Ubuntu
linux-source-2.6.20 package

throttling <not supported> on Core2 T7200

Bug #112104 reported by Alan Pope 🍺🐧🐱 🦄
2
Affects Status Importance Assigned to Milestone
linux-source-2.6.20 (Ubuntu)
Invalid
Undecided
Unassigned

Bug Description

Binary package hint: linux-source-2.6.20

New Toshiba M400 with Core2Duo T7200@2GHz.

$ cat /proc/acpi/processor/CPU?/throttling
<not supported>
<not supported>

The Ondemand speed govenor is throttling the cores to 1Ghz and the fans work so it's not overheating, but it would be preferable to have it scale properly with powernowd.

$ powernowd
Found 2 scalable units: -- 1 'CPU' per scalable unit
cpu0: 1000Mhz - 65535Mhz (5 steps)
cpu1: 1000Mhz - 65535Mhz (5 steps)

Revision history for this message
Alan Pope 🍺🐧🐱 🦄 (popey) wrote :
Revision history for this message
Alan Pope 🍺🐧🐱 🦄 (popey) wrote :
Revision history for this message
Alan Pope 🍺🐧🐱 🦄 (popey) wrote :
Revision history for this message
Alan Pope 🍺🐧🐱 🦄 (popey) wrote :
Revision history for this message
Alan Pope 🍺🐧🐱 🦄 (popey) wrote :
Revision history for this message
Alan Pope 🍺🐧🐱 🦄 (popey) wrote :
Revision history for this message
Alan Pope 🍺🐧🐱 🦄 (popey) wrote :

I have checked the BIOS and turned off the speed restrictor there, now the gnome frequency applet and /proc/cpuinfo both report 2GHz. So no frequency scaling going on.

powernowd reports 2 steps. 2GHz and 65536Mhz

Revision history for this message
Alan Pope 🍺🐧🐱 🦄 (popey) wrote :

http://acpi.sf.net/dsdt/index.php indicated that dumping out using acpidump seems like a good idea.

Revision history for this message
Alan Pope 🍺🐧🐱 🦄 (popey) wrote :

I don't know if this is related or should be a separate bug, but here goes.

The raw performance of the CPU seems to be quite sub-optimal. As a very very very simple an arbitrary benchmark I run the following on every machine I own:-

time perl -e 'for($i=0;$i<1e8;$i++) { }'

On my core2duo desktop at 2.66GHz takes about 6 seconds. Friends of mine have tested (repeatedly) and a 1.8Ghz core2duo takes about 10 seconds, so I would expect my 2.0GHz core2duo laptop to take somewhere between the two. Instead it takes 20 seconds. This is consistent whatever is currently running, booted to a console or into X.

http://hants.lug.org.uk/cgi-bin/wiki.pl?HugoRandomBenchmark shows some comparisons.

Yes, I know it's no benchmark, but it is a good quick gauge of raw cpu performance.

Is this related?

Revision history for this message
Alan Pope 🍺🐧🐱 🦄 (popey) wrote :

I ran the linux firmware kit bootable cd, here's the results - not good. Looks like this may be a firmware issue. Will bring it up on LKML.

Revision history for this message
Alan Pope 🍺🐧🐱 🦄 (popey) wrote :

Fixed by BIOS update 3.5 from toshiba

Changed in linux-source-2.6.20:
status: Unconfirmed → Rejected
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.
Subscribing...

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.