Any way to upgrade CMOS clock??

Hi!

> Thanks for your reply. Daily syncing is out of the
> question for the reason I stated in another reply
> in this thread.

These programs must be making many log entries if they could be thrown
off by a mere two second adjustment.

I have only one more suggestion to offer as a troubleshooting point.
You can try to determine where the time drift is occurring with a
simple BASIC program found here:

http://www.gilanet.com/ohlandl/CPU/MC146818AJ.html

William

> -----Original Message-----
> From: [private.php?do=newpm&u=]
> Posted At: Tuesday, May 15, 2007 12:55 PM
> Posted To: alt.sys.pc-clone.dell
> Conversation: Any way to upgrade CMOS clock??
> Subject: Re: Any way to upgrade CMOS clock??
>
> Hi!
>
> > Thanks for your reply. Daily syncing is out of the
> > question for the reason I stated in another reply
> > in this thread.
>
> These programs must be making many log entries if they could be thrown
> off by a mere two second adjustment.
>
> I have only one more suggestion to offer as a troubleshooting point.
> You can try to determine where the time drift is occurring with a
> simple BASIC program found here:
>
> http://www.gilanet.com/ohlandl/CPU/MC146818AJ.html
>
> William

Relying on the time to be unique is, no offense, bad code. Been bad
code since I started programming in the 70's.

Timecode is a good start, but it needs a sequence number added to it, at
a minimum. If there is a collision, then increment the sequence number
and try again -- until it works. Heck, a clock down to seconds works
properly with this code.

Tom

On May 15, 1:05 pm, "Tom Scales" <tjsca...@gmail.com> wrote:
> > -----Original Message-----
> > From: wm_wa...@hotmail.com [mailto:wm_wa...@hotmail.com]
> > Posted At: Tuesday, May 15, 2007 12:55 PM
> > Posted To: alt.sys.pc-clone.dell
> > Conversation: Any way to upgrade CMOS clock??
> > Subject: Re: Any way to upgrade CMOS clock??
>
> > Hi!
>
> > > Thanks for your reply. Daily syncing is out of the
> > > question for the reason I stated in another reply
> > > in this thread.
>
> > These programs must be making many log entries if they could be thrown
> > off by a mere two second adjustment.
>
> > I have only one more suggestion to offer as a troubleshooting point.
> > You can try to determine where the time drift is occurring with a
> > simple BASIC program found here:
>
> >http://www.gilanet.com/ohlandl/CPU/MC146818AJ.html
>
> > William
>
> Relying on the time to be unique is, no offense, bad code. Been bad
> code since I started programming in the 70's.
>
> Timecode is a good start, but it needs a sequence number added to it, at
> a minimum. If there is a collision, then increment the sequence number
> and try again -- until it works. Heck, a clock down to seconds works
> properly with this code.
>
> Tom- Hide quoted text -
>
> - Show quoted text -

Where did you take the assumption that the time is used to indicate
sequence? Bad programming is based on bad assumptions.
The time is used to indicate time! End of assumptions!

When the time is not reliable and it has to be corrected then the
error handling depends on the amount and nature of the data
affected by the corrections. In order to address the whole issue
you would have to know all the specific details. Every programmer
knows that and stays away from making any assumptions.
You claim to be a programmer and make wild assumptions.

