Page 1 of 1

Possible to soft-undervolt Asus AM2NPV-VM? (slightly urgent)

Posted: Tue Sep 12, 2006 9:21 am
by ziphnor
Hi,

I thought i had found the right parts for my parents HTPC ( see http://forums.silentpcreview.com/viewtopic.php?t=34102 ), but as usual finding a good mobo is causing problems. I was hoping to place the order tonight as my dad is really looking forward to his new HTPC(their DVD recorder broke down so they cant watch DVDs or record anything ;), hence the slight urgency.

I was originally betting on the Asus M2NPV-VM because it has Firewire, all the necessary TV out stuff(they will need this until they buy a new TV) and an onboard SPDIF header(for later upgrade to HDMI video card).

However, now i hear that the M2NPV-VM offers no vCore control in the BIOS, which is a serious problem. I was planning to do some serious undervolting using RMClock, do i need BIOS vcore support to do that, or will it work anyway?

Re: Possible to soft-undervolt Asus AM2NPV-VM? (slightly urg

Posted: Tue Sep 12, 2006 9:53 am
by QuietOC
ziphnor wrote:However, now i hear that the M2NPV-VM offers no vCore control in the BIOS, which is a serious problem. I was planning to do some serious undervolting using RMClock, do i need BIOS vcore support to do that, or will it work anyway?
As long as that motherboard fully supports Cool'n'Quiet you can change Vcore and CPU multiplier using software like RMClock. The only BIOS support required is that it enables Cool'n'Quiet.

Posted: Tue Sep 12, 2006 10:09 am
by ziphnor
Thanks! Cool&Quiet is supported, so i guess it isnt a problem.

Thanks again.

Posted: Tue Sep 12, 2006 12:20 pm
by Rusty075
CrystalCPUID works very well on that board. I prefer that program to RMClock, I think it's easier to use and more stable, but it's personal preference either way.

By sure to read jan's article about using CrystalCPUID to configure your own custom CnQ settings...its very helpful.

Posted: Thu Sep 14, 2006 8:58 am
by Nil Einne
I can also confirm that soft undervolting works well with Crystal (haven't worked out how to get RM to work yet)

Posted: Fri Sep 15, 2006 1:10 pm
by ziphnor
Thanks for the feedback, i have no special preference for RMClock, its just that i had problems with Crystal on my own HTPC.

The important thing is that it can be done.

Posted: Mon Sep 18, 2006 4:18 pm
by autoduel
I use RMClock to under-volt my AMD64 3000+ to 0.975V (reported as 0.894v by CPU-Z).

I love the Asus M2NPV-VM board. It has been working like a dream in the HTPC.

Posted: Sun Oct 22, 2006 9:53 pm
by plympton
I've got it running somewhere around .95 - 1.0v on my Athlon 3800+ Energy Efficient Proc. Running about 80 watts total with 1 GB and a 500 GB WD HD. Turned off C'n'Q and just running flat out with Windows Media Center 2005. Not quite a Tivo, but at least it's pulling in ATSC.

-Dan

Posted: Sun Oct 22, 2006 11:29 pm
by revloc8
I have been thinking of gettin this board as an HTPC as well. I would like to get my hands on an AMD 3800+ EESFF if possible, but we'll see how it goes. I gotta say that I am extremely encouraged by the undervolting results you are all posting, but I was surprised when plympton said he is using around 80 watts(undervolted!). I read an article on Tom's Hardware that said 54W was possible with the same board, 2GB RAM, a Raptor HDD, etc.
(http://www.tomshardware.com/2006/09/25/ ... index.html)
I realize every system is different, but with the same core parts a difference of 26W is quite large, especially considering the undervolt. For a 24/7 system, 26W can add up over time, thus the knitpicking. So, my question is: Whats going on here? Do you have a video card in there? Or is Tom's talking outta their hiney as they sometimes do?

Posted: Mon Oct 23, 2006 7:00 pm
by StrongSilentType
Rusty075 wrote:CrystalCPUID works very well on that board. I prefer that program to RMClock, I think it's easier to use and more stable, but it's personal preference either way.

