[RFC] Enabling invariant TSC timecounter on SMP

[Andriy Gapon]

on 03/06/2011 19:28 Jung-uk Kim said the following:
> Unlike Intel, AMD did not guarantee "all TSCs reset to zero with RESET 
> IPI" before Bulldozer[1].  In fact, I tried to measure deltas between 
> cores when I started hacking on it using some crude heuristics, 
> somewhat like the OpenSolaris hack[2].  Basically, a dual-core AMD 
> Family 10h processor showed noticeably larger deltas than *two* 
> dual-core Intel Woodcrest Xeons'.

You are right.  I haven't had any problems with my Athlon II system with forced
smp_tsc, but testing shows that it has a measurable difference in TSC between
cores.  E.g. with the "tsc_check" code:

cpus 0-1, min_write_time = 186, tdelta = 316
cpus 0-1, TSCs are considered to be OUT of sync


> [1] I couldn't find any clues from their publicly available documents 
> whether they will implement (or need) additional mechanism for 
> multi-socket Bulldozer platforms.  It only says something like "all 
> TSCs are synchronized with a clock source in north bridge".  We will 
> see when AMD Valencia & Interlagos are available. :-)
> [2] Unfortunately, there is no way to accurately measure it with 
> current generation hardware.

Yeah, quite unfortunate.

-- 
Andriy Gapon