> -----Original Message-----
> From: [private.php?do=newpm&u=]
> Posted At: Tuesday, May 15, 2007 2:04 PM
> Posted To: alt.sys.pc-clone.dell
> Conversation: Any way to upgrade CMOS clock??
> Subject: Re: Any way to upgrade CMOS clock??
>
> On May 15, 1:05 pm, "Tom Scales" <tjsca...@gmail.com> wrote:
> > > -----Original Message-----
> > > From: wm_wa...@hotmail.com [mailto:wm_wa...@hotmail.com]
> > > Posted At: Tuesday, May 15, 2007 12:55 PM
> > > Posted To: alt.sys.pc-clone.dell
> > > Conversation: Any way to upgrade CMOS clock??
> > > Subject: Re: Any way to upgrade CMOS clock??
> >
> > > Hi!
> >
> > > > Thanks for your reply. Daily syncing is out of the
> > > > question for the reason I stated in another reply
> > > > in this thread.
> >
> > > These programs must be making many log entries if they could be
> thrown
> > > off by a mere two second adjustment.
> >
> > > I have only one more suggestion to offer as a troubleshooting
> point.
> > > You can try to determine where the time drift is occurring with a
> > > simple BASIC program found here:
> >
> > >http://www.gilanet.com/ohlandl/CPU/MC146818AJ.html
> >
> > > William
> >
> > Relying on the time to be unique is, no offense, bad code. Been bad
> > code since I started programming in the 70's.
> >
> > Timecode is a good start, but it needs a sequence number added to
it,
> at
> > a minimum. If there is a collision, then increment the sequence
> number
> > and try again -- until it works. Heck, a clock down to seconds
works
> > properly with this code.
> >
> > Tom- Hide quoted text -
> >
> > - Show quoted text -
>
> Where did you take the assumption that the time is used to indicate
> sequence? Bad programming is based on bad assumptions.
> The time is used to indicate time! End of assumptions!
>
> When the time is not reliable and it has to be corrected then the
> error handling depends on the amount and nature of the data
> affected by the corrections. In order to address the whole issue
> you would have to know all the specific details. Every programmer
> knows that and stays away from making any assumptions.
> You claim to be a programmer and make wild assumptions.

Wow, sorry. Don't get your dander up. Time is also an inaccurate
method of determining sequence. For example, at the time change when the
clocks are set back an hour, you have an hour of inaccurate sequence.
Much better ways.

Been a programmer for 27 years.

Well, you are right. We do not know a lot about what you are trying to do,
only the brief bit that you told us. A hardware clock board may be absolutely
the right answer for one or even two computers and someone with the ability to
modify whatever software grabs the time of day from the regular system clock.
Or, as both Tom and I suggested in different ways, correct for the loss of time
ticks in the software that analyzes the data. No matter what, though, there is
no magic bullet solution to motherboard clock with a design that dates back to
the intro of the IBM AT back in 1984 or even 1983. I promise not to make any
assumptions or offer any further suggestions to solve a problem about which I
have been told very little... Ben Myers

On 15 May 2007 11:04:14 -0700, "" <> wrote:

>On May 15, 1:05 pm, "Tom Scales" <tjsca...@gmail.com> wrote:
>> > -----Original Message-----
>> > From: wm_wa...@hotmail.com [mailto:wm_wa...@hotmail.com]
>> > Posted At: Tuesday, May 15, 2007 12:55 PM
>> > Posted To: alt.sys.pc-clone.dell
>> > Conversation: Any way to upgrade CMOS clock??
>> > Subject: Re: Any way to upgrade CMOS clock??
>>
>> > Hi!
>>
>> > > Thanks for your reply. Daily syncing is out of the
>> > > question for the reason I stated in another reply
>> > > in this thread.
>>
>> > These programs must be making many log entries if they could be thrown
>> > off by a mere two second adjustment.
>>
>> > I have only one more suggestion to offer as a troubleshooting point.
>> > You can try to determine where the time drift is occurring with a
>> > simple BASIC program found here:
>>
>> >http://www.gilanet.com/ohlandl/CPU/MC146818AJ.html
>>
>> > William
>>
>> Relying on the time to be unique is, no offense, bad code. Been bad
>> code since I started programming in the 70's.
>>
>> Timecode is a good start, but it needs a sequence number added to it, at
>> a minimum. If there is a collision, then increment the sequence number
>> and try again -- until it works. Heck, a clock down to seconds works
>> properly with this code.
>>
>> Tom- Hide quoted text -
>>
>> - Show quoted text -
>
>Where did you take the assumption that the time is used to indicate
>sequence? Bad programming is based on bad assumptions.
>The time is used to indicate time! End of assumptions!
>
>When the time is not reliable and it has to be corrected then the
>error handling depends on the amount and nature of the data
>affected by the corrections. In order to address the whole issue
>you would have to know all the specific details. Every programmer
>knows that and stays away from making any assumptions.
>You claim to be a programmer and make wild assumptions.
>
>

