8rdavcore/nf7vore: Software controlled undervolting!

The forum for non-component-related silent pc discussions.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

jib
Posts: 24
Joined: Fri Jan 17, 2003 12:21 am

8rdavcore/nf7vore: Software controlled undervolting!

Post by jib » Sun Nov 30, 2003 2:25 am

Hi all..

Normally I'm just lurking on these forums as I'm too lazy to post myself, but yesterday I found this incredible utility for software-controlling my Epox 8RDA+. I'm not really sure if this has been mentioned here before, but my jaw just dropped as I realized this is the prefect solution! No more reboots! Undervolting and overclocking just became a lot easier.

There's two versions of the util, one for the Epox 8rda-series and one for the Abit NF7. Its under active development so be sure to let the author know you like it if you do. The program even exist in a Linux version :lol::lol: which I hope will be continued!

What makes it really unbeliveable is that the program has the ability to set vcore below what the BIOS announces as minimum. In other words you could go from an 1,1v undervolt to 1,650v and overclock just using this program.

Here's a screenshot of how it looks, from an earlier version

Image
Seems like the Epox owners are the ones giving the most feedback, at least at AOA. There's a similiar thread for NF7 too but not much happening there.

I've emailed the author, hasw, about this thread. Let's hope he drops in and says hi :wink:

al bundy
Posts: 667
Joined: Thu Feb 20, 2003 5:38 pm
Location: Chicago, IL

Post by al bundy » Sun Nov 30, 2003 3:14 am

This looks very promising indeed.

8)

Ralf Hutter
SPCR Reviewer
Posts: 8636
Joined: Sat Nov 23, 2002 6:33 am
Location: Sunny SoCal

Post by Ralf Hutter » Sun Nov 30, 2003 4:29 am

So have you tried it yet? Let us know how it works.

Oh, and use something like Motherboard Monitor to check your Vcore to make sure that software app actually does lower it.

GenghiS_KhaN
Posts: 210
Joined: Sat Jul 19, 2003 1:40 pm
Location: Kortrijk, Belgium

Post by GenghiS_KhaN » Sun Nov 30, 2003 5:26 am

Just installed it, seems to be OK uptill now, I was able to change the vcore, mbm recognized it, so did my BIOS when i rebooted and checked it

jib
Posts: 24
Joined: Fri Jan 17, 2003 12:21 am

Post by jib » Sun Nov 30, 2003 6:12 am

Ralf Hutter wrote:So have you tried it yet? Let us know how it works.
Sure.

Here's a pic of my current undervolted settings

.. and this is overclocked/standard settings. Temps a bit too high still, haven't installed my 120mm exhaust fan yet :wink:

Edit: pics removed

I use an applebred 1.4ghz cpu, its not as good as I was hoping for but still ok. Undervolts well here, I hope to cool it with minimal fan speeds.

As the prog reads from the same sensors as speedfan (uses the same driver) and the others I see no need to verify that what it says is correct.
Last edited by jib on Wed Dec 03, 2003 4:51 am, edited 1 time in total.

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Sun Nov 30, 2003 8:47 am

I'm running it on my NF7 right now.

Works excellent. It mis-identifies the multiplier (which he warns is a possibility in the Readme, and is really of no consequence anyway) It also has trouble with the SYS sensor, at least I hope it does, or my mobo is running at 208°.

The fan reading doesn't like my FPS fan, it shows it at double what it should. (I had to adjust the divisor in MBM5 to get it to read right there, nf7core is probably locked at the defualt divisor reading)

But those unimportnat sensor readings aside, it rocks!

The AutoFSB feature works smoothly, and is something alot of people have been looking for for a long time.

On my machine I have it set down to 1.4v, and at idle the FSB drops down to 66mhz. Giving me a heat output of 21.7 Watts. (A deduction of 65% from the stock of 62.1)

GenghiS_KhaN
Posts: 210
Joined: Sat Jul 19, 2003 1:40 pm
Location: Kortrijk, Belgium

Post by GenghiS_KhaN » Sun Nov 30, 2003 8:52 am

hm, I also have the 208°c problem but this is not really bad but now it keeps saying 'Apply new settings...' when I hit apply...
Doesnt do a thing

It worked once but now... :cry:

It rocked when it worked :)

edit: it seems that I can only apply some settings, I couldnt enter 1.475V but i can enter 1.45V.
if it doesnt want to apply the settings it'll just keep on saying: "apply new settings..." having some problems with that right now

