Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David J Taylor]

"Terje Mathisen" <"terje.mathisen at tmsw.no"> wrote in message 
news:j2sg37-6ja2@ntp.tmsw.no...

[]
My tmsw.dyndns.org S1 server (FreeBSD 8, Garmin 18LVC on the roof) is a 
very old Dell Latitude laptop sitting in my (cold) attic.


The hard drive doesn't need to run very often at all, except for 
flushing updates to the stats files, and the ntp server part has behaved 
quite nicely, including riding out any brownouts we might have had.


Terje


You're lucky to have a very old laptop - none of mine have a serial port. 
But, yes, I can see and agree that it's a low-usage application.


Cheers,
David 


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Terje Mathisen]

David J Taylor wrote:

"George White"  wrote in message

 gives the status of freeBSD on the
EeePC. What I'm not sure about is the long-term supply of replacement
batteries.


Again, batteries wouldn't matter here. But I think it's the router-like
form factor I would prefer, and a serial port is essential.


My tmsw.dyndns.org S1 server (FreeBSD 8, Garmin 18LVC on the roof) is a 
very old Dell Latitude laptop sitting in my (cold) attic.


The hard drive doesn't need to run very often at all, except for 
flushing updates to the stats files, and the ntp server part has behaved 
quite nicely, including riding out any brownouts we might have had.


Terje

--
- 
"almost all programming can be viewed as an exercise in caching"

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David J Taylor]

"George White"  wrote in message 
news:pine.gso.4.64.1001301646190.3...@halifax.chebucto.ns.ca...

[]

I'm not happy about leaving a laptop on 24 hours a day, in any case.


We do it routinely.  Laptops are very commonly used to control
instruments and log data on oceanographic vessels.  One benefit
is that they are immume to the usual power glitches.


Thanks for that, George.  The one time I left a laptop on doing something 
the HD didn't work the next day.  Ever since then I have equated laptop 
and fragile.  Of course, in this case you might not even need an HD.



There is already an ample supply of netbooks with broken screens.

 gives the status of freeBSD on the 
EeePC.  What I'm not sure about is the long-term supply of replacement

batteries.


Again, batteries wouldn't matter here.  But I think it's the router-like 
form factor I would prefer, and a serial port is essential.


Thanks for the information, though.

Cheers,
David 


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[George White]

On Mon, 25 Jan 2010, David J Taylor wrote:

"Richard B. Gilbert"  wrote in message 
news:yeedncqvudbjkcdwnz2dnuvz_g6dn...@giganews.com...

[]
I think that that the closest you can come "off the shelf" would be a 
Laptop.  You are not likely to find one in that price range.


I'm not happy about leaving a laptop on 24 hours a day, in any case.


We do it routinely.  Laptops are very commonly used to control
instruments and log data on oceanographic vessels.  One benefit
is that they are immume to the usual power glitches.

There is already an ample supply of netbooks with broken screens.

 gives the status of freeBSD on the 
EeePC.  What I'm not sure about is the long-term supply of replacement

batteries.

PHK has created GPS clocks using a "single board computer" and a GPS 
receiver.  I don't know what it cost him.  See:

http://phk.freebsd.dk/soekris/pps/


Thanks for your pointer, Richard.  He mentions US $220.  I don't suppose he'd 
want to ship a pre-configured unit, though.  It's certainly close.


Cheers,
David 



--
George White  
189 Parklea Dr., Head of St. Margarets Bay, Nova Scotia  B3Z 2G6

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Terje Mathisen]

David J Taylor wrote:



notjeff: "I just recently setup my sheevaplug as a stratum-1 NTP server."



Thanks, again. However: "The jitter was usually between 1 and 3 ms" is
about a thousand times worse than I was hoping to achieve. Something to
keep an eye on, though.


You need a real hw PPS signal to get down to single-digit us jitter, the 
only alternative is to have an expensive timing source which you can 
query and which includes the hardware to timestamp each request.


From reading the ntpd driver sources I think the Palisade has such a mode?

Terje
--
- 
"almost all programming can be viewed as an exercise in caching"

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David J Taylor]

notjeff: "I just recently setup my sheevaplug as a stratum-1 NTP 
server."




Thanks, again.  However: "The jitter was usually between 1 and 3 ms" is 
about a thousand times worse than I was hoping to achieve.  Something to 
keep an eye on, though.


Cheers,
David 


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[E-Mail Sent to this address will be added to the BlackLists]

David J Taylor wrote:
> BlackList wrote in message
>> 
>
> Thanks for that.  Interesting, as it seems that you /may/
>  be able to get access to a serial I/O (needed for the GPS)
>  according to the diagram here:
> 
>
> I wonder whether anyone has ported NTP to this platform,
>  and what accuracy they have seen?


notjeff: "I just recently setup my sheevaplug as a stratum-1 NTP server."


-- 
E-Mail Sent to this address 
  will be added to the BlackLists.

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[E-Mail Sent to this address will be added to the BlackLists]

David J Taylor wrote:
> BlackLists wrote:
>> 
>
> Thanks for that.  Interesting, as it seems that you /may/
>  be able to get access to a serial I/O (needed for the GPS)
>  according to the diagram here:
>
> 
>
> I wonder whether anyone has ported NTP to this platform,
>  and what accuracy they have seen?

Seems implied by this:


-- 
E-Mail Sent to this address 
  will be added to the BlackLists.

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David Malone]

"David J Taylor" 
 writes:

>Thanks for that.  Interesting, as it seems that you /may/ be able to get 
>access to a serial I/O (needed for the GPS) according to the diagram here:

>  http://www.marvell.com/platforms/Marvell_PlugComputer_DevKit.pdf

>I wonder whether anyone has ported NTP to this platform, and what accuracy 
>they have seen?

I believe FreeBSD has been ported to that platform:

http://wiki.freebsd.org/FreeBSD/arm

Though I'm not sure of the exact status of the port. Warner would
probably know.

David.

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David J Taylor]

"E-Mail Sent to this address will be added to the BlackLists" 
 wrote in message 
news:hjnmt9$q9...@news.eternal-september.org...