What he said. Perhaps my sharing my experience came across as
insulting, but I was trying to help. The bigger challenge would be
syncing ALL of the clocks. Do you use a server clock? What about LAN
or even WAN latency? Query the server's clock and the client on the
local lan would get a different answer than the client on the WAN.
Clocks are notoriously unreliable. Computer operators are worse. I
can't tell you how many times I've had to recover from someone
incorrectly setting the clock. What if the CMOS battery starts to die.
You could end up with all of your transactions in 1980.

We don't know enough to help. We would be happy to.

Tom

> -----Original Message-----
> From: Ben Myers [private.php?do=newpm&u=]
> Posted At: Tuesday, May 15, 2007 4:38 PM
> Posted To: alt.sys.pc-clone.dell
> Conversation: Any way to upgrade CMOS clock??
> Subject: Re: Any way to upgrade CMOS clock??
>
> Well, you are right. We do not know a lot about what you are trying
> to do,
> only the brief bit that you told us. A hardware clock board may be
> absolutely
> the right answer for one or even two computers and someone with the
> ability to
> modify whatever software grabs the time of day from the regular system
> clock.
> Or, as both Tom and I suggested in different ways, correct for the
loss
> of time
> ticks in the software that analyzes the data. No matter what,
though,
> there is
> no magic bullet solution to motherboard clock with a design that dates
> back to
> the intro of the IBM AT back in 1984 or even 1983. I promise not to
> make any
> assumptions or offer any further suggestions to solve a problem about
> which I
> have been told very little... Ben Myers
>
> On 15 May 2007 11:04:14 -0700, ""
> <> wrote:
>
> >On May 15, 1:05 pm, "Tom Scales" <tjsca...@gmail.com> wrote:
> >> > -----Original Message-----
> >> > From: wm_wa...@hotmail.com [mailto:wm_wa...@hotmail.com]
> >> > Posted At: Tuesday, May 15, 2007 12:55 PM
> >> > Posted To: alt.sys.pc-clone.dell
> >> > Conversation: Any way to upgrade CMOS clock??
> >> > Subject: Re: Any way to upgrade CMOS clock??
> >>
> >> > Hi!
> >>
> >> > > Thanks for your reply. Daily syncing is out of the
> >> > > question for the reason I stated in another reply
> >> > > in this thread.
> >>
> >> > These programs must be making many log entries if they could be
> thrown
> >> > off by a mere two second adjustment.
> >>
> >> > I have only one more suggestion to offer as a troubleshooting
> point.
> >> > You can try to determine where the time drift is occurring with a
> >> > simple BASIC program found here:
> >>
> >> >http://www.gilanet.com/ohlandl/CPU/MC146818AJ.html
> >>
> >> > William
> >>
> >> Relying on the time to be unique is, no offense, bad code. Been
bad
> >> code since I started programming in the 70's.
> >>
> >> Timecode is a good start, but it needs a sequence number added to
> it, at
> >> a minimum. If there is a collision, then increment the sequence
> number
> >> and try again -- until it works. Heck, a clock down to seconds
> works
> >> properly with this code.
> >>
> >> Tom- Hide quoted text -
> >>
> >> - Show quoted text -
> >
> >Where did you take the assumption that the time is used to indicate
> >sequence? Bad programming is based on bad assumptions.
> >The time is used to indicate time! End of assumptions!
> >
> >When the time is not reliable and it has to be corrected then the
> >error handling depends on the amount and nature of the data
> >affected by the corrections. In order to address the whole issue
> >you would have to know all the specific details. Every programmer
> >knows that and stays away from making any assumptions.
> >You claim to be a programmer and make wild assumptions.
> >
> >

Hi!

> Where did you take the assumption that the time is
> used to indicate sequence?

That's what time is *for*. It gives us a good way to know in what
sequence things should or did happen...not that you probably need such
an explanation, as most of us are quite experienced users of time. :-)

One suggestion I left off in my original post...perhaps you should try
your application on another computer system to see if it keeps better
time.

