From: Lonnie A. <li...@lo...> - 2012-12-29 22:22:07
|
Hi Devs, Just documenting an interesting observation... On my multi-core Atom boxes, with the minor kernel change from 2.6.35.13 to .14, the testing of the 'tsc' clocksource seems to be more strict, resulting in the 'hpet' clocksource being chosen instead of the 'tsc' clocksource. $ cat /sys/devices/system/clocksource/clocksource0/current_clocksource hpet -- Console snippet -- ACPI: Core revision 20100428 Enabling APIC mode: Flat. Using 1 I/O APICs ..TIMER: vector=0x30 apic1=0 pin1=2 apic2=-1 pin2=-1 CPU0: Intel(R) Atom(TM) CPU D525 @ 1.80GHz stepping 0a Booting Node 0, Processors #1 Initializing CPU#1 TSC synchronization [CPU#0 -> CPU#1]: Measured 54 cycles TSC warp between CPUs, turning off TSC clock. Marking TSC unstable due to check_tsc_sync_source failed #2 Initializing CPU#2 #3 Ok. Initializing CPU#3 Brought up 4 CPUs ... HPET: 3 timers in total, 0 timers will be used for per-cpu timer Switching to clocksource hpet -- From my research. the 'hpet' clocksource seems to be the more robust choice, but 'tsc' is simply faster to access. Lonnie |