P5B Deluxe & P5W DH overclocking, save 20W+

All about them.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

P5B Deluxe & P5W DH overclocking, save 20W+

Post by smilingcrow » Wed Sep 13, 2006 4:11 pm

I’ve already posted here about how the newer BIOSs for the P5B Deluxe affect voltage when overclocking. But after measuring how significant the actual power consumption that can be saved is, I figured it is worth a separate thread to highlight this. Here’s the situation with regard the P5B Deluxe using BIOS revision 0614:

If you overclock in the BIOS and leave the VCore setting in the BIOS on auto, the VCore gets increased above stock and becomes fixed. This means that Speedstep will have no affect on VCore and neither will RMClock. You may well not realise this is happening as CPU-Z continues to show the VCore value as being variable; it changes with Speedstep or RMClock.
Fortunately there is a way to overcome this problem; overclock the FSB with
Clockgen 1.0.4.8 Beta. This allows you to regain control over the VCore using Speedstep or RMClock.

Using Clockgen I overclocked an E6600 to 3GHz using a FSB of 333 and with RMClock I set the VCore to the following settings: Multiplier 6 = 1.05V, Multiplier 9 = 1.212V. This resulted in CPU-Z displaying voltages of 1.128V & 1.288V.
I then measured the power consumption at idle and when running two instances of Prime95. I did this for the two different methods of overclocking and these were the results.

Idle/Load:
BIOS O/C = 108/153W
Clockgen O/C = 85/134W

That’s a saving of 23W at idle and 19W at full load.

I mention the Asus P5W DH Deluxe in the thread title as I have a strong suspicion that recent BIOS revisions have made it function in the way that the P5B Deluxe does.
Could someone that has this board please test to see if this is the case. If you don’t have a power meter there are other ways to test for this functionality.

Notes.
1. I found Clockgen to be stable although I didn’t test it much above 400 MHz. It happily went down to 166 MHz and I didn’t try setting it lower than that.
2. The clock generator that the P5B Deluxe uses is the Cypress CY28551; you configure Clockgen to use it. I don’t know if the P5 (non deluxe) uses the same clock generator.
3. The system at the 2GHz/3GHz setting was tested as being Prime95 stable for 8 and 3 hours respectively.

NeoNSX
Posts: 80
Joined: Sun Sep 10, 2006 5:04 am

Post by NeoNSX » Fri Oct 06, 2006 5:39 pm

This is a great tip; I'm surprised nobody has responded until now. Thanx for posting.

maukka
Posts: 12
Joined: Thu Dec 09, 2004 5:11 am
Location: Finland

Post by maukka » Sat Oct 07, 2006 2:23 am

I don't have a power meter, but judging from the temperatures reported by coretemp, this behaviour does not occur with P5W DH (BIOS 1407).

I'm using an E6600 and FSB set to 333 in BIOS. Vcore is set to AUTO because manual selection is not available when C1E and Speedstep are enabled.

Changing VID from rmclock definitely affects the temps. I can only select a minimum VID of 1.162V, but at 6x333MHz the cpu temperature drops a good 7C compared to 1.325V when stressing with Intel TAT.

Temps also drop 4C at 9x333 when changing VID from 1.325V to 1.287V (orthos stable).

So I would hazard a guess that with P5W DH power consumption does decrease when rmclock is used to lower the VID and BIOS is used for overclocking

onix
Posts: 3
Joined: Sun Oct 08, 2006 4:39 am

Post by onix » Sun Oct 08, 2006 4:47 am

I'm using CrystalCPUID instead of RMClock (it has three states :wink: ) but even thought the multiplier ratio has been changed to 7 in the bios (0711) and it shows as startup multiplier under RMClock, I sometimes get a 6x multiplier both in RMClock and CrystalCPUID's real time clock :shock:
I've set the three states to a 7x multiplier with different voltages, where does that 6x multiplier get from?
PS: E6300 + P5B Deluxe

