Re: [PATCH] Relevance of 8254 calibration.

On Mon, 26 Nov 2001, Warner Losh wrote: > Well, I'll bet your system time has drifted about a second or four in > those two days. When you say system time do you mean "hardware" time? or the running kernel's time? The hardware time doesn't work anyway since the box's RTC is stuck somewhere in 20

Re: [PATCH] Relevance of 8254 calibration.

On Mon, 26 Nov 2001, Warner Losh wrote: > Yes. Almost *ALL* PCs in the field aren't exactly 11931282Hz. > There's a lot of variance in this. PC have such crappy oscillators > that calibration is required. The "slight" variation can be as large > as +-300Hz, which is huge. :-(. > > But I'm a li

Re: [PATCH] Relevance of 8254 calibration.

On Mon, 26 Nov 2001, Warner Losh wrote: > The higher levels of NetBSD does this if you are running ntpd. Ditto > Linux. Thanks for the pointer, i'm going to check out the NTP stuff in both OS' just now. > I measure phase differences in oscelators to sub-pico second level in > my day job :-).

Re: [PATCH] Relevance of 8254 calibration.

In message <[EMAIL PROTECTED]> Zwane Mwaikambo writes: : When you say system time do you mean "hardware" time? or the running : kernel's time? Output of date command, which is kept by the kernel. : The hardware time doesn't work anyway since the box's RTC : is stuck somewhere in 2057 but the ha

Re: [PATCH] Relevance of 8254 calibration.

In message <[EMAIL PROTECTED]> Zwane Mwaikambo writes: : PS the box runs fine with the patch right now (2 day uptime) I can't do : anything which would write to the RTC without it blocking for a long time, : but i only do project compiles on the box anyway. Well, I'll bet your system time has dr

Re: [PATCH] Relevance of 8254 calibration.

In message <[EMAIL PROTECTED]> Zwane Mwaikambo writes: : hmm cross referencing here (forgive me ;) NetBSD nor Linux do this : calibration and NetBSD runs on just about anything ;) The higher levels of NetBSD does this if you are running ntpd. Ditto Linux. I measure phase differences in oscel

Re: [PATCH] Relevance of 8254 calibration.

In message <[EMAIL PROTECTED]> Zwane Mwaikambo writes: : I'd just like to know wether there are any i386 boxes which don't have a : timer_freq of 1193182Hz I know there might be slight variations but i : still have to come across a box that doesn't, which leads me to the : purpose of this email.