powernow-k7 for XP-M 2400+ CPU

Hi,

could any kind soul please send me the powernow-k7 output for his Athlon XP-M 2400+ CPU (cpuid 0x7A0)?

Background: my Asus A7V8X BIOS does not support Athlon mobile CPUs (and probably never will). But it is fairly easy to modify powernow-k7 in a way to set the PST data on the module command line, fill a fake struct pst accordingly and get a working setup. This could be a workaround for all the people who run mobile CPUs on desktop boards or broken notebook BIOSes and would like to save some power. OK we will not be able to set the voltage but setting the clock rate would help anyway. Collecting PST data for various cpuids would help here BTW.

Remark: my Asus board w/ BIOS 1014 (latest) recognizes the CPU as an Athlon MP 1800. Duh. So the powernow and acpi modules do not get any chance to do anything because they don't get advertised a mobile CPU.

Yes, I am doing this on my own risk, but as long as you don't overclock there should be no problem at all. Be sure the patch will contain a warning in capital letters not to use this feature for overclocking...

Ciao,
hm

F u cn rd ths u cnt spl wrth a dm!

Hi,

I'm not sure that you can do it in a so straightforward way, as far as I know, mobile and desktop Athlon CPUs use slightly different tables for the VID codes, so setting a FID/VID combination that is known for a Mobile CPU could not have the same effect on a desktop CPU.

I have some experince in that because I upgraded my 1GHz Mobile Duron to Athlon XP mobile 1900+ on my VAIO FX-502, and I had to hack the cpufreq code to add custom table for the CPU scaling

I used the CrystalCPUID at
http://crystalmark.info/?lang=en
to determine the FID/VID pairs that worked for me.

You may also want to see this article about mobile<->desktop Athlons CPUs:

http://www.cpuheat.wz.cz/html/AXP_multiplier/AXP_Multiplier.htm

A table with the differences in setting VID is given there, may be you will have to modify also the VID states table that is defined in the cpufreq driver.

There is even a better page at
http://people.overclockers.ru/xKVtor/records
but unfortunately it is in Russian. :( I don't really know how good is the altavista web translator but you could try using it if the info on the first page is not enough.

Hope this helps a bit,

Alex

http://www.linux.org.uk/mailman/listinfo/cpufreq

__________________________________
Do you Yahoo!?
Friends. Fun. Try the all-new Yahoo! Messenger.
http://messenger.yahoo.com/

Well, I will post probably a patch for this difference. But for the Cool'n Quiet technology as you refer I guess, it's not possible to change VID anyway. Then, I do consider it more as comestic bug for kernel logs and such.

If the motherboard do allow to change the voltage at runtime, that is by another way, and I'm afraid this is specific to Asus.

Of course, I may be wrong. Some motherboard do have a VRM mobile part instead of a desktop one.

Cheers,

Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.

a dmesg please for the powernow-k7 stuff? Not all athlon are powernow capable at first.

Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.

This is why I said "a Athlon 0x7A0 XP-M 2400+" because this is what I have on my board.

So again, could any kind soul please send me the powernow-k7 dmesg output for his Athlon XP-M 2400+ CPU (cpuid 0x7A0)?

And yes, this straightforward patch _is_ possible. I tried this with some fancy figure I made up but I'd rather use the "correct" ones, and it works.

Thanks!

Cahn's Axiom:
	When all else fails, read the instructions.

Yes, I can probably help you somehow. But send the dmesg at first. You only waste my time if you don't post that damn log.

Cheers,

Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.

My CPU says,

powernow: PowerNOW! Technology present. Can scale: frequency and voltage.
powernow: Found PSB header at c00f0810
powernow: Table version: 0x12
powernow: Flags: 0x1 (Desktop voltage regulator)
powernow: Settling Time: 100 microseconds.
powernow: Has 1 PST tables. (Only dumping ones relevant to this CPU).
powernow: No PST tables match this cpuid (0x7a0)
powernow: This is indicative of a broken BIOS.
powernow: See http://www.codemonkey.org.uk/projects/cpufreq/powernow-k7.shtml

and the CPU is seen by the BIOS as

processor : 0
vendor_id : AuthenticAMD
cpu family : 6
model : 10
model name : AMD Athlon(TM) MP
stepping : 0
cpu MHz : 1800.924
cache size : 512 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mp mmxext 3dnowext 3dnow
bogomips : 3563.52

But it is a 2400+ XP-M with a Barton core. I suspect the BIOS says Athlon MP because the MP's and the XP-M's L5.3 bridges are open which confuses the BIOS. The CPU runs at 133x13,5, i.e. within its specs. I do not intend to overclock it although the mobiles are said to be well overclockable.

The only thing I need is the powernow-dmesg output of a similar CPU on whatever machine, and then I will patch the powernow-k7 module to have it fake a similar PST. Simple enough as it seems.

TIA!!!

"In short, _N is Richardian if, and only if, _N is not Richardian."

Hi Harald,

Ok. Then the bios detect somehow that the processor is powernow capable but is buggy in regard to generate the PST table at POST.
It should be helpfull that you run that little toy:

http://poup.poupinou.org/cpufreq/powernow-k7/dump_psb.c

Just to see what's there in this PST.

Also the flags from the bios indicate that there is a desktop voltage regulator, and AFAIK it's not possible to change voltage from that.

I suspect more that the BIOS is not able to says if an Athlon is mobile, or not. Actually there is very few differences (in detection code, I mean) to sort a MP from a mobile one.

Cheers,

Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.