Needed: A little CPU voltage and frequency 101

[eightbit]

It has been a LONG time since I have done this. Long story short, I discovered an old Alienware with an AMD Athlon 1200 CPU (Thunderbird) and I noticed the VCORE in the BIOS is reporting at 1.81V. That can't be right as the spec for this CPU is 1.75V:

 

http://www.cpu-world.com/CPUs/K7/AMD-Athlon 1200 - A1200AMS3C.html

 

The motherboard is an old FIC AD11 board, and amazingly I was able to find the manual and switch settings. Looking at the board, I see the switches are all off and the jumpers for autodetect for the CPU core voltage and frequency are both enabled. So, the board is taking control of this and applying what it thinks are the proper settings. But again, the core voltage is wrong...so the board is not detecting this correctly. I suspect it has always been this way....but I would like to correct it. Looking at the manual here is the voltage core settings:

 

 

 

So, jumper in disable and SW3 to OFF looks like the plan here.

 

 

But then there is a second set of switched for frequency ratio:

 

 

 

 

It might sound stupid, but what do I choose here? Is frequency voltage times the ratio? Bus speed times the ratio? And if bus speed times ratio, is it 133mhz bus or 266mhz bus as the processor is using double data rate?

 

 

I used to install these all of the time, but my memory is failing me recently Any help with the proper settings is greatly appreciated!

Edited February 14, 2020 by eightbit

[eightbit]

I think it might be 133 bus x9 for 1197? Or is my memory failing me? 

[carlsson]

Yes, it matches what the CPU World page says. Perhaps you were too tired to read what the site you linked to writes. Another way to see it, if it would be 1200/266 that is a multiplier of 4.5 which is below the lowest possible choice on your motherboard.

[eightbit]

Yeah, I was definitely too tired Thank you carlsson!

[mr_me]

That website could be wrong on the voltage.  Specifications from AMD would be better.

[eightbit]

The voltage on the Thunderbirds above 700MHz is 1.75V....that information is everywhere including old reviews of those chips. However, when I set these settings manually to 1.75V and the x9 multiplier for the 133MHz bus, the northbridge fan does not spin and the computer does not boot. It will only boot when using the auto settings. Very strange. Either I am wrong about the CPU or the documentation is wrong...or some other unknown problem!

[carlsson]

Do you have another power supply to test with, in case it needs 1.80V in order to pull 1.75V for the CPU? Just a stab in the dark here.

[eightbit]

3 minutes ago, carlsson said:

Do you have another power supply to test with, in case it needs 1.80V in order to pull 1.75V for the CPU? Just a stab in the dark here.

 

Good point. I don;t have another PSU (yet) but I actually have one coming. The power supply needed for these Athlon chips is special in the fact that the CPU draws from the +5v rail instead of using the +12v rail as they do today. Because of that it is very hard to source a new PSU that will be sufficient enough for these CPU's. The +5v rail on this PSU is supplying 30A for example.

[ChildOfCv]

Could the motherboard be trying to overclock it?  That typically involves increasing voltage beyond official specs.

[eightbit]

8 hours ago, ChildOfCv said:

Could the motherboard be trying to overclock it?  That typically involves increasing voltage beyond official specs.

It could be. I mean, with the default letting the motherboard detect it is working fine. But the VCORE hovers at 1.81 and 1.79. Hmmm....

 

The fan and heatsink for the CPU were cleaned thoroughly and the thermal compound (or what used to be compound) was cleaned/chopped off and new compound applied. The fan works perfectly and the BIOS is reporting the CPU sitting idle at around 42C....which from what I read is normal for the Thunderbird processors. They have a max rating of 95C. I guess it is OK as is...unless someone else with one of these has any additional advice.

 

And, if my some miracle someone else has an FIC AD11 motherboard and has the driver disk I will be forever in your debt

 

 

And funny, history repeats itself:

 

https://forums.anandtech.com/threads/ad11-overvolting-athlon-by-default.590445/

 

Edited February 15, 2020 by eightbit

[eightbit]

I found my switch issue. The first set of switches on this board provide the multiplier, and the second set the voltage. I thought it was the other way around!

 

In any case, adjusted again and it boots fine, but the BIOS *still* reports the VCORE at 1.81, even though it is force set through the dipswitches to 1.750. Whatever....  Guess that's why you don't hear about FIC motherboards nowadays...lol!

 

But, I do have to say, the board has been through hell (it was super dirty and musty smelling, and an exploded CR-2032 battery...something I never saw!) and it still works after 20 years. And, no caps are swollen. So I guess I have to give some props to FIC for durability!

[ChildOfCv]

2 hours ago, eightbit said:

And funny, history repeats itself:

 

https://forums.anandtech.com/threads/ad11-overvolting-athlon-by-default.590445/

 

Hmmm, looks like that user never got an answer either.  It could possibly be a sensor issue, but with a programmable voltage, I'd expect the sensor to also be the feedback--in other words, if it's targeting 1.75 and reading 1.80, it should be cutting back.

 

I guess a firmware bug isn't out of the question though, like maybe it's translating the read value into the wrong voltage for display, and nobody thought it was worth fixing.