[]




Thanks for that.  Interesting, as it seems that you /may/ be able to get 
access to a serial I/O (needed for the GPS) according to the diagram here:


 http://www.marvell.com/platforms/Marvell_PlugComputer_DevKit.pdf

I wonder whether anyone has ported NTP to this platform, and what accuracy 
they have seen?


David 


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[E-Mail Sent to this address will be added to the BlackLists]

David J Taylor wrote:
>>> 
>>> 
>>
>> 
>
> Thanks, most interesting - I knew about neither of those products.
>
> http://www.veracityuk.com/products/timenet/timenet.php
>  "Accuracy: Ethernet NTP ±100ms overall" is not what I
>   call "good timekeeping".
>
> http://www.gpsntp.com/economic-ntpserver/
>  "NTP time stamp resolution: +/- 15 usec" - nice that it
<   has SNMP, but no specification of accuracy!
>
> You can't buy directly from either site, which discourages
>  an impulse purchase.
>
> I already have the GPS, and was looking for a simple computer
>  (small and low-powered) where I could run a FreeBSD or
>  similar system.  Will see what emerges from the router
>  or David Lord's approaches.



-- 
E-Mail Sent to this address 
  will be added to the BlackLists.

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David Lord]

David J Taylor wrote:








Thanks, most interesting - I knew about neither of those products.

http://www.veracityuk.com/products/timenet/timenet.php
 "Accuracy: Ethernet NTP ±100ms overall" is not what I call "good 
timekeeping".


http://www.gpsntp.com/economic-ntpserver/
 "NTP time stamp resolution: +/- 15 usec" - nice that it has SNMP, but 
no specification of accuracy!


You can't buy directly from either site, which discourages an impulse 
purchase.


I already have the GPS, and was looking for a simple computer (small and 
low-powered) where I could run a FreeBSD or similar system.  Will see 
what emerges from the router or David Lord's approaches.


I've moved away from using ntp over ethernet as main method
for keeping systems in sync. Network and system load along
with temperature variations cause relative havoc which I'll
see if I can cure using pps from the radioclock. It could
be possible to stabilise system clock oscillators but I'm
not sure if that is effective or practical on modern pcs and
it really involves too much work vs pps via parallel port
(last system I purchased has neither serial nor parallel).

The ADM5120P may become base for generating a pps signal to be
distributed around network via rs422 and possibly 433MHz tx/rx
if that doesn't add too much variation. I'm not sure my
programming skills are up to this though.

I also have a 486dx with a very stable (at least long term)
crystal but lacking ram (I still have another 3x 486 but only
40MB ram total to split between four pcs).

My order for some D25 connectors from CPC turned up on Friday,
D25 hoods + all the extra items I'd added to take value up to
give free delivery, but stock discrepancy and no D25 bodies.
Anyway I had a search through my rubbish last night and found
a couple of connectors.

Meanwhile due to cold weather I've had heating turned up and
server using Conrad MSF receiver has had periods of much
reduced offsets (< 300us), so suspicion that the ttl out from
Conrad is on borderline for the rs232 on that server seems
confirmed and I'll give parallel port method a try on a
different system later this week.

David

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Martin Burnicki]

Evandro Menezes wrote:
> On Jan 25, 8:15 am, Martin Burnicki 
> wrote:
>>
>> Yes, it helps indeed ;-)
> 
> I'm glad it does.
> 
> And just for completion, the current generation of AMD processors
> derive the CPU clock from the memory controller clock instead of the
> FSB.  Moreover, the memory controller clock may also be changed on the
> fly and is thus subject to power management policies too, effectively
> making its TSC not invariant if the memory controller power is
> managed.  However, I don't know of any OS that does this yet.

Argh, I just began to hope things could become easier in the future. I'm
sure the OS maintainers will find a way to fiddle with this to degrade
timekeeping ;-)

Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David J Taylor]

Thanks, most interesting - I knew about neither of those products.

http://www.veracityuk.com/products/timenet/timenet.php
 "Accuracy: Ethernet NTP ±100ms overall" is not what I call "good 
timekeeping".


http://www.gpsntp.com/economic-ntpserver/
 "NTP time stamp resolution: +/- 15 usec" - nice that it has SNMP, but no 
specification of accuracy!


You can't buy directly from either site, which discourages an impulse 
purchase.


I already have the GPS, and was looking for a simple computer (small and 
low-powered) where I could run a FreeBSD or similar system.  Will see what 
emerges from the router or David Lord's approaches.


Cheers,
David 


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[E-Mail Sent to this address will be added to the BlackLists]

BlackLists wrote:
> David J Taylor wrote:
>> I would still very much like to have a low-powered (watts)
>>  system running NTP perhaps with a good (for timekeeping)
>>  FreeBSD version.  Something the size of a home router,
>>  with a serial port for the GPS.  Looking for better
>>  than (say) ten microsecond accuracy.  About US $100-150.
>
> 
> 



-- 
E-Mail Sent to this address 
  will be added to the BlackLists.

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[E-Mail Sent to this address will be added to the BlackLists]

David J Taylor wrote:
> I would still very much like to have a low-powered (watts)
>  system running NTP perhaps with a good (for timekeeping)
>  FreeBSD version.  Something the size of a home router,
>  with a serial port for the GPS.  Looking for better
>  than (say) ten microsecond accuracy.  About US $100-150.




-- 
E-Mail Sent to this address 
  will be added to the BlackLists.

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Richard B. Gilbert]

David J Taylor wrote:
"Richard B. Gilbert"  wrote in message 
news:yeedncqvudbjkcdwnz2dnuvz_g6dn...@giganews.com...

[]
I think that that the closest you can come "off the shelf" would be a 
Laptop.  You are not likely to find one in that price range.


I'm not happy about leaving a laptop on 24 hours a day, in any case.

PHK has created GPS clocks using a "single board computer" and a GPS 
receiver.  I don't know what it cost him.  See:

http://phk.freebsd.dk/soekris/pps/


Thanks for your pointer, Richard.  He mentions US $220.  I don't suppose 
he'd want to ship a pre-configured unit, though.  It's certainly close.