Edit2: :oops: I did something wrong, for some reason I already applied the settings and tried to apply them again... thats when I had the 'problem'
Last edited by GenghiS_KhaN on Sun Nov 30, 2003 10:04 am, edited 2 times in total.

wsc
Posts: 161
Joined: Sat Jun 21, 2003 9:25 am
Location: Oklahoma

Post by wsc » Sun Nov 30, 2003 9:19 am

I can't wait to finish my NF7-s system and try it out! :D

Inexplicable
Posts: 226
Joined: Sat Sep 06, 2003 5:59 am
Location: Finland

Post by Inexplicable » Sun Nov 30, 2003 10:12 am

This is kind of inspiring. Too bad I don't have either of these mobos. :( However, I know that my Abit IS7-E has a Winbond W83303D regulator chip on it, which is supposed to have an I2C interface. Does anyone happen to have programming info?

Poking around a little I found that the I2C devices at addresses 0x2e and 0x44 on the SMBUS have registers that seem to change consistently with my BIOS settings. I think I've found the right places to change VCORE, DDIMM, and AGP voltage. Unfortunately I can't figure out how to undervolt. My bios has exactly 16 different VCORE settings, ranging from VID to VID+0.375V. Looks like VCORE is controlled by a 4-bit register, so undervolting the IS7-E might actually be impossible. :(

No luck finding the FSB clock generator yet. Anybody have any info?

GenghiS_KhaN
Posts: 210
Joined: Sat Jul 19, 2003 1:40 pm
Location: Kortrijk, Belgium

Post by GenghiS_KhaN » Sun Nov 30, 2003 1:38 pm

I contacted the guy about the high temp with the nf7s motherboard:
"Hi,

yes, Systemp will be corrected:

0.7.6b:
- Added VCore-presets to AutoFSB
- Fixed sys-temp display (nf7vcore)
- Added second AutoFSB delay option
- Moves now to tray if minimized
- Corrected Vdd trackbar setting at startup (8rdavcore)

Regards,
Sebastian
"
Last edited by GenghiS_KhaN on Mon Dec 01, 2003 1:08 pm, edited 1 time in total.

mikmac
Posts: 2
Joined: Sun Nov 30, 2003 3:05 pm

Post by mikmac » Sun Nov 30, 2003 3:20 pm

is there any possibility for program like this one for Asus A7N8X ??
Im just looking program like that from a month or so...

Hasw
Posts: 7
Joined: Sun Nov 30, 2003 3:20 am
Contact:

Post by Hasw » Tue Dec 02, 2003 6:21 am

Version 0.7.6b is now available:

http://www.hasw.net/8rdavcore/

Changes:

0.7.6b:
- Added VCore-presets to AutoFSB
- Fixed sys-temp display (nf7vcore)
- Added second AutoFSB delay option
- Moves now to tray if minimized
- Corrected Vdd trackbar setting at startup (8rdavcore)
- Increased VCore change delay
- Changed CPU load code
- Added "reset settings before exit" option
- Fixed "Current Vagp" displays Vdd
- Added max. CPU temp to AutoFSB
- Added some hints

Inexplicable
Posts: 226
Joined: Sat Sep 06, 2003 5:59 am
Location: Finland

Post by Inexplicable » Tue Dec 02, 2003 10:44 am

Hasw -- a couple of questions for you...

Do you still make the source code available? I found an old version on freshmeat but nothing since. I'm interested in writing a similar Linux utility for the Abit IS7 series. Or rather, I'm interested in making a daemon that adjusts vcore and FSB automatically based on load.

How did you come up with the programming info? Disassembling the bios and monitoring the smbus come to mind.

Hasw
Posts: 7
Joined: Sun Nov 30, 2003 3:20 am
Contact:

Post by Hasw » Tue Dec 02, 2003 11:38 am

I'm working parallel on a new Linux-version. The methods used in version 0.5.5 haven't much changed. Try to find out how the voltages are controlled on your board (looking for ICs on the mainboard) and try to get datasheets for them. If it's not possible to get them, "monitor" the devices for changes.

Ralf Hutter
SPCR Reviewer
Posts: 8636
Joined: Sat Nov 23, 2002 6:33 am
Location: Sunny SoCal

Post by Ralf Hutter » Tue Dec 02, 2003 12:08 pm

Hasw wrote:I'm working parallel on a new Linux-version. The methods used in version 0.5.5 haven't much changed. Try to find out how the voltages are controlled on your board (looking for ICs on the mainboard) and try to get datasheets for them. If it's not possible to get them, "monitor" the devices for changes.
Any plans to port this to P4 boards?

Inexplicable
Posts: 226
Joined: Sat Sep 06, 2003 5:59 am
Location: Finland

Post by Inexplicable » Tue Dec 02, 2003 12:34 pm

Hasw -- Yeah, I've been eyeballing the mobo, all right. Vcore on Abit IS7 is regulated using Intercil's ISL6556B. I've got the datasheet but the problem is the chip doesn't have an i2c interface. Presumably there's an additional chip that implements it but so far I've been unable to locate it on the PCB. I think I've found the device on the smbus, though. However, the registers seem write protected and I have no idea how to enable them.

I was wondering if the NF7 has a similar interface and, if so, how do you unprotect the registers?

Unfortunately I don't have the necessary equipment to monitor the smbus. I suppose I could rig something with a second PC... Any suggestions for existing tools?

Hasw
Posts: 7
Joined: Sun Nov 30, 2003 3:20 am
Contact:

Post by Hasw » Tue Dec 02, 2003 1:16 pm

@Ralf Hutter: I don't have a P4, so it will be difficult adding support for one without information and a possibility to test it.

@Inexplicable: The ATXP1 registers are not write-protected on the NF7. Maybe you have found the SPD EEPROMS (0x50-0x54)?
I've used a Atmel microcontroller which sends the I2C data to a RS232 interface.

Inexplicable
Posts: 226
Joined: Sat Sep 06, 2003 5:59 am
Location: Finland

Post by Inexplicable » Tue Dec 02, 2003 1:26 pm

I see the SPDs as well but I've got the following at device address 0x44. Byte #3 changes when I change the vcore in BIOS. Based on the few safe voltages I can use, the register seems to correspond to the 6-bit input signal accepted by the Intersil chip, assuming a constant 0.375V offset to the output voltage.

0 1 2 3 4 5 6 7 8 9 a b c d e f 0123456789abcdef
00: 00 00 0f 38 00 00 00 00 00 00 00 00 00 00 00 00 ..?8............

Ralf Hutter
SPCR Reviewer
Posts: 8636
Joined: Sat Nov 23, 2002 6:33 am
Location: Sunny SoCal

Post by Ralf Hutter » Tue Dec 02, 2003 1:28 pm

Hasw wrote:@Ralf Hutter: I don't have a P4, so it will be difficult adding support for one without information and a possibility to test it.
What kind of board do you need to do this with? Will any P4 board work, even on that's not already undervoltable in the BIOS, or can you make any board work, no matter what? I have a spare Intel 875PBZ board that I might be willing to loan you if you think it might work. Where are you located?

Hasw
Posts: 7
Joined: Sun Nov 30, 2003 3:20 am
Contact:

Post by Hasw » Tue Dec 02, 2003 2:02 pm

@Inexplicable: Have you searched for this device on the board? It can be far away from the voltage controller.

@Ralf Hutter: No, only boards that are able to change the VCore over software. If the VCore-ID is directly connected to the voltage controller it is not possible. And different manufactures using different methods to control the VCore, so the software only runs on this type of board. Thanks for your offer but I have also no P4 CPU.

Inexplicable
Posts: 226
Joined: Sat Sep 06, 2003 5:59 am
Location: Finland

Post by Inexplicable » Tue Dec 02, 2003 3:14 pm

Hasw wrote:@Inexplicable: Have you searched for this device on the board? It can be far away from the voltage controller.
I've been trying but I guess I'd have to take the mobo out to get a really good look. This is my main rig, soI'been avoiding taking it apart.

GenghiS_KhaN
Posts: 210
Joined: Sat Jul 19, 2003 1:40 pm
Location: Kortrijk, Belgium

Post by GenghiS_KhaN » Wed Dec 03, 2003 4:32 am

hm I think to new version messes with my MBM5 temps.

They are really weird -55 cpu -55 case and it only occured when I had the program running in the background for a while. Now I'll see if it does the same with the program NOT running.

jib
Posts: 24
Joined: Fri Jan 17, 2003 12:21 am

Usage guide

Post by jib » Wed Dec 03, 2003 11:03 pm

Ok as you may have noticed I made this small usage guide, it expanded from being just my own notes to becoming almost useful.

