Skip to contentSkip to conversation

Can't lower vcore => "vid trans failed"

DISCUSSION Manfred Langthaller 2 participants 3 messages Jun 18, 2007

lore
Share & tools
lore
#0 lore raw

Hi!

Mainboard: MSI K8MM3-V H (MSI-7181 Ver: 2.0) BIOS v3.1 (latest)
CPU: AMD mobile Sempron 3000+ (SMS3000BQX2LF, TDP 25W) Debian 4.0
Kernel: 2.6.21.5

If I insert module powernow-k8 3 P-States are recognized, but the voltages are too high.

powernow-k8: Found 1 Mobile AMD Sempron(tm) Processor 3000+ processors (version 2.00.00)
powernow-k8:    0 : fid 0xa (1800 MHz), vid 0xa
powernow-k8:    1 : fid 0x8 (1600 MHz), vid 0xc
powernow-k8:    2 : fid 0x0 (800 MHz), vid 0x13
powernow-k8: ph2 null fid transition 0xa

0: VID = 0xa = 01010 = 1.3V
1: VID = 0xc = 01100 = 1.250V
2: VID = 0x13 = 10011 = 1.075V

So I dumped the DSDT, fixed some errors, patched the kernel with "acpi-dsdt-initrd-v0.8.4-2.6.21.patch" changed the _PSS values ("control" and "status" fields) in DSDT.dsl and inserted the fixed dsdt.aml to the initrd.

Now after inserting module powernow-k8 dmesg says:

powernow-k8: Found 1 Mobile AMD Sempron(tm) Processor 3000+ processors (version 2.00.00)
powernow-k8:    0 : fid 0xa (1800 MHz), vid 0x13 => 1.075V
powernow-k8:    1 : fid 0x8 (1600 MHz), vid 0x16 => 1.000V
powernow-k8:    2 : fid 0x0 (800 MHz), vid 0x1d => 0.825V
powernow-k8: ph2 null fid transition 0xa

Changing the scaling_govenor to "ondemand" leads to following errors.

powernow-k8: vid trans failed, vid 0x1d, curr 0x13
powernow-k8: transition frequency failed
Time: acpi_pm clocksource has been installed.
Clocksource tsc unstable (delta = -277822049 ns)
(appending nohz=off highres=off to the kernel should fix the clocksource error)

All in all the vcore cannot be set lower than 0x13, even 0x13 at 1800MHz works perfectly.

Where is the problem?
Is it the board, the kernel, or me?

Thanks a lot,

Manfred

#1 lore raw
Quoted 11 lines

Mainboard: MSI K8MM3-V H (MSI-7181 Ver: 2.0) BIOS v3.1 (latest)
CPU: AMD mobile Sempron 3000+ (SMS3000BQX2LF, TDP 25W) Debian 4.0
Kernel: 2.6.21.5

If I insert module powernow-k8 3 P-States are recognized, but the voltages are too high.

powernow-k8: Found 1 Mobile AMD Sempron(tm) Processor 3000+ processors
(version 2.00.00)

Quoted 8 lines
powernow-k8:    0 : fid 0xa (1800 MHz), vid 0xa
powernow-k8:    1 : fid 0x8 (1600 MHz), vid 0xc
powernow-k8:    2 : fid 0x0 (800 MHz), vid 0x13
powernow-k8: ph2 null fid transition 0xa

0: VID = 0xa = 01010 = 1.3V
1: VID = 0xc = 01100 = 1.250V
2: VID = 0x13 = 10011 = 1.075V

These voltages are consistently 0.1V above minimum specification, but your laptop motherboard may have needed some margin. If you mess with them, you may have problems while transitioning. As you've noticed.