Cheers,
David


The only way to be sure he will or will not sell a pre-configured unit 
is to ask him.


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Evandro Menezes]

On Jan 25, 8:15 am, Martin Burnicki 
wrote:
>
> Yes, it helps indeed ;-)

I'm glad it does.

And just for completion, the current generation of AMD processors
derive the CPU clock from the memory controller clock instead of the
FSB.  Moreover, the memory controller clock may also be changed on the
fly and is thus subject to power management policies too, effectively
making its TSC not invariant if the memory controller power is
managed.  However, I don't know of any OS that does this yet.

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David J Taylor]

"Richard B. Gilbert"  wrote in message 
news:yeedncqvudbjkcdwnz2dnuvz_g6dn...@giganews.com...

[]
I think that that the closest you can come "off the shelf" would be a 
Laptop.  You are not likely to find one in that price range.


I'm not happy about leaving a laptop on 24 hours a day, in any case.

PHK has created GPS clocks using a "single board computer" and a GPS 
receiver.  I don't know what it cost him.  See:

http://phk.freebsd.dk/soekris/pps/


Thanks for your pointer, Richard.  He mentions US $220.  I don't suppose 
he'd want to ship a pre-configured unit, though.  It's certainly close.


Cheers,
David 


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Richard B. Gilbert]

David J Taylor wrote:
"Evandro Menezes"  wrote in message 
news:b3618b89-24cd-44a9-8e25-78887d18e...@f12g2000yqn.googlegroups.com...

On Jan 21, 2:52 pm, "David J Taylor"  wrote:


BTW: I still have systems running which are older than Core2, one 
still has

an Intel PIII 550MHz (and keeps good time as a stratum-1 server with
Windows 2000).


I guess that it would qualify for a low-power processor these days, so
who cares about managing its power and messing with its TSC?

;-)


Low-power in what sense?  Work done, or watts consumed?  

I would still very much like to have a low-powered (watts) system 
running NTP perhaps with a good (for timekeeping) FreeBSD version.  
Something the size of a home router, with a serial port for the GPS.  
Looking for better than (say) ten microsecond accuracy.  About US $100-150.


Cheers,
David


I think that that the closest you can come "off the shelf" would be a 
Laptop.  You are not likely to find one in that price range.


PHK has created GPS clocks using a "single board computer" and a GPS 
receiver.  I don't know what it cost him.  See:

http://phk.freebsd.dk/soekris/pps/

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Martin Burnicki]

Evandro Menezes wrote:
> On Jan 21, 10:42 am, Martin Burnicki 
> wrote:
>> Ah, that's interesting. I did know that new CPU cores may not suffer from
>> switched CPU clocks anymore, but I didn't know this is because they are
>> driven by the FSB clock. So I assume the QPC clock frequency reported by
>> Windows can also correspond to the FSB clock.
> 
> Only indirectly.  You certainly know that the CPU clock is a multiple
> of the FSB clock.  The unit of the result returned by RDTSC is still
> CPU clock ticks, always.  So, when the CPU clock multiplier is changed
> from, say, 3.5 to 1.0, due to a power management decision, the TSC
> circuitry is changed accordingly, so that its unit is the same as the
> CPU's.  The result is that when measuring the CPU clock, one will
> still get it right.
> 
> However, the precision of the TSC is not 1 CPU clock tick anymore, but
> the FSB multiplier.  So it cannot be used so easily to measure how
> many CPU clock cycles a sequence of instructions takes anymore.
> AFAIK, on AMD processors it's still possible to chose between variant
> and invariant TSC.  But I think that the tendency is for BIOS makers
> to not offer this option and just enable the invariant TSC, since only
> developers care about its precision (and they can use a performance
> counter for the same purpose).
> 
> HTH

Yes, it helps indeed ;-)

Interesting details I didn't know, yet.

Thanks,
Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Uwe Klein]

David J Taylor wrote:
"Evandro Menezes"  wrote in message 
news:b3618b89-24cd-44a9-8e25-78887d18e...@f12g2000yqn.googlegroups.com...



On Jan 21, 2:52 pm, "David J Taylor"  wrote:



BTW: I still have systems running which are older than Core2, one 
still has

an Intel PIII 550MHz (and keeps good time as a stratum-1 server with
Windows 2000).



I guess that it would qualify for a low-power processor these days, so
who cares about managing its power and messing with its TSC?

;-)



Low-power in what sense?  Work done, or watts consumed?  

I would still very much like to have a low-powered (watts) system 
running NTP perhaps with a good (for timekeeping) FreeBSD version.  
Something the size of a home router, with a serial port for the GPS.  
Looking for better than (say) ten microsecond accuracy.  About US $100-150.



Unpack and Work or Fiddle a Bit ?

For Fiddle a Bit:
There is a wide spectrum of hardware available.
Take any of the Low Cost Thin clients ( Linux on ARM )
Take any of the "InternetRadio" sets ( same, Linux on ARM )

Take any of the Low Cost Router/WlanAccessPoint Hardware that
can have OpenWRT or similar installed. ( Linux on usually ARM )
http://en.wikipedia.org/wiki/List_of_router_or_firewall_distributions
Some even have ntpd in their package repository ;-)

uwe

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David J Taylor]

"Evandro Menezes"  wrote in message 
news:b3618b89-24cd-44a9-8e25-78887d18e...@f12g2000yqn.googlegroups.com...

On Jan 21, 2:52 pm, "David J Taylor"  wrote:


BTW: I still have systems running which are older than Core2, one still 
has

an Intel PIII 550MHz (and keeps good time as a stratum-1 server with
Windows 2000).


I guess that it would qualify for a low-power processor these days, so
who cares about managing its power and messing with its TSC?

;-)


Low-power in what sense?  Work done, or watts consumed?  

I would still very much like to have a low-powered (watts) system running 
NTP perhaps with a good (for timekeeping) FreeBSD version.  Something the 
size of a home router, with a serial port for the GPS.  Looking for better 
than (say) ten microsecond accuracy.  About US $100-150.


