 | |
Expected accuracy of time clock on Intel Core-2 with Vista??
Hardware (General Discussions) General Hardware Discussions (alt.comp.hardware)
 |
| | Thread Tools | Search this Thread | Display Modes |
#1
 
| |||
| |||
|
Expected accuracy of time clock on Intel Core-2 with Vista?? -
05-14-2007
, 10:49 AM
#2
| |||
| |||
|
|
What is the expected accuracy for the time clock on a PC with Intel Core-2 6300 at 1.86GHz, 32-bit Vista? My Vista synchronizes with an Internet server on Sat at 9pm. I checked on Sat after 9pm and the PC time clock was accurate to every second. On Sunday evening it was 2 seconds fast, today on Monday it is 4 seconds fast, i.e. it seems to gain 2 seconds per day. Does it mean that the latest PC technology still cannot ensure time clock accuracy at least -/+ 1 second per week? Are there any expansion boards available to rectify the problem? |
time standard and that has almost nothing to do with "the latest PC
technology". It has almost everything to do with saving a penny. As for
expansion boards, why bother? If you have a reliable internet connection
you can synchronize your clock more frequently and achieve +/- 1 second
accuracy very easily. I synchronize my machines once per day using a
free program called atomic.exe but there are others out there which are
equally free and which serve the same purpose. You can even, with a bit
of registry diddling cause Windoze to update more frequently than their
pre-ordained weekly interval although for some reason MS seems to have a
difficult time with this relatively simple task.
There are numerous sites and pages covering the whole subject of time
synchronization and a search will turn them up readily. You can even, if
you feel daring enough set up your own time server synchronized directly
to the national standard and then synchronize other computers on a
network directly to that.
--
John McGaw
[Knoxville, TN, USA]
http://johnmcgaw.com
#3
| |||
| |||
|
|
mar... (AT) hotmail (DOT) com wrote: What is the expected accuracy for the time clock on a PC with Intel Core-2 6300 at 1.86GHz, 32-bit Vista? My Vista synchronizes with an Internet server on Sat at 9pm. I checked on Sat after 9pm and the PC time clock was accurate to every second. On Sunday evening it was 2 seconds fast, today on Monday it is 4 seconds fast, i.e. it seems to gain 2 seconds per day. Does it mean that the latest PC technology still cannot ensure time clock accuracy at least -/+ 1 second per week? Are there any expansion boards available to rectify the problem? What it means is that your PC uses the cheapest possible crystal for its time standard and that has almost nothing to do with "the latest PC technology". It has almost everything to do with saving a penny. As for expansion boards, why bother? If you have a reliable internet connection you can synchronize your clock more frequently and achieve +/- 1 second accuracy very easily. I synchronize my machines once per day using a free program called atomic.exe but there are others out there which are equally free and which serve the same purpose. You can even, with a bit of registry diddling cause Windoze to update more frequently than their pre-ordained weekly interval although for some reason MS seems to have a difficult time with this relatively simple task. There are numerous sites and pages covering the whole subject of time synchronization and a search will turn them up readily. You can even, if you feel daring enough set up your own time server synchronized directly to the national standard and then synchronize other computers on a network directly to that. -- John McGaw [Knoxville, TN, USA]http://johnmcgaw.com |
The problem with frequent synchronizations is that it may lead to
timestamps
that are out of order. For example, if you have a server program
running in the
background 24/7 around the clock that collects real-time events and
saves a
timestamp of each event you may end up with events that are out of
order
when a sync sets the clock back a few seconds. Syncing once per week
Sat/Sun night is fine but not more frequent than that.
Besides, Internet synchronization is rather primitive because the
protocol
is very primitive. You always have to check visually the time after
so called
synching. I would rather do it only once per week.
#4
| |||
| |||
|
|
What is the expected accuracy for the time clock on a PC with Intel Core-2 6300 at 1.86GHz, 32-bit Vista? My Vista synchronizes with an Internet server on Sat at 9pm. I checked on Sat after 9pm and the PC time clock was accurate to every second. On Sunday evening it was 2 seconds fast, today on Monday it is 4 seconds fast, i.e. it seems to gain 2 seconds per day. Does it mean that the latest PC technology still cannot ensure time clock accuracy at least -/+ 1 second per week? Are there any expansion boards available to rectify the problem? I use "Atomic clock sync" from |
It keeps the clock synced to the second.
#5
| |||
| |||
|
|
I use "Atomic clock sync" fromwww.worldtimeserver.com. It keeps the clock synced to the second. |
to the "second", i.e. it may be one second or 5 seconds.
Since the protocol is primitive all syncing programs
are primitive to the "second", i.e. it may be one second or
5 seconds.
The issue here is not syncing but the accuracy of the
PC time clock or any way to use an expansion
board with highly accurate clock so that the syncing
is reduced to a minimum.