I was going to post it right in this thread but somehow I managed to push the new topic button instead of reply :roll:

Hasw
Posts: 7
Joined: Sun Nov 30, 2003 3:20 am
Contact:

Post by Hasw » Sat Dec 06, 2003 9:25 am

New version 0.7.8b:

http://www.hasw.net/8rdavcore/
http://www.hasw.net/nf7vcore/

0.7.8b:
- Now shows Fans > 670 RPM
- Added sensor update frequency
- Added FSB change delay for AutoFSB
- Now it's possible to change VCore and FSB in one step
- Save settings on exit
- Added -A command-line option for AutoFSB
- Added more debug-messages to giveio service handler
- Changed CPU load code

egghat
Posts: 73
Joined: Fri Mar 07, 2003 5:38 am

CPU temperature 208 degrees celsius ?!?

Post by egghat » Mon Dec 08, 2003 5:13 am

I have an Epox 8RDA3+. For some strange reasons the CPU temperature is 208 degrees. Therefore AutoFSB doesn't work correctly, cause the CPU temperature is too high and AutoFSB refuses to increase the FSB even when the load is high.

Anyone knows a solution?

Bye egghat

Hasw
Posts: 7
Joined: Sun Nov 30, 2003 3:20 am
Contact:

Post by Hasw » Mon Dec 08, 2003 9:14 am

Have you a negative CPU temperature?

egghat
Posts: 73
Joined: Fri Mar 07, 2003 5:38 am

Post by egghat » Mon Dec 08, 2003 10:09 am

Hasw wrote:Have you a negative CPU temperature?
No, it's plus 208 degrees :?

CPUCool reports -34 degrees. MBM also when I select Winbond diode, but temperature is (supposedly) correct, when I choose Winbond 1,2 or 3.

Now, OK, this is really weird. After I started MBM suddenly I get the correct temperature in your wonderful little tool too.
It's now 35 degrees like sensor 2 in MBM. Normally I never start MBM (perhaps I should ;-) ).

Btw., 8rdavcore shows the second temperature fine.

Weird. I'll reboot my computer tomorrow to check what 8rdavcore reports, after the computer is freshly booted.

bye egghat.

Hasw
Posts: 7
Joined: Sun Nov 30, 2003 3:20 am
Contact:

Post by Hasw » Tue Dec 09, 2003 10:15 am

I have meant your real temperature ;-) If I see this now right, your normal CPU temperatur is 35 degrees.

silvervarg
Posts: 1283
Joined: Wed Sep 03, 2003 1:35 am
Location: Sweden, Linkoping

Post by silvervarg » Wed Dec 31, 2003 4:41 am

I have tried this program out on my NF7-M board, and it works great.
Is it possible to let the program change the multiplier setting as well?

For some reason I get less stable as I lower the FSB to 100 MHz (lowest setting). It is rock stable at 110 MHz.
Works great at 1.1Vcore 11*110 MHz on my Barton 2500+ (going aprox XP1700+) dissipating a mere 18.2W.
Could be my RAM that don't like that slow speed.
Also I can't reach 200 MHz FSB with stability. ~195MHz seems to be the limit. Probably RAM that is stopping me here.

Also rock stable at:
1.2Vcore 11*133 MHz (~XP2000+). 28.8W max.
1.425Vcore 11*166 MHz (XP2500+ stock speed). 50.7W max.

Stock CPU settings: 1.65Vcore, 11*166. 68.3W max.

Stability tested with prime95 on all settings.

jib
Posts: 24
Joined: Fri Jan 17, 2003 12:21 am

Post by jib » Thu Jan 08, 2004 1:48 am

silvervarg wrote:I have tried this program out on my NF7-M board, and it works great.
Is it possible to let the program change the multiplier setting as well?
In the words of hasw ..
This multiplier registers are outputs...so they are used. But
changing this on runtime it does not change something. I think the
multi is only read by the chipset on a reset. Maybe there is a way,
but I think the chances are good that the computer crashes because
of the "big" speed difference in a short time.
Other news worth mentioning, the 8rdavcore/nf7vcore utils has now been merged into one program, and a new version has been out for a while. This version fixes some bugs and adds fan speeds controlling for the NF7 (yay!!) .. this possibly does not work on all/any of the 8rda series due to a different hardware design.

The development and discussion about this util still goes on at the Epox AOA Forum so head over there if you are interested.

Post Reply