I think the "auto" is supposed to be controlled through the bios. I also would not trust this with a valuable low voltage CPU, because it would first need to post at a default, probably high, voltage.

Note that socket 7 can use either single or dual voltage CPUs. If you put a single voltage CPU in the socket, it will short the two different voltages together, inside the CPU, no matter what the motherboard is trying to do. Trying to measure voltages with no CPU installed can also be complicated. There are many threads about problems related to single/dual voltages, VRMs, different power implementations for socket 7, etc.

Hi ChrisK,

What I understood about the “auto voltage” function on these Gigabyte socket 7 boards, is that it can only switch between 3.5V single-voltage supply and 2.8V Vcore for dual-voltage CPUs. For any CPU with a different voltage requirement, you have to set that voltage manually with the DIP-switches 4,5,6,7 as show on page 47 of the manual.

I assume this auto-switching makes use of CPU pin AL-1 (VCC2DET#), which is pulled low on dual-voltage CPUs only.

Looking at the VID input table of the HIP6008 and compare it with the SW 4,5,6,7 table in the manual, it appears that SW4 is connected to VID2, SW5 to VID1, SW6 to VID0, and SW7 to VID3.
Note that the “auto” setting is the same as the 2.8V setting so essentially there is no manual 2.8V setting and it depends on the installed CPU if you get 2.8V or 3.5V with this setting.

Theorizing how this works, I looked at the VID3:0 pattern for 2.8V and 3.5V and noticed that you can change from 2.8V to 3.5V by pulling VID0, VID1, and VID2 low. So this must be where these 3 transistors on the VID0..2 lines are for.
Then there must be some logic that takes the auto/2.8V SW4..7 pattern and combines it with the VCC2DET# signal to let these 3 transistors conduct or not.

If this works differently on your board, there may be some fault. Reading an occasional 3.17V is not how this circuit should work.

Looking into the BIOS, I didn’t find any voltage control there. So this auto-switching must be done in hardware on the board.

Greetings, Jan

Hi Jan,

just had another look on the board and it is completely as you said.
After a lot of searching and measuring I eventually could identify the components AL1 is connected to. They are placed below the IDE ports, so quite a bit away.
AL1 as coming from the CPU socket is pulled high, then fed into an inverter which is driving a transistor which drives the three transistors connected to VID0, 1 and 2. Hmmmpf.
There's even an unpopulated jumper for tying AL1 low manually...

So pulling AL1 low should disable these three transistors and allow setting the voltage for the CPU core freely with the DIP switches.
When I have another minute or two I will test this out (need to set up the board and everything first) and then we'll see what we get. But it can only be like that.

Guess I was confused by the 3.5V which I considered too much for most single voltage CPUs.
And I would also have expected some more "magic" from this Auto Voltage feature 😉
But it is what it is.

Big Thanks again for your support and giving ideas to check for!

BTW:
I had a short test with the Unicore BIOS you gave me here .
What I can say at this point is that it is working but unfortunately looses some features of the last official F.4 version like voltage monitoring in BIOS, Vcore display on POST screen and also USB keyboard support (while it's working in BIOS it curiously doesn't on the DOS prompt anymore).
I haven't yet tested any other "modern world features" such as K6-2/-III(+) support, UDMA and large drive support because of the auto voltage thing. Will give it another more extensive shot as soon as possible.

nickles rust wrote on 2026-03-17, 13:36:

I think the "auto" is supposed to be controlled through the bios. I also would not trust this with a valuable low voltage CPU, because it would first need to post at a default, probably high, voltage.

Note that socket 7 can use either single or dual voltage CPUs. If you put a single voltage CPU in the socket, it will short the two different voltages together, inside the CPU, no matter what the motherboard is trying to do. Trying to measure voltages with no CPU installed can also be complicated. There are many threads about problems related to single/dual voltages, VRMs, different power implementations for socket 7, etc.

Hi nickles rust,

The different implementation of an “Auto Vcore” function on several socket 7 boards surprised me as well. Gigabyte did this (limited) function in hardware, without any BIOS involvement, on several of their GA-586xx boards.

If ChrisK’s Auto Vcore issue was on one of those (semi-)jumperless boards, your remark about BIOS control of the Vcore would have been absolutely correct.
However, socket 7 boards that are truly jumperless are quite rare, and the so called semi-jumperless boards are even more rare.

These semi-jumperless boards usually still have jumpers for FSB and Multiplier, but can control the CPU Voltage through the BIOS.
I’ve only ever encountered 4 of those, the Freetech P5F85/87 and the Shuttle HOT-566, HOT-569, and HOT-569A. The Shuttle boards even can control the CPU Voltage either through jumpers, or via the BIOS. They use a special jumper pack to put over the Vcore jumpers to enable BIOS Vcore control.

True jumperless socket 7 boards that I’ve encountered were made by Abit (IT5V, AX5, PX5, TX5), Chaintech (5TDM2), PCChips (M590), QDI (Titanium IB, IB+, IIB, IIIB, Advance 3), and Zida (TX100/TX100-3D). So only 12 in total.

I have also been wondering how these jumperless boards control the initial CPU voltage before the reset line goes low and the CPU starts executing code.
Without jumpers or dipswitches to set the voltage for the CPU, and with the wide range of required voltages for various S7 CPUs, you need the motherboard to provide a sufficient high voltage for 2.8V/2.9V Vcore CPUs to start reliably but not too high for 2.1V/2.2V models to get damaged.

For dual voltage CPUs a 2.6V or 2.7V Vcore is sufficient to start higher Vcore models and not too high to damage low Vcore parts. I’ve found that a 2.6V initial voltage value is often used in these jumperless boards. After a few milliseconds the BIOS will take over and set the correct value for the installed CPU.
For single voltage CPUs however, this 2.6V would be too low. But this is solved by pulling the Voltage up to 3.4V by using the CPUs VCC2DET# pin, that tells the CPU voltage circuit if the CPU is a single- or dual voltage model.

The tricky part on these jumperless boards starts when the BIOS takes over the Vcore control. If the BIOS is too old to recognize later 2.0V- 2.2V Vcore CPU models, they will receive a too high voltage and may be damaged. When I started to experiment with a 2.2V K6-2 CPU on my jumperless Abit AB-PX5, I measured a 3.2V Vcore on the CPU, ouch!! ☹
So that’s why patching a BIOS for K6-2(+)/III(+) support on these jumperless boards is even more important than on the usual jumpered boards. 😉

Cheers, Jan