all voltages on zero but still running

[the-fallen]

Strange. But also strange that there is a Frequency that is just 1MHz higher than the prior one.

Try to set VIDs to the second-highest frequency and check it then. Does it seems to be stable then?

I saw those "strange magic frequency" on ubuntu, too. But that was nearly an year ago. Maybe there is a piece of software that misintepretates something.

By intels documentation there can not be a FID higher than 63 because they reserved 6 Bits on a register for FID, one is not used and one is displaying the "DID" (we call it DID - Divisor ID and it is used to change the BUS Frequency). Maybe something n your system gets stuck with that.

Using windows, do you normally have only 4 frequency steps ? Then there should be 4 too in linux because both system get those information from the BIOS.

[evil]

i am using ubuntu.
i have also windows 7, what can i use to check FIDs since rmclock is not supported?

[the-fallen]

If Rightmarks CPUClock utility is not yet working the only thing i know is NHC.

Which version of ubuntu are you using? (I am running 9.10 - maybe you can test if your system is working better/correct using a boot-usbstick with ubuntu 9.10 beta on it)

[evil]

damn, i posted rmclock results days ago and didn't realized it actually didn't (attachment too big)

here it is (again...)

[evil]

since you are working on a new release, i'll wait for that, but i believe that rmclock screenshot could give you some interesting information abour the weird frequencies of this kind of cpus

[the-fallen]

Thanks for the screenshot.
It may take some time until the new release and even than it needs to get mature until I can say its ready for broad use but I will bring it to public fur users that want to test it as soon as it is complete.

[evil]

ok let me know if you need anything from this cpu
for now i find it weird that the fid in ubuntu are different than windows.
there is a different number of them, and there is the first and the last with a particular name.

[evil]

update on the situation, could be useful for you.

i upgraded from jaunty to Katatonic Koala, then i installed the pae-phc kernel.
even here as soon as i try any vid with the max freq (2668), it automatically scales down to 2667, no way to force it to stay at 2668.
so what i did is to skip that freq and undervolt the others.
result (with safe margin):
74:43 10:27 8:2 6:2 134:2

unfortunately i noticed that with those settings, the mouse arrow seems stuttering, thus performance is not optimal.
i have no idea what could be the reason, maybe te fact that there are 2 vids much different from the rest and it has a lower transition to change so much voltage up and down? (just a wild guess...)

[the-fallen]

At the meantime we found out about those "strange" FIDs.

The 0.3.x - patches does not interpref the Byte where the FID is stored correctly.

Your first FID ( 74 ) is FID 10 with another Bit set that enables Non-Integer Bus-Ratio (it adds "0.5" to the FID set.) So the real FID is 10.5 and marks an OP-Point known as "IDA".

The IDA OP-Point is not supposed to be set and used manually.

The other strange FID ( 134 ) is in real FID "6" but with a Bit set that enables "SLFM".

SLFM does nothing else but telling the CPU to only use each second clock tick. So internally the CPU is halving the BUS frequency.


The next Version of PHC ( 0.4 ) will support all these things. It will be released soon.