliers, but yours might be different. I've read of people who can use .950 all the time with no worries at all clock speeds. The only way to tell is to do some testing on your machine, I use a multithreaded tool to stress both cores to 100%. Check out WPrime

The results are worth playing around with settings for a little while, there is quite a bit of heat reduction to be found.

[jeepcoma]

eyestrain, you're right on. There doesn't seem to be much, if any, difference in temps with the different clock speeds when the cpu is at idle. I'm guessing that it would be due to the advanced halt states working, but I don't really know. I've basically reached the same conclusion as you that I should just leave it at full speed all the time. The drop in voltage has far more affect on temperature than clock speed.

[eyestrain]

Thanks for the confirmation.

In my case, idling at .95V 1ghz seemed to settle at about the same temp as idling forced at 1.0125V 2ghz. Oddly, there was a much larger (not proportional) temp difference when I forced idle to 1.2V.

Under load does seem where the biggest benefit of undervolting comes. That's why I'm jealous of .975 or .95, since my CPU made errors (eventually) even at 1.0V.