#6
| |||
| |||
|
|
On May 14, 12:42 pm, John McGaw <nob... (AT) nowh (DOT) ere> wrote: mar... (AT) hotmail (DOT) com wrote: What is the expected accuracy for the time clock on a PC with Intel Core-2 6300 at 1.86GHz, 32-bit Vista? My Vista synchronizes with an Internet server on Sat at 9pm. I checked on Sat after 9pm and the PC time clock was accurate to every second. On Sunday evening it was 2 seconds fast, today on Monday it is 4 seconds fast, i.e. it seems to gain 2 seconds per day. Does it mean that the latest PC technology still cannot ensure time clock accuracy at least -/+ 1 second per week? Are there any expansion boards available to rectify the problem? What it means is that your PC uses the cheapest possible crystal for its time standard and that has almost nothing to do with "the latest PC technology". It has almost everything to do with saving a penny. As for expansion boards, why bother? If you have a reliable internet connection you can synchronize your clock more frequently and achieve +/- 1 second accuracy very easily. I synchronize my machines once per day using a free program called atomic.exe but there are others out there which are equally free and which serve the same purpose. You can even, with a bit of registry diddling cause Windoze to update more frequently than their pre-ordained weekly interval although for some reason MS seems to have a difficult time with this relatively simple task. There are numerous sites and pages covering the whole subject of time synchronization and a search will turn them up readily. You can even, if you feel daring enough set up your own time server synchronized directly to the national standard and then synchronize other computers on a network directly to that. -- John McGaw [Knoxville, TN, USA]http://johnmcgaw.com The problem with frequent synchronizations is that it may lead to timestamps that are out of order. For example, if you have a server program running in the background 24/7 around the clock that collects real-time events and saves a timestamp of each event you may end up with events that are out of order when a sync sets the clock back a few seconds. Syncing once per week Sat/Sun night is fine but not more frequent than that. Besides, Internet synchronization is rather primitive because the protocol is very primitive. You always have to check visually the time after so called synching. I would rather do it only once per week. |
good old NTP is getting a bit long in the tooth but there are probably a
half-dozen other protocols out there and some are certainly good for
better than 1 second accuracy. NTP4 comes immediately to mind as being
good for 1ms accuracy. And there are other solutions using a combination
of synchronization, drift analysis, and corrections applied many times
per second which can keep clocks across a facility accurate to better
than 100ms even if external reference time is lost for some period.
Simple and relatively inexpensive (inexpensive if one really has need of
such accuracy, that is) GPS synchronization equipment and software can
readily give continuous +/- 50ms accuracy directly traceable to NIST.
And it doesn't even need an add-in card of any sort.
--
John McGaw
[Knoxville, TN, USA]
http://johnmcgaw.com
#7
| |||
| |||
|
|
On May 14, 12:42 pm, John McGaw <nob... (AT) nowh (DOT) ere> wrote: The problem with frequent synchronizations is that it may lead to timestamps that are out of order. For example, if you have a server program running in the background 24/7 around the clock that collects real-time events and saves a timestamp of each event you may end up with events that are out of order when a sync sets the clock back a few seconds. Syncing once per week Sat/Sun night is fine but not more frequent than that. |
It seems like you're trying to fix something that isn't a
problem.
#8
| |||
| |||
|
|
The issue here is not syncing but the accuracy of the PC time clock or any way to use an expansion board with highly accurate clock so that the syncing is reduced to a minimum. |
Then buy some high grade crystal somewhere and replace the
one on the board.
#9
| |||
| |||
|
|
What is the expected accuracy for the time clock on a PC with Intel Core-2 6300 at 1.86GHz, 32-bit Vista? My Vista synchronizes with an Internet server on Sat at 9pm. I checked on Sat after 9pm and the PC time clock was accurate to every second. On Sunday evening it was 2 seconds fast, today on Monday it is 4 seconds fast, i.e. it seems to gain 2 seconds per day. Does it mean that the latest PC technology still cannot ensure time clock accuracy at least -/+ 1 second per week? Are there any expansion boards available to rectify the problem? |
No, it means the delay between you and the timeserver over the network
is variable because obviously you don't have a dedicated line over which
only your traffic is allowed. I sync my clock once per hour and I use
whatever is the fastest responding timeserver from a long list of them.
That means that I do NOT use the time service in Windows since you can
only specify a couple timeservers there. The geographically closest
timeserver is not necessarily the fastest over the Internet (you could
have one in town a mile away but your connection goes off on the ISP's
backbone to a major hub, like Chicago, and then bounces back). You have
no control over the path taken to a timeserver or any server. You have
no control over the saturations of those interconnected networks.
I use Socketwatch but there are probably others that can also poll
around trying to find whatever is the then currently fastest *accessed*
timeserver from your host point. My local university provides 2 NTP
servers. Yet they are not always the fastest (least delay) to get the
current time. I might end up using one at Purdue or at a university in
Germany. All depends which is fastest to respond at the time of the
poll.
You could repeatedly resync using the same server over a network and
notice that each one varies in its offset from before. You are
synchronizing over a non-dedicated network. If you want absolute
accuracy that is repeatable then you need to not sync across a TCP
network and instead somehow hook into the radio transmission. Control
towers for airports do NOT time sync over the Internet, nor anyone else
that needs repeatable results on every time check.
So just what operation are you performing in real-time where a
difference of, say, 2 seconds between various hosts in your intranet
will cause severe distress in results?
#10
| |||
| |||
|
|
maruk2 wrote: The issue here is not syncing but the accuracy of the PC time clock or any way to use an expansion board with highly accurate clock so that the syncing is reduced to a minimum. Then buy some high grade crystal somewhere and replace the one on the board. |
A better crystal and tighter timing circuit probably won't help much.
The OP is synchronizing to an NTP server over the Internet, not using
the radio broadcast.
|
« Previous Thread
|
Next Thread »
| Thread Tools | Search this Thread |
| Display Modes | |
Linear Mode
| |
Powered by vBulletin Version 3.5.3
Copyright ©2000 - 2008, Jelsoft Enterprises Ltd.
Copyright ©2000 - 2008, Jelsoft Enterprises Ltd.