| View previous topic :: View next topic |
| Author |
Message |
Ant P.
Veteran
Joined: 18 Apr 2009
Posts: 1920
Location: UK
|
 Posted: Fri Dec 28, 2012 6:03 pm Post subject: |
 |
|
| Just applied this patch to zen-sources 3.7.1 and it seems to be working well. I can post an updated diff for that if you'd like. |
|
| Back to top |
|
 |
_______0
Guru
Joined: 15 Oct 2012
Posts: 335
|
| Posted: Fri Dec 28, 2012 6:16 pm Post subject: |
|
|
cool, I'd like to try since disabling ohci driver resulted in several USB ports not working anymore :/
What about the timer question?? |
|
| Back to top |
|
|
_______0
Guru
Joined: 15 Oct 2012
Posts: 335
|
| Posted: Tue Jan 08, 2013 1:32 pm Post subject: |
|
|
sorry if it's a dumb question, but how to apply the patch? I want to use it.
I am drooling over this :/ |
|
| Back to top |
|
|
roarinelk
Guru
Joined: 04 Mar 2004
Posts: 381
|
| Posted: Tue Jan 08, 2013 5:09 pm Post subject: |
|
|
| _______0 wrote: | | roarinelk wrote: | | maudeb wrote: | 4 965
...but I still can't see HPET_MSI-edge (BLK_DEV_FD is not set). I'd really like if someone could port this on the kernel tree.
|
a) it's overrated (APIC timer has a much higher resolution than HPET),
b) you need to set CONFIG_HPET=n AND have a HPET with 3 or more timers for one of the
entries to appear. |
this is interesting, how to pick APIC timer?? I don't see it in /sys/bus/clocksource/devices/clocksource0/available_clocksource
|
go with tsc if it is in there, it's the most precise one (runs with CPU clock).
The APIC timer is the "LOC" line in /proc/interrupts, I *think* it's used for the scheduler
| Quote: |
And how do you determine HPET is overrated and apic timer is better??
|
the hpet is an awkward to program 14.something MHz timer in the southbridge;
the apic is the interrupt controller in the cpu which usually runs at bus clock and
raises an irq to the cpu it is attached to (whereas the hpet usually raises global
interrupts to all cpus).
Last edited by roarinelk on Tue Jan 08, 2013 5:12 pm; edited 1 time in total |
|
| Back to top |
|
|
roarinelk
Guru
Joined: 04 Mar 2004
Posts: 381
|
| Posted: Tue Jan 08, 2013 5:09 pm Post subject: |
|
|
| _______0 wrote: | sorry if it's a dumb question, but how to apply the patch? I want to use it.
I am drooling over this :/ |
you should start by reading the wiki sections regarding kernel sources and the patch utility! |
|
| Back to top |
|
|
_______0
Guru
Joined: 15 Oct 2012
Posts: 335
|
| Posted: Wed Jan 09, 2013 11:17 am Post subject: |
|
|
wow, your true!!
| Code: | | hpet0: 3 comparators, 32-bit 14.318180 MHz counter |
compared to:
| Code: | | tsc: Detected 3110.206 MHz processor |
however hpet wins in floating point numbers, hpet "14.318180 MHz" against tsc "3110.206 MHz". Does it mean hpet is more accurate? You know, more decimals (floating point) higher precision, moreover there are THREE hpet's.
| roarinelk wrote: | | go with tsc if it is in there, it's the most precise one (runs with CPU clock). |
According to the wiki http://en.wikipedia.org/wiki/RDTSC TSC has serious drawbacks with synching. And it's not entirely clear late in the paragraph whether the fixes to stabilize the TSC clock can substitute HPET.
| roarinelk wrote: | The APIC timer is the "LOC" line in /proc/interrupts, I *think* it's used for the scheduler
|
The line in /proc/interrupts:
| Code: | | LOC Local timer interrupts |
does not belong exclusively to TSC timer, I am using hpet and there's still activity in there.
So the question remains, HOW DO YOU DETERMINE TSC IS BETTER???
PS: PLZ SEND ME TEH CODEZ!! No seriously, is that hard to copy and paste the command so I can patch this kewl thing immediately??? I am not saying not to RTFM, but in the meantime, while I RTFM, I can have the kernel nicely patched. pls cut and paste here how to patch this. Besides you didn't tell me which wiki to read. |
|
| Back to top |
|
|
|
Unsupported Software