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Sun Oct 08, 2006 5:02 am

maukka wrote:So I would hazard a guess that with P5W DH power consumption does decrease when rmclock is used to lower the VID and BIOS is used for overclocking
Sounds very promising.
onix wrote:I'm using CrystalCPUID instead of RMClock (it has three states :wink: )
CrystalCPUID has 3 states because Cool n Quiet typically only used up to 3 states I think! RMClock allows you to use every multiplier with a Core (2) Duo; my E6600 has 4 and my T2300 has 5. Your E6300 only has 2 multiplier settings: 6 & 7.
onix wrote:I'm using CrystalCPUID instead of RMClock (it has three states :wink: ) but even thought the multiplier ratio has been changed to 7 in the bios (0711) and it shows as startup multiplier under RMClock, I sometimes get a 6x multiplier both in RMClock and CrystalCPUID's real time clock :shock:
I've set the three states to a 7x multiplier with different voltages, where does that 6x multiplier get from?PS: E6300
6 is the multiplier that Speedstep will use at idle. I’m not sure that CrystalCPUID is so great with Core (2) Duo, RMClock has more features and finer control.

onix
Posts: 3
Joined: Sun Oct 08, 2006 4:39 am

Post by onix » Mon Oct 09, 2006 2:50 am

Thanks for your suggestion, but I still prefer the three states CrytalCPUID provides. I don't need multiplier managment, really, more like "voltage managment" :wink:
After disabling CE1 and Speedstep completely, the 6x multiplier dissapeared :)
CrytalCPUID continues working correctly even with Speedstep disabled in the BIOS (I thought you had to leave it there and disable it in windows)

smilingcrow
*Lifetime Patron*
Posts: 1809
Joined: Sat Apr 24, 2004 1:45 am
Location: At Home

Post by smilingcrow » Mon Oct 09, 2006 3:49 am

onix wrote:Thanks for your suggestion, but I still prefer the three states CrytalCPUID provides. I don't need multiplier managment, really, more like "voltage managment" :wink:
After disabling CE1 and Speedstep completely, the 6x multiplier dissapeared :)
CrytalCPUID continues working correctly even with Speedstep disabled in the BIOS (I thought you had to leave it there and disable it in windows)
I’m not clear what you are trying to achieve here, but I do get the impression that you don’t fully understand the process involved.
Your E6300 only has two multipliers values, hence two speeds; therefore it only needs two voltage values. If you disable C1E & Speedstep then your CPU will be fixed at 1.86 GHz and it will only require one voltage value.

CrystalCPUID and RMClock allow you to set an individual VCore value for individual multiplier values. Just because CrystalCPUID shows 3 settings, doesn’t mean that your CPU has 3 settings. CrystalCPUID shows 3 settings as a hangover from when it worked with AMD CPUs only.
For this reason and others, I think RMClock is much more useful when working with Core (2) Duo CPUs. It will show you the individual multipliers that your CPU supports and allow you to select which of them you wish to use and to set their VCore values. It seems much clearer to me in showing how the process works. Whereas CrystalCPUID can be confusing, which is why I think you haven’t fully grasped the concept at hand. CrystalCPUID has misled you. :(

onix
Posts: 3
Joined: Sun Oct 08, 2006 4:39 am

Post by onix » Tue Oct 10, 2006 7:58 am

What am I trying to achieve?
To give the processor the voltage it needs, acording to ussage. Maybe I am misled, somehow, because I don't see temp going down when the 1.05 setting is in use :(

Peroni
Posts: 26
Joined: Sun Jun 25, 2006 12:52 am

Post by Peroni » Sat Jan 13, 2007 5:01 pm

Sorry to bring up an old post but I was wondering if someone figured out a way to dinamically change the CPU speed and voltage based on CPU load with the P5B Deluxe.

I'm surprised nodoby has created an application that can change multiplier, voltage and FSB speed all at the same time and automatically.
For instance my E6300 could run at 3.2 GHz with 1.4v while playing games and something like 1GHz @ 1v for when I'm only surfing the net.

Post Reply