Cheers,
David 


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Evandro Menezes]

On Jan 21, 10:42 am, Martin Burnicki 
wrote:
> Ah, that's interesting. I did know that new CPU cores may not suffer from
> switched CPU clocks anymore, but I didn't know this is because they are
> driven by the FSB clock. So I assume the QPC clock frequency reported by
> Windows can also correspond to the FSB clock.

Only indirectly.  You certainly know that the CPU clock is a multiple
of the FSB clock.  The unit of the result returned by RDTSC is still
CPU clock ticks, always.  So, when the CPU clock multiplier is changed
from, say, 3.5 to 1.0, due to a power management decision, the TSC
circuitry is changed accordingly, so that its unit is the same as the
CPU's.  The result is that when measuring the CPU clock, one will
still get it right.

However, the precision of the TSC is not 1 CPU clock tick anymore, but
the FSB multiplier.  So it cannot be used so easily to measure how
many CPU clock cycles a sequence of instructions takes anymore.
AFAIK, on AMD processors it's still possible to chose between variant
and invariant TSC.  But I think that the tendency is for BIOS makers
to not offer this option and just enable the invariant TSC, since only
developers care about its precision (and they can use a performance
counter for the same purpose).

HTH

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Evandro Menezes]

On Jan 21, 2:52 pm, "David J Taylor"  wrote:
>
> BTW: I still have systems running which are older than Core2, on still has
> an Intel PIII 550MHz (and keeps good time as a stratum-1 server with
> Windows 2000).

I guess that it would qualify for a low-power processor these days, so
who cares about managing its power and messing with its TSC?

;-)

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Martin Burnicki]

Terje Mathisen <"terje.mathisen at tmsw.no"> wrote:
> Martin Burnicki wrote:
>> Alan wrote:
>>> However when I turn -M on the bottom line changes to Current 0.977 ms
>>>
>>> Is that really correct as even I can tell 0.977 is less than the
>>> "minimun" 1.000 ms" in the previous line?
>>
>> I think 1 ms is just the nominal value and 0.977 is due to rounding
>> errors or an inexact measurement interval.
> 
> Timer ticks on most versions of Win* are derived from the CMOS clock
> chip which can generate interrupts at any power of two rate, from 1 Hz
> to 32 KHz.

Ah, I know the features of the CMOS clock chip but I haven't been aware this
chip is used to generate the timer tick IRQs.

> 1024 Hz corresponds to ~977 us.

Yep, and 64 Hz corresponds to 15.625 ms, the default system time increment
up to Windows XP.

Under Vista the clockres tool from sysinternals shows:

Maximum timer interval: 15.600 ms
Minimum timer interval: 0.500 ms
Current timer interval: 1.000 ms

So Vista seems to use a different timer for the timer tick, maybe the HPET
which is also used for QPC under Vista?

Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Terje Mathisen]

Martin Burnicki wrote:

Alan wrote:

However when I turn -M on the bottom line changes to Current 0.977 ms

Is that really correct as even I can tell 0.977 is less than the
"minimun" 1.000 ms" in the previous line?


I think 1 ms is just the nominal value and 0.977 is due to rounding errors
or an inexact measurement interval.


Timer ticks on most versions of Win* are derived from the CMOS clock 
chip which can generate interrupts at any power of two rate, from 1 Hz 
to 32 KHz.


1024 Hz corresponds to ~977 us.

Terje

--
- 
"almost all programming can be viewed as an exercise in caching"

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David J Taylor]

Windows chooses the most precise, invariant frequency source available
in the system.  As I said before, Intel's Core2 processors have a TSC
that doesn't vary with the CPU clock (though its precision is not as
good as it seems).  Only newer AMD processors sport the same feature
and even then it must be enabled by a BIOS option that's not always
available.  Therefore, the likelihood of an AMD system using the PM
timer is greater than new Intel systems.

In these days, it's moot to worry about the precision of the
QueryPerformanceCounter function.  Windows will use the most precise
time source in the system and only very old systems would not have at
least the PM timer.

HTH


Evandro, thanks for the details.  I've never worried about the precision 
of the QueryPerformanceCounter function in my own software, I'm just 
reporting the values as a clue to why we see NTP behaving as it does.


BTW: I still have systems running which are older than Core2, on still has 
an Intel PIII 550MHz (and keeps good time as a stratum-1 server with 
Windows 2000).


Cheers,
David 


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Rob]

Rob  wrote:
> Martin Burnicki  wrote:
>>> Is that really correct as even I can tell 0.977 is less than the
>>> "minimun" 1.000 ms" in the previous line?
>>
>> I think 1 ms is just the nominal value and 0.977 is due to rounding errors
>> or an inexact measurement interval.
>
> I thought the timer interval was 1/1024 Hz?
  ^^ ms, of course.

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Rob]

Martin Burnicki  wrote:
>> Is that really correct as even I can tell 0.977 is less than the
>> "minimun" 1.000 ms" in the previous line?
>
> I think 1 ms is just the nominal value and 0.977 is due to rounding errors
> or an inexact measurement interval.

I thought the timer interval was 1/1024 Hz?

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Martin Burnicki]

Evandro Menezes wrote:
> On Jan 21, 4:43 am, Martin Burnicki 
> wrote:
>> AFAIK this should only affect the TSCs, so the setting should not matter
>> at all if the PM timer is used.
> 
> Another detail on RDTSC: newer cores, such as Intel's Core2 and AMD's
> Phenom, return a constant-rate readout by RDTSC.  In actuality, the
> counter that RDTSC is driven by the FSB clock instead of by the CPU
> clock and, even if the later changes due to power management, it's
> adjusted accordingly.

Ah, that's interesting. I did know that new CPU cores may not suffer from
switched CPU clocks anymore, but I didn't know this is because they are
driven by the FSB clock. So I assume the QPC clock frequency reported by
Windows can also correspond to the FSB clock.

