Messages in this thread | | | Date | Mon, 17 May 1999 19:06:45 -0700 | From | dave madden <> | Subject | Re: gettimeofday non-monotonic on 2.2.7 SMP |
| |
=>From: Simon Kirby <sim@netnation.com> =>... =>My dual celeron 300As @ 450 MHz (2.2.9) running your program gives: => =>Mon May 17 18:05:40 1999: fwd:774496/0:back max 0000000 =>Mon May 17 18:05:41 1999: fwd:913473/0:back max 0000000
That's what I expect, but what I get is:
Mon May 17 18:57:17 1999: fwd:362637/4:back max 0000003 Mon May 17 18:57:18 1999: fwd:866801/9:back max 0000003 Mon May 17 18:57:19 1999: fwd:873531/7:back max 0000004 Mon May 17 18:57:20 1999: fwd:872292/8:back max 0000003 Mon May 17 18:57:21 1999: fwd:873949/9:back max 0000003
with max [i.e. maximum clock decrement detected] going up to 5000-10000 (5-10ms) when I'm writing a CD or scrolling a Netscape window.
Somebody else also reported success with a Dell SMP box running 2.2.9, so I compiled a copy for myself, but it didn't make any difference from 2.2.7.
=>Another thing..Perhaps try enabling "MTRR" in the kernel if it's not =>already to synchronize TSCs at bootup. I'm not sure if MTRR does this or =>not, actually, so maybe try flashing a new BIOS if all else fails. I know =>some BIOSes don't synchronize the TSCs properly, and this would confuse =>gettimeofday()...
I'm using MTRR, but that code doesn't mention the TSC. My box is an Intel N440BX motherboard with the latest BIOS; what are your Celerons on?
Thanks for trying the program!
d.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |