Gigabyte EP35-DS3
Corsair 520 watt PSU
Latest bios (F3)
Intel E8400 Stepping 0, Revision E0

I am only able to undervolt it at full speed to 1.1 volt. Also, having activated SpeedStep, using RM Clock, I am only able to combine FIDS with a minimum value of 1.1 volt. (In bios I can use much lower)

Is it possible to unlock these VID settings to enable lower values?

In Vista, I found no clear guide on how to enable SpeedStep, any suggested readings?

Is there anything else that I can do to improvde the undervolting stability?

Thanks,
Erik

This is all I get with EPU fom P5Q-E.

My Q9550 only undervolts to 1.125V at 100% - anything less than that is prone to crashing. With SpeedStep it would idle at 1.08V - seeing as it's always running a folding VM with 2 cores in the background, I use less power than leaving voltage at "auto". It hosts a website so is on 24/7.

It seems some people have undervolted successfully when they have also underclocked. Have you considered lowering the cpu multiplier as well?

Thanks all for your feedback. Concerning the lowered multi I havn´t tried it yet. WIll give it a goo tomorrow.

What is the speed inpact of lowering the FSB?

It seems I´ve come to a dead end. The lowest available VIM for E8400 in RMClock is 1.1 volt.

Using BIOS I am able to access lower values, but since I´m in need of dynamic adjustment, being able to change it in Windows is crucial.

Is there some way to overcome the minimum SpeedStep available VIM?

if you can turn those 0's into 1's there is your only way, but i dont know how exactly to do that, probably if VID 5 is isolated from the socket, not sure what to do

I finally got a MoBo that will undervolt (Gigabyte EP45-UD3P) and I think my results with the E8400 are pretty decent.

3.6GHz: 1.232V idle / 1.200V stress testing

3.0GHz: 1.056V idle / 1.040V stress testing

2.0GHz: 0.880V idle / 0.864V stress testing

I think that the silicon would be fine at 2.0GHz even lower than that, but those voltages seem to be the lower limit of what the system can handle. Any lower and CPU-Z won't read, a few notches still lower and it won't boot.

Those are all stable BTW. 10k FFTs in prime for an hour, then Orthos blend overnight.

Same here, I have EIST and C1E/2E/4E enabled. Dont use RM clock I saw the same 1.1v bottom limit. I found on my Gigabyte board that it wont go down to lower than 1.1v unless I have the Vcore on auto or normal.

So on my UD3P here's my new Q9650, that runs almost the same clocks at the same voltages as my E8400 E0.

@gechu
Speedstep P-states are on-chip and can't be changed by BIOS or windows apps. If you manually undervolt your chip in BIOS, it will remain at that VID through all "speedsteps" and Speedstep would only change the multiplier. If your chip can undervolt to a level near or below the P-state minimum of 1.1v and remain stable at load, then go that route and let Speedstep set the multiplier dynamically. I have a Q9450 set that way (.99375 VID, 0.960 Vcore idle, 0.944 Vcore load)and it's stable at all multipliers and loads. If your chip is not a good undervolter, you may choose to underclock the FSB to get there, but then you would trade off cpu performance and memory bandwidth for power savings.

I have an E8200 C0, but I'll still post my results.
Through BIOS, the lowest VCore I can run at stock speed 2.667 GHz is 1.1125v, which gives CPU-Z idle voltage of 1.072v and load 1.056.
Overclocking it to 400 MHz bus for a 3.2 GHz clock speed requires BIOS voltage 1.2750v. In Windows CPU-Z reports a idle/load voltage of 1.232/1.216v.
The CPU is passively cooled by a Ninja Mini, so the temperatures are quite high, which might require higher voltages than usual.

0,800V (minimum that my BIOS settings offer) is reached with 1,8GHz underclocking for E8400. I have not tested this voltage with faster clock speeds, but i have stress tested it for an hour so it seems pretty stable.

Update for those still interested in this ancient CPU.

Put my memory out of the equation by lowering the memory multiplier. Then I lowered the VCORE as low as I could (0.9) and set the CPU multiplier to 6, and finally began raising the FSB. I managed to reach an FSB of 360 (6x360 = 2160Mhz) and yet have a 100% stable system.

BIOS says VCOM 0.9, but CPU-Z showed 0,864. Wonder what is real... Regardless of that I might remove the word "disappointed" from the topic title