> But it all boils down to the HAL, which will use whatever's necessary
> to make sure that what QueryPerformanceCounter returns is independent
> of the CPU clock.

There may still be limitations depending on the age of specific items, e.g.
the HAL can only decide which timer to use if the pros and cons of the
timers have been known to the programmers, and appropriate code has been
integrated into the HAL, when the specific HAL version was released.

Anyway, thanks for the hints.


Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Evandro Menezes]

On Jan 21, 9:51 am, "David J Taylor"  wrote:
>
> Thanks, Martin.  I'm beginning to wish I'd never asked, as I have Windows
> XP SP3 here and yet the performance counter is running at 2.4GHz (Intel
> E6600 dual-core).  Or does the PM_Timer only apply to AMD systems?

Windows chooses the most precise, invariant frequency source available
in the system.  As I said before, Intel's Core2 processors have a TSC
that doesn't vary with the CPU clock (though its precision is not as
good as it seems).  Only newer AMD processors sport the same feature
and even then it must be enabled by a BIOS option that's not always
available.  Therefore, the likelihood of an AMD system using the PM
timer is greater than new Intel systems.

In these days, it's moot to worry about the precision of the
QueryPerformanceCounter function.  Windows will use the most precise
time source in the system and only very old systems would not have at
least the PM timer.

HTH

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David J Taylor]

"Martin Burnicki"  wrote in message 
news:mp2m27-dp3@gateway.py.meinberg.de...

[]
AFAIK if the TSC is used then the clock frequency reported for QPC 
matches

the CPU's clock full clock frequency, and IIRC then according to some MS
docs the reported frequency does not even change when the CPU's clock is
decreased for power saving. IMO that also wouldn't make much sense.

From what I've seen the reported QPC clock frequency depends on which 
timer

circuit is being used. Some months ago I've already posted what I had
found:
http://lists.ntp.org/pipermail/questions/2009-March/022159.html

But of course I can not *guarantee* this is always correct ;-)

In that article there are also some hints which Windows versions / 
service
packs use which timer to implement QPC. I've just reread that article 
and

found this perfectly matches what Alan has observed, i.e. Windows XP SP3
should use the PM timer even if no /usepmtimer flag has been added in
Windows' boot.ini.

Of course it does not explain the problems Alan is observing.


Regards,

Martin



Thanks, Martin.  I'm beginning to wish I'd never asked, as I have Windows 
XP SP3 here and yet the performance counter is running at 2.4GHz (Intel 
E6600 dual-core).  Or does the PM_Timer only apply to AMD systems?


The AMD Vista and Windows-7 desktop systems show 3.579...MHz, an Intel 
Windows-7 portable 1.757MHz (PIT?), an Intel Vista portable 14.3MHz 
(HPET), and an Intel Windows Vista Desktop 14.3MHz.


All frequencies you listed, and nothing which helps Alan.  Seemed to have 
a peculiar timer resolutions though - I never seen 3.906 ms or 1.953 ms. 
What software is setting those?


Cheers,
David 


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Evandro Menezes]

On Jan 21, 4:43 am, Martin Burnicki 
wrote:
> AFAIK this should only affect the TSCs, so the setting should not matter at
> all if the PM timer is used.

Another detail on RDTSC: newer cores, such as Intel's Core2 and AMD's
Phenom, return a constant-rate readout by RDTSC.  In actuality, the
counter that RDTSC is driven by the FSB clock instead of by the CPU
clock and, even if the later changes due to power management, it's
adjusted accordingly.

But it all boils down to the HAL, which will use whatever's necessary
to make sure that what QueryPerformanceCounter returns is independent
of the CPU clock.

HTH

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Martin Burnicki]

Alan wrote:
> I'll try a few experiments. Motherboard is ASRock ALiveNF6p-VSTA, a
> fairly mainstream manufacturer board with NVIDIA Chipset and onboard
> graphics. That or very similar architecture boards are quite widely in
> use.
> 
> I do have a copy of OpenSolaris I stuck on another partition to play
> with some time ago so I can install NTPD on it and watch what happens
> when I get a chance.

Remember whether timekeeping works correctly depends at least on:

- which timer circuit is being used
- whether that timer hardware works correctly or not
- whether that timer is handled correctly by the OS, or not

so it's possible OpenSolaris is a perfect timekeeper on that board.

> Might also try a fresh install of Windows on 
> another parition to see if I still see the problem.
> 
> In the meantime what I've noticed is that if I fire up most multimedia
> apps (web browser plugins for example) then the timer resolution gets
> set to 1.953 ms or 3.906 ms and in this case NTPD DOES MANAGE TO SYNCH
> THE TIME!! (although it drifts a bit before resynching) It seems that it
> is only when the timer interval drops to 0.977 ms either set by NTPD or
> something else (Windows Media player sets it to this for example) that
> we enter a time-warp.
> 
> So in  summary
> 
> Current timer interval: 15.625 ms - No problem
> Current timer interval   3.906 ms - No problem
> Current timer interval   1.953 ms - No problem
> Current timer interval   0.977 ms - Wild time drift

AFAIK the MM timer resolution can be set in 1 ms steps, so the nominal
values for what you've observed should be 4 ms, 2 ms, and 1 ms.

I have not yet had a closer look at the clockres tool from sysinternals
mentioned by Evandro Menezes, but that program just seems to count the MM
timer callbacks during a 15.625 ms system time interval, or vice-versa. See
this computation:

15.625 ms / 16 = 0.97656 ms

clockres shows the same 0.977 ms interval on a Win XP SP3 system here when
ntpd is running with -M. This system has a AMD Athlon 64 X2 4400+ CPU and
ntpd is working fine.