By sure to read jan's article about using CrystalCPUID to configure your own custom CnQ settings...its very helpful.
There's a bug in CrystalCPUID which if you have your hard drive suspended because of inactivity, then if it goes back to active mode, the voltage and the multiplier of the CPU will be set to maximum.

I email this to the author a while ago (it's always reproducible), but got no reply and no update for a few months ... at which time I found RMClock and stop checking. You may want to check if this is fix or not.

Note for the archives: Cpus don't have a 4x multiplier!

Posted: Mon Oct 23, 2006 9:45 pm
by plympton
That Tom's article is a bit annoying - no details on what voltages they were running, etc. Disappointing - no way to reproduce the results!

After getting all kinds of crashes with RMClock - I finally figured out that RMClock is trying to set a 4x multiplier on my AM2 chip - BOOM! - down goes the system every time. Anything less than 0.800 volts and - BOOM!

Looks like I can get an 5x (1 GHz) with 0.800 volts and about 61 watts, and 10x (2 GHz) with 1.000 volts and about 70 watts. It survives a reboot (nice) so far, and feels pretty stable. Not sure the 9 watts is worth the hassle of the RMClock pain I had to go through, but oh-well-it's-done.

This is with 65w 3800+ EE AM2 proc,
1 GB Memory (1 stick, 667 MHz, Kingston),
500 GB WD SATA drive,
Stock cooler,
NSK 2400 with stock PS, and 1 120mm fan on low,
Cat's Eye DTA-150 Atsc tuner card,
Running Windows Media Center 2005, Rollup-2, etc..

If it survives standby's, boots up to make recordings on time, I'll be a happy guy. My Tivo used 22 watts 100% of the time - if this uses 70 watts, but only in use some of the time, it'll be a wash power wise, but I get ATSC in the deal! :-)

-Dan

Re: Note for the archives: Cpus don't have a 4x multiplier!

Posted: Wed Oct 25, 2006 7:41 am
by plympton
plympton wrote: If it survives standby's, boots up to make recordings on time, I'll be a happy guy. My Tivo used 22 watts 100% of the time - if this uses 70 watts, but only in use some of the time, it'll be a wash power wise, but I get ATSC in the deal! :-)
Well, apparently, MCE doesn't sleep/wake too well - it's always working too hard in the background to sleep (recording Live TV). You have to install all kinds of workaround to get it to sleep, and they don't work 100% (too fine a line between recording a program and just sitting there)

Also, waking from sleep last night, I got the jagged screen of death - scrambled video out. Probably because the voltage was too low (1.00v).

Looks like I'll live with 60-70w all the time... :-/.

-Dan

Re: Note for the archives: Cpus don't have a 4x multiplier!

Posted: Fri Oct 27, 2006 10:00 am
by Felger Carbon
plympton wrote:Looks like I'll live with 60-70w all the time.
Is that the wall-socket power measured by a wattmeter (Power Angel, Kill A Watt), or some other measurement?

Re: Note for the archives: Cpus don't have a 4x multiplier!

Posted: Fri Oct 27, 2006 9:54 pm
by plympton
Felger Carbon wrote:
plympton wrote:Looks like I'll live with 60-70w all the time.
Is that the wall-socket power measured by a wattmeter (Power Angel, Kill A Watt), or some other measurement?
Kill-a-watt. Great investment for this kind of stuff!

Posted: Sat Oct 28, 2006 2:07 pm
by signal64
Not sure if it helps, but in order to get an accurate voltage reading on an A8N-VM (and the CSM version) I had to use CPU-Z and edit it's ini to use VCORE=3.

Not sure if that applies to this VM board as well.

By default, using PCPROBE II and other utils they reported higher VCORE voltages that weren't actually being deliverd.

For instance, under load 1.40v woud be listed as 1.46v. Idle 1.10v would be listed as 1.18v.

If your trying to undervolt using the default numbers you end up trying to drive the voltage down below what you think your setting them to.

I've seen some posts in regards to this on other ASUS boards and maybe worth checking out for this one.

If VCORE=3 gives bad numbers, try VCORE=2.