Quoted 9 lines

Now after inserting module powernow-k8 dmesg says:

powernow-k8: Found 1 Mobile AMD Sempron(tm) Processor 3000+ processors
(version 2.00.00)

powernow-k8:    0 : fid 0xa (1800 MHz), vid 0x13 => 1.075V
powernow-k8:    1 : fid 0x8 (1600 MHz), vid 0x16 => 1.000V
powernow-k8:    2 : fid 0x0 (800 MHz), vid 0x1d => 0.825V
powernow-k8: ph2 null fid transition 0xa

vid at 1800 MHz should be 1.2V (0xe)
vid at 1600 MHz should be 1.15V (0x10)
vid at 800 MHz shoul dbe 0.975V (0x17)

I don't know why you think you can operate at 0.825V but it's not a supported frequency on the data sheet.

Quoted 8 lines

powernow-k8: vid trans failed, vid 0x1d, curr 0x13
powernow-k8: transition frequency failed
Time: acpi_pm clocksource has been installed. Clocksource tsc unstable (delta = -277822049 ns) (appending nohz=off highres=off to the kernel should fix the clocksource
error)

Quoted 5 lines

All in all the vcore cannot be set lower than 0x13, even 0x13 at 1800MHz works perfectly.

Where is the problem?
Is it the board, the kernel, or me?

Your revised ACPI table is more than a little strange.

-Mark Langsdorf
Operating System Research Center
AMD

#2 lore raw
Quoted 44 lines · Mark
Quoted 13 lines

Mainboard: MSI K8MM3-V H (MSI-7181 Ver: 2.0) BIOS v3.1 (latest)
CPU: AMD mobile Sempron 3000+ (SMS3000BQX2LF, TDP 25W) Debian 4.0
Kernel: 2.6.21.5

If I insert module powernow-k8 3 P-States are recognized, but the voltages are too high.

powernow-k8: Found 1 Mobile AMD Sempron(tm) Processor 3000+ processors (version 2.00.00)
powernow-k8:    0 : fid 0xa (1800 MHz), vid 0xa
powernow-k8:    1 : fid 0x8 (1600 MHz), vid 0xc
powernow-k8:    2 : fid 0x0 (800 MHz), vid 0x13
powernow-k8: ph2 null fid transition 0xa
Quoted 3 lines

0: VID = 0xa = 01010 = 1.3V
1: VID = 0xc = 01100 = 1.250V
2: VID = 0x13 = 10011 = 1.075V

These voltages are consistently 0.1V above minimum specification, but your laptop motherboard may have needed some margin. If you mess with them, you may have problems while transitioning. As you've noticed.

Quoted 7 lines

Now after inserting module powernow-k8 dmesg says:

powernow-k8: Found 1 Mobile AMD Sempron(tm) Processor 3000+ processors (version 2.00.00)
powernow-k8:    0 : fid 0xa (1800 MHz), vid 0x13 => 1.075V
powernow-k8:    1 : fid 0x8 (1600 MHz), vid 0x16 => 1.000V
powernow-k8:    2 : fid 0x0 (800 MHz), vid 0x1d => 0.825V
powernow-k8: ph2 null fid transition 0xa

vid at 1800 MHz should be 1.2V (0xe)
vid at 1600 MHz should be 1.15V (0x10)
vid at 800 MHz shoul dbe 0.975V (0x17)

I don't know why you think you can operate at 0.825V but it's not a supported frequency on the data sheet.

Quoted 8 lines

powernow-k8: vid trans failed, vid 0x1d, curr 0x13
powernow-k8: transition frequency failed
Time: acpi_pm clocksource has been installed.
Clocksource tsc unstable (delta = -277822049 ns)
(appending nohz=off highres=off to the kernel should fix the clocksource error)

All in all the vcore cannot be set lower than 0x13, even 0x13 at 1800MHz works perfectly.

Quoted 11 lines
Quoted 2 lines

Where is the problem?
Is it the board, the kernel, or me?

Your revised ACPI table is more than a little strange.

-Mark Langsdorf
Operating System Research Center
AMD

Hi Mark & Thanks a lot!

First of all, it's a desktop board and I'm just playing around ;-) Second I searched for a datasheet but did'n find one.

Thank you for clarification!

Have a nice day,

Manfred