Here are a few thoughts I've also already posted some time ago, see:
http://lists.ntp.org/pipermail/questions/2009-December/025210.html

> Even if under Vista/Windows 7 the system time increments in 1 ms steps, 
> the nominal standard tick count is still ~15600 (15601 on a Vista machine
> here), i.e ~15.6 ms. Since this is not an integral multiple of 1 ms there
> must be some math which converts from 1 ms steps to 15.6 ms steps, and
> that math may suffer from rounding errors.
> 
> AFAICS this is still the basic problem as under XP or earlier, when the MM
> timer has been set: The MM timer ticks at 1 ms, but the system time ticks
> at 15.625 ms, and there also needs to be a conversion from one tick rate
> to the other.
> 
> The difference in Vista/7 vs. 2000/XP seems to be that
> GetSystemTimeAsFiletime returns values from the 1 ms "tick domain" for the
> newer systems whereas it returns values from the 15 ms "tick domain" on
> older systems.

Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Dave Hart]

On Thu, Jan 21, 2010 at 11:01 UTC, Martin Burnicki wrote:
> Dave Hart, isn't there a way to force disabling time interpolation with your
> binaries even if the system time increments in 15.625 ms steps?

No, the opposite is available (forcing its use on), but the only way
to disable interpolation is to use ntpd -M on Windows Vista or newer,
where after cranking the multimedia timer to highest resolution, ntpd
will notice the system clock is stepping in less than 4ms increments
and disable interpolation:

if (os_clock_precision < 4 * 1 && !getenv("NTPD_USE_INTERP_DANGEROUS")) {
msyslog(LOG_INFO, "using Windows clock directly");
} else {
winnt_use_interpolation = TRUE;

It is interesting to me that the brokenness is the same on 4.2.4p8
(which has the old Windows interpolation code) and 4.2.6 (which has
the new interpolation introduced around 4.2.5p162).  Both codebases
are trying to do the same thing, maintain a mapping between the
performance counter timeline and the system clock timeline.  Since
both are equally hosed, I am relieved to believe it's not indicating a
problem with the new interpolation code.

Although it's a longshot, you can try forcing ntpd to use the
processor TSC instead of QueryPerformanceCounter.  To do this, you
need to determine your processor frequency very accurately (and I
can't point to a good tool for that offhand).  Then add --pccfreq=X
where X is your processor frequency in Hz.  For a nominally 400MHz
space heater of mine, the magic value is --pccfreq=398125000.  If you
don't get this within a few PPM of the correct figure, expect ntpd to
go wild immediately.  If your TSC rate (CPU frequency essentially)
varies over time, it will break this config.  When using the TSC (PCC)
instead of QueryPerformanceCounter, ntpd 4.2.6 will automatically lock
thread affinity to a single processor for threads which use the
counter.

Cheers,
Dave Hart
___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Martin Burnicki]

David J Taylor wrote:
> Is:
> 
>   Performance Counter Frequency: 3.580
> 
> a guarantee that the TSC isn't being used?

AFAIK if the TSC is used then the clock frequency reported for QPC matches
the CPU's clock full clock frequency, and IIRC then according to some MS
docs the reported frequency does not even change when the CPU's clock is
decreased for power saving. IMO that also wouldn't make much sense.

>From what I've seen the reported QPC clock frequency depends on which timer
circuit is being used. Some months ago I've already posted what I had
found:
http://lists.ntp.org/pipermail/questions/2009-March/022159.html

But of course I can not *guarantee* this is always correct ;-)

In that article there are also some hints which Windows versions / service
packs use which timer to implement QPC. I've just reread that article and
found this perfectly matches what Alan has observed, i.e. Windows XP SP3
should use the PM timer even if no /usepmtimer flag has been added in
Windows' boot.ini.

Of course it does not explain the problems Alan is observing.


Regards,

Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Alan]

I'll try a few experiments. Motherboard is ASRock ALiveNF6p-VSTA, a 
fairly mainstream manufacturer board with NVIDIA Chipset and onboard 
graphics. That or very similar architecture boards are quite widely in use.


I do have a copy of OpenSolaris I stuck on another partition to play 
with some time ago so I can install NTPD on it and watch what happens 
when I get a chance. Might also try a fresh install of Windows on 
another parition to see if I still see the problem.


In the meantime what I've noticed is that if I fire up most multimedia 
apps (web browser plugins for example) then the timer resolution gets 
set to 1.953 ms or 3.906 ms and in this case NTPD DOES MANAGE TO SYNCH 
THE TIME!! (although it drifts a bit before resynching) It seems that it 
is only when the timer interval drops to 0.977 ms either set by NTPD or 
something else (Windows Media player sets it to this for example) that 
we enter a time-warp.


So in  summary

Current timer interval: 15.625 ms - No problem
Current timer interval   3.906 ms - No problem
Current timer interval   1.953 ms - No problem
Current timer interval   0.977 ms - Wild time drift

More digging when I get a chance.

Martin Burnicki wrote:

David J Taylor wrote:

"Alan"  wrote in message
news:z9i5n.58805$q36.5...@newsfe19.ams2...

Would like to get to the bottom of this as well. Using 4.4.6-o with -M ,
I get "Frequency error 3030 PPM exceeds tolerance 500 PPM". Considering
that without =M the frequency modification is only about 5 PPM then
obvioulsy something very strange is going on. It seems that even if the
timer precision is increased by another program then time immediately
starts to drift rapidly by hundreds of milliseconds.


Yes, looks like timing is seriously broken on your hardware.

Dave Hart, isn't there a way to force disabling time interpolation with your
binaries even if the system time increments in 15.625 ms steps?
Maybe this could be wort a try in this special case.


Alan,

Our experience was that the switching between normal and high-resolution
timers caused steps of many milliseconds (I don't recall the exact figure)
which really messed up NTP.  So either run with no MM timers at all, or
run with the MM timers permanently enabled, and NTP recognises that
change, and adjusts accordingly.  Have NTP start the MM timers was one
solution, and hence the -M option.

It might be helpful to know what the event log says with both sets of
startup parameters, as there may be a clue there which Dave Hart, the
person closest to this code, can interpret.

I must confess to having nagging doubts about AMD (but with no good
reason),


AFAIK the CPU type (Intel or AMD, CPU family ...) should not matter if the
PM timer is used for QPC instead of the TSCs.

However, as I've mentioned in a different reply, the problem may be due to a
fault chipset.

