Re: Call-gate-like mechanism

Patrick Schaaf wrote:
> As far as I can see, the only sane fix would be an RDTSC that returned
> not only the cycle count, but also a processor ID.

Why not do it the same way as on IBM mainframes, where the microsec 
system clock is synchronized across all cpus?

Yeah, I realize that this would fail if you had different speed grades 
in the same system.

OTOH, if timing is really critical for some app, then it makes sense to 
do some processor affinity mapping, and lock the relevant process to a 
single cpu. With multiple processes/threads, each process could be on a 
different cpu, it simply couldn't be easily migrated after startup.

Terje
--
- <[EMAIL PROTECTED]>
"almost all programming can be viewed as an exercise in caching"