William

On 15 May 2007 11:04:14 -0700, ""
<> wrote:

>On May 15, 1:05 pm, "Tom Scales" <tjsca...@gmail.com> wrote:
>> > -----Original Message-----
>> > From: wm_wa...@hotmail.com [mailto:wm_wa...@hotmail.com]
>> > Posted At: Tuesday, May 15, 2007 12:55 PM
>> > Posted To: alt.sys.pc-clone.dell
>> > Conversation: Any way to upgrade CMOS clock??
>> > Subject: Re: Any way to upgrade CMOS clock??
>>
>> > Hi!
>>
>> > > Thanks for your reply. Daily syncing is out of the
>> > > question for the reason I stated in another reply
>> > > in this thread.
>>
>> > These programs must be making many log entries if they could be thrown
>> > off by a mere two second adjustment.
>>
>> > I have only one more suggestion to offer as a troubleshooting point.
>> > You can try to determine where the time drift is occurring with a
>> > simple BASIC program found here:
>>
>> >http://www.gilanet.com/ohlandl/CPU/MC146818AJ.html
>>
>> > William
>>
>> Relying on the time to be unique is, no offense, bad code. Been bad
>> code since I started programming in the 70's.
>>
>> Timecode is a good start, but it needs a sequence number added to it, at
>> a minimum. If there is a collision, then increment the sequence number
>> and try again -- until it works. Heck, a clock down to seconds works
>> properly with this code.
>>
>> Tom- Hide quoted text -
>>
>> - Show quoted text -
>
>Where did you take the assumption that the time is used to indicate
>sequence? Bad programming is based on bad assumptions.
>The time is used to indicate time! End of assumptions!
>
>When the time is not reliable and it has to be corrected then the
>error handling depends on the amount and nature of the data
>affected by the corrections. In order to address the whole issue
>you would have to know all the specific details. Every programmer
>knows that and stays away from making any assumptions.
>You claim to be a programmer and make wild assumptions.
>
>

I'm NOT a programmer AND with all due respect to your
application/solution reading all the replies here seems to indicate to
me that more details are needed now to help you further in your
SPECIFIC case. Otherwise it looks to me that we can't help you much
more.

Last FWIW, it just irks me when anyone one of us try to help you and
you reply in anger , ie: Tom Scales. Whether he made an assumption
or not isn't the point; he was trying to help you for free and you
get mad at him. Well you get my drift !!! And FWIW, Tom is one of
the few here who is most knowledgeable so you would be eliminating a
good source of information... just a word to the wise.

<> wrote in message
news: ups.com...
> As the starting point, assume a PC with Intel Core-2 6300
> at 1.86GHz, 32-bit Vista
>
> The CMOS clock (is it still called CMOS clock?) drifts about
> 2 secs/day on my machine. Is there any way to replace it
> with a much more precise clock?

No.


> Are there any expansion
> cards available with high quality clocks to replace
> the default CMOS clock. I do not want the drift to
> exceed 1 second/week.
>

Seems as if those would be CMOS dependent, so my opinion is no.


> Please do not bring up the synchronization issue with
> Internet servers. I want to synchronize only once/week
> at the most and not every day.
>

I wasn't going to bring it up as internet synchronization and Windows time
should have little nothing to do with the RTC of CMOS, which was the
original premise of your request.

Have a nice day.


Stew

Well, you are in luck, they are BACK.

Actually, Bulova "reintroduced" them recently and they are making them
again. It's kind of a unique watch mechanism with a very unique sound
to it.

Your old one might be repairable.


RnR wrote:
> On Tue, 15 May 2007 11:53:23 -0400, Barry Watzman
> <> wrote:
>
>> Two seconds per day is not a big drift. When Bulova introduced the
>> "Accutron" tuning-fork watch in th 1960's, I think that they guaranteed
>> it accurate to 2 minutes per month, and your PC clock is twice that
>
> Funny you mention this as my neighbor was a watch maker who worked for
> them and I got one as a gift. I still have it but it doesn't work
> (battery??).