[12:09] <popey> last comment on bug 112104 specifically
[12:49] <mjg59> popey: Like I said, the lack of throttling isn't a problem
[12:50] <mjg59> So regardless of anything else, the bug is mistitled :)
[12:50] <popey> heh
[12:51] <popey> suggestions welcome then
[12:53] <mjg59> But yeah, the scaling stuff looks utterly funted
[12:53] <mjg59> What hardware is this?
[12:53] <popey> tosh tablet laptop
[12:53] <mjg59> Other than that, the failures are fairly harmless
[12:53] <mjg59> Does Windows scale it correctly?
[12:54] <popey> it doesnt feel quick though
[12:54] <popey> dunno, dont use windows
[12:54] <mjg59> Right, the scaling brokenness is probably killing performance
[12:54] <mjg59> If you run something cpu intensive, do the values in /proc/cpuinfo rise?
[12:55] <popey> it always sits at 2GHz
[12:55] <popey> never changes
[12:55] <mjg59> Yeah, that's fucked
[12:55] <popey> my apologies if i disappear.. on ropey wifi at UDS
[12:56] <mjg59> Also, don't use powernowd directly - the init script should just be setting up ondemand and leaving it up to the kernel
[12:56] <popey> ok
[12:56] <mjg59> Can you try unloading speedstep_centrino and load acpi-cupfreq instead?
[12:57] <popey> sure
[12:57] <popey> speedstep_centrino isnt currently loaded
[12:57] <mjg59> ...
[12:57] <mjg59> Ok, so that's a pretty poor sign to begin wth
[12:57] <popey> acpi_cpufreq is
[12:57] <mjg59> Ok, that might explain it
[12:57] <mjg59> In that case, try unloading acpi-cpufreq and load speedstep-centrino
[12:58] <popey> ok
[12:58] <popey> FATAL: Error inserting speedstep_centrino (/lib/modules/2.6.20-15-generic/kernel/arch/i386/kernel/cpu/cpufreq/speedstep-centrino.ko): No such device
[12:58] <mjg59> Anything appear in dmesg then?
[12:58] <popey> no, but I get a lot of these all the time:-
[12:59] <popey> [ 5424.824000]  APIC error on CPU0: 40(40)
[12:59] <popey> [ 5424.824000]  APIC error on CPU1: 40(40)
[01:00] <mjg59> Ok. I think you'll need to work with Ben to figure out why speedstep-centrino isn't working. In theory, the acpi_cpufreq stuff should work.
[01:00] <popey> heh, spoke to ben the other day, he said "it's broken"
[01:00] <mjg59> Well, that's the obvious conclusion
[01:00] <popey> well, actually said "sounds liek hardware failure"
[01:01] <mjg59> What was the bios option that you changed actually called?
[01:01] <popey> sorry, can't remember, it was something about cpu scaling
[01:01] <popey> but not the word scaling
[01:02] <popey> will check in the morning when i reboot
[01:02] <mjg59> It may be that that needs to be enabled
[01:02] <popey> is this the kind of thing that might be fixed by a bios update or is it more likely to be a kernel thing do you think?
[01:02] <mjg59> Sounds like a bios thing
[01:03] <popey> ok, will re-enable when i boot in the morning and try the speedstep_centrino module, thanks
[01:11] <popey> meh, couldn't waity
[01:11] <popey> it is called "Dynamic CPU Frequency Scaling" - I had it set to "Always high", now it is "Dynamically Switchable"
[01:11] <mjg59> Ok
[01:11] <popey> (last option is "always low")
[01:12] <mjg59> "Dynamically Switchable" sounds like the right value
[01:12] <popey> cpuinfo says 1GHz
[01:12] <mjg59> Ok. If you do something that loads the CPU (like dd if=/dev/zero of=/dev/null) does that rise?
[01:12] <popey> alan@mother:/proc/acpi/processor/CPU0$ cat throttling
[01:12] <popey> <not supported>
[01:12] <popey> one mo
[01:13] <popey> yes
[01:13] <popey> cpuinfo says 2GHz now
[01:13] <mjg59> Ok. So that works.
[01:13] <mjg59> throttling isn't the same thing as frequency scaling
[01:13] <popey> and back to 1Ghz
[01:13] <popey> ah, ok
[01:14] <mjg59> Throttling just drops the clock rate of the processor. It doesn't drop the voltage. As a result, it's entirely useless for anything other than emergency limitation, and even then scaling is preferable.
[01:14] <popey> right
[01:14] <mjg59> So don't worry about the lack of throttling
[01:14] <popey> wilco
[01:14] <mjg59> Based on your descriptions, everything now sounds fine except for the oddly low performance
[01:15] <mjg59> Which is somewhat trickier to track down...
[01:15] <mjg59> Your laptop probably has less cache than the desktop machine, but I would expect that to make a huge difference in the case you're describing
[01:15] <popey> kind of a bummer when i was hoping to do some cpu intensive stuff with this thing :(
[01:15] <mjg59> I'd recommend some more solid benchmarking
[01:16] <popey> ok, the unix bench thing?
[01:16] <popey> Byte Unix Benchmark
[01:16] <mjg59> For instance
[01:16] <mjg59> I've never done any heavy CPU benchmarking, so...
[01:16] <popey> http://hants.lug.org.uk/cgi-bin/wiki.pl?ByteUNIXBenchmark is the kind of thing
[01:17] <popey> which seems relatively comprehensive
[01:19] <popey> thanks for the time spent on this mjg59, another beer in the nudge bank
[12:58] <Fahuadai> I have heard that 7.04 has inproved support for broadcom wireless cards including the bcm43** series. I've been trying for a while to get wpa2 working on a university network with no luck so far.
[12:59] <Fahuadai> would upgrading to 7.04 help?
[06:53] <elcuco> hi
[06:54] <elcuco> I am using Kubuntu 6.10 (in 2 hours 7.04) on a Lenovo 3000 N100, and found this site:
[06:54] <elcuco> https://wiki.ubuntu.com/LaptopTestingTeam/Lenovo3000N100
[06:55] <elcuco> some things are not exactly corrent in that wiki page.
[06:55] <elcuco> for example external monitor did not work for me on the 7.04 livecd
[06:55] <elcuco> screen is not detected out of the box (915 resolution)
[08:46] <crimsun> elcuco: you need to use xserver-xorg-video-intel, of coure.
[08:46] <crimsun> course ^
[08:46] <crimsun> forget xserver-xorg-video-i810 + 915resolution
[11:47] <peterka> Hi