The Linux kernel identifies quite a number of problematic hardware and
displays appropriate warnings at startup, so booting a current Linux system
(maybe from a Live CD) on that machine and watching the startup messages
*may* give some hints.

about whether you have another program setting the time (check 
that w32time.exe is not running - Show Processes from all users), and

perhaps something in the BIOS.  One final idea (which there was no option
on my test system) might be to start with just one CPU active in the BIOS.


I doubt the problem may be due to a different time sync program running
since the problem occurs if and only if the MM timer tick rate is changed.

Martin


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David J Taylor]

I doubt the problem may be due to a different time sync program running
since the problem occurs if and only if the MM timer tick rate is 
changed.


Martin


Agreed, Martin.  Just clutching at straws with all the things I have seen 
which can cause problems on Windows.  Thanks also for your notes on the 
different processors and timers.  Is:


 Performance Counter Frequency: 3.580

a guarantee that the TSC isn't being used?

Cheers,
David 


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Martin Burnicki]

David J Taylor wrote:
> "Alan"  wrote in message
> news:z9i5n.58805$q36.5...@newsfe19.ams2...
>> Would like to get to the bottom of this as well. Using 4.4.6-o with -M ,
>> I get "Frequency error 3030 PPM exceeds tolerance 500 PPM". Considering
>> that without =M the frequency modification is only about 5 PPM then
>> obvioulsy something very strange is going on. It seems that even if the
>> timer precision is increased by another program then time immediately
>> starts to drift rapidly by hundreds of milliseconds.

Yes, looks like timing is seriously broken on your hardware.

Dave Hart, isn't there a way to force disabling time interpolation with your
binaries even if the system time increments in 15.625 ms steps?
Maybe this could be wort a try in this special case.

> Alan,
> 
> Our experience was that the switching between normal and high-resolution
> timers caused steps of many milliseconds (I don't recall the exact figure)
> which really messed up NTP.  So either run with no MM timers at all, or
> run with the MM timers permanently enabled, and NTP recognises that
> change, and adjusts accordingly.  Have NTP start the MM timers was one
> solution, and hence the -M option.
> 
> It might be helpful to know what the event log says with both sets of
> startup parameters, as there may be a clue there which Dave Hart, the
> person closest to this code, can interpret.
> 
> I must confess to having nagging doubts about AMD (but with no good
> reason),

AFAIK the CPU type (Intel or AMD, CPU family ...) should not matter if the
PM timer is used for QPC instead of the TSCs.

However, as I've mentioned in a different reply, the problem may be due to a
fault chipset.

The Linux kernel identifies quite a number of problematic hardware and
displays appropriate warnings at startup, so booting a current Linux system
(maybe from a Live CD) on that machine and watching the startup messages
*may* give some hints.

> about whether you have another program setting the time (check 
> that w32time.exe is not running - Show Processes from all users), and
> perhaps something in the BIOS.  One final idea (which there was no option
> on my test system) might be to start with just one CPU active in the BIOS.

I doubt the problem may be due to a different time sync program running
since the problem occurs if and only if the MM timer tick rate is changed.

Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Martin Burnicki]

Alan wrote:
> Here's what I see from clockres without the -M option on
> 
> Max 15.625 ms
> Minimum 1.000 ms
> Current 15.625 ms

This seems to indicate the internal timer ticks at 15.625 ms interval by
default, but if the MM timer is set to highest resolution the timer tick
changes to 1 ms.

However, under Windows up to XP/Server 2003 the system time returned by the
GetSystemTimeAsFileTime call still increments in 15.625 ms steps even if
the MM timer has been set to 1 ms and thus the internal tick rate has been
increased. You can check this with my wclkres tool.

> However when I turn -M on the bottom line changes to Current 0.977 ms
> 
> Is that really correct as even I can tell 0.977 is less than the
> "minimun" 1.000 ms" in the previous line?

I think 1 ms is just the nominal value and 0.977 is due to rounding errors
or an inexact measurement interval.

Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Martin Burnicki]

Alan wrote:
> Thanks for the reply,
> 
> I've checked and I have /usepmtimer in boot.ini and wclkres -p reports
> Performance Counter Frequency: 3.580

So your system does indeed use the PM timer to implement the QPC function.
 
> I also have Cool N Quiet turned off in BIOS.

AFAIK this should only affect the TSCs, so the setting should not matter at
all if the PM timer is used.
 
> As an experiment I tried seeting affinity to processor 0 and retrying
> with the -M switch. No improvement

That's what I had expected. The affinity should also show an effect if the
TSCs are used for QPC, which is not the case.

> - Jitter is at 100+ millseconds 
> within a minute and the clock starts being regulaly stepped forward a
> few minutes later. With affinity set and without the -M switch I synch
> to within a few milliseconds with jitter of a few millseconds almost
> immediately. ALthough both offset and jitter jump up by about 20 msec
> depending on what I run but then settle down again as I expect under
> Windows without -M switch
> 
> Overnight (without the -M option) with the system idle and only running
> NTPD, both offset and jitter on all configured server clocks dropped to
> about 1 msec or less. Again even if the system is "idle" the time drift
> rate rockets as soon as -M is turned on.
> 
> Anything else I might look at?

We are also using QPC in the driver software for our PCI cards. Quite some
time ago we had a customer who also had timing problems with our driver. it
turned out that the problem was the chipset on the mainboard which returned
inconsistent timestamps.

As you have observed, if you don't use the -M flag system time seems to jump
(20 ms in your case) if some other application sets the multimedia timer to
highest resolution. However, this is just what ntpd does on startup if the
-M switch has been specified. So I'd expect that in your case ntpd runs
stable without -M, but I'm afraid if another app changes the MM timer then
time disciplination will be as poor as with the -M flag specified.

Anyway, I'm actually, out of ideas what you could try to improve the
situation.

Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David J Taylor]

"Alan"  wrote in message 
news:z9i5n.58805$q36.5...@newsfe19.ams2...
Would like to get to the bottom of this as well. Using 4.4.6-o with -M , 
I get "Frequency error 3030 PPM exceeds tolerance 500 PPM". Considering 
that without =M the frequency modification is only about 5 PPM then 
obvioulsy something very strange is going on. It seems that even if the 
timer precision is increased by another program then time immediately 
starts to drift rapidly by hundreds of milliseconds.


Alan,

Our experience was that the switching between normal and high-resolution 
timers caused steps of many milliseconds (I don't recall the exact figure) 
which really messed up NTP.  So either run with no MM timers at all, or 
run with the MM timers permanently enabled, and NTP recognises that 
change, and adjusts accordingly.  Have NTP start the MM timers was one 
solution, and hence the -M option.


It might be helpful to know what the event log says with both sets of 
startup parameters, as there may be a clue there which Dave Hart, the 
person closest to this code, can interpret.


I must confess to having nagging doubts about AMD (but with no good 
reason), about whether you have another program setting the time (check 
that w32time.exe is not running - Show Processes from all users), and 
perhaps something in the BIOS.  One final idea (which there was no option 
on my test system) might be to start with just one CPU active in the BIOS.


Cheers,
David 


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Alan]

Would like to get to the bottom of this as well. Using 4.4.6-o with -M , 
I get "Frequency error 3030 PPM exceeds tolerance 500 PPM". Considering 
that without =M the frequency modification is only about 5 PPM then 
obvioulsy something very strange is going on. It seems that even if the 
timer precision is increased by another program then time immediately 
starts to drift rapidly by hundreds of milliseconds.


David J Taylor wrote:

Alan,

I hope you get this fixed.  Here's what you can get with XP with the MM 
timer enabled:



Windows XP LAN-synched to a stratum-1 server, 32s poll for local 
servers, 1024s poll for Internet backup servers:


 http://www.satsignal.eu/mrtg/narvik_ntp-b.html


Windows XP acting as a stratum-1 server:

 http://www.satsignal.eu/mrtg/feenix_ntp_2.html


Both PCs running  "ntpd 4.2.6-o Dec 09 11:48:30.27 (UTC-00:00) 2009 (1)" 
from Dave Hart's site.  Both PCs have Intel, not AMD processors.  Note 
the reduced offset on the monthly graphs in week 1 and week 2 - it was 
so cold here we kept the heating on 24 hours a day.  PC Narvik has been 
rebooted about three times since the middle of week 2, hence the three 
positive transients.


On PC Narvik, Services Manager, Path to executable is:

 C:\Program Files\NTP\bin\ntpd.exe -M -g -c "C:\Program 
Files\NTP\etc\ntp.conf"


Now you happen to mention processors, I realise that the PC where the 
timekeeping is poor with the more recent versions of NTP (i.e. 4.2.5 and 
later) both have AMD processors.  Maybe a coincidence?


Cheers,
David


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[David J Taylor]

Alan,

I hope you get this fixed.  Here's what you can get with XP with the MM 
timer enabled:



Windows XP LAN-synched to a stratum-1 server, 32s poll for local servers, 
1024s poll for Internet backup servers:


 http://www.satsignal.eu/mrtg/narvik_ntp-b.html


Windows XP acting as a stratum-1 server:

 http://www.satsignal.eu/mrtg/feenix_ntp_2.html


Both PCs running  "ntpd 4.2.6-o Dec 09 11:48:30.27 (UTC-00:00) 2009 (1)" 
from Dave Hart's site.  Both PCs have Intel, not AMD processors.  Note the 
reduced offset on the monthly graphs in week 1 and week 2 - it was so cold 
here we kept the heating on 24 hours a day.  PC Narvik has been rebooted 
about three times since the middle of week 2, hence the three positive 
transients.


On PC Narvik, Services Manager, Path to executable is:

 C:\Program Files\NTP\bin\ntpd.exe -M -g -c "C:\Program 
Files\NTP\etc\ntp.conf"


Now you happen to mention processors, I realise that the PC where the 
timekeeping is poor with the more recent versions of NTP (i.e. 4.2.5 and 
later) both have AMD processors.  Maybe a coincidence?


Cheers,
David 


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Alan]

Here's what I see from clockres without the -M option on

Max 15.625 ms
Minimum 1.000 ms
Current 15.625 ms

However when I turn -M on the bottom line changes to Current 0.977 ms

Is that really correct as even I can tell 0.977 is less than the 
"minimun" 1.000 ms" in the previous line?




Evandro Menezes wrote:

On Jan 20, 8:23 am, Martin Burnicki 
wrote:

I've written a little tool which can reports clock frequency for the QPC
API:http://www.meinberg.de/download/utils/windows/wclkres-1.2.zip


Or just run this Microsoft tool from the command-line:

\\live.sysinternals.com\tools\clockres

It's also available from http://live.sysinternals.com/Tools/Clockres.exe

HTH


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Evandro Menezes]

On Jan 20, 8:23 am, Martin Burnicki 
wrote:
>
> I've written a little tool which can reports clock frequency for the QPC
> API:http://www.meinberg.de/download/utils/windows/wclkres-1.2.zip

Or just run this Microsoft tool from the command-line:

\\live.sysinternals.com\tools\clockres

It's also available from http://live.sysinternals.com/Tools/Clockres.exe

HTH

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Evandro Menezes]

On Jan 20, 8:23 am, Martin Burnicki 
wrote:
>
> Also, changes in the CPU clock frequency by AMD's Cool'n'Quiet or Intel's
> Speedstep can mess up the QPC results.

Actually, on systems running Windows XP or later if the HAL deternines
that this is the case, then RDTSC is not used for this function.

HTH

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Alan]

Thanks for the reply,

I've checked and I have /usepmtimer in boot.ini and wclkres -p reports
Performance Counter Frequency: 3.580

I also have Cool N Quiet turned off in BIOS.

As an experiment I tried seeting affinity to processor 0 and retrying 
with the -M switch. No improvement - Jitter is at 100+ millseconds 
within a minute and the clock starts being regulaly stepped forward a 
few minutes later. With affinity set and without the -M switch I synch 
to within a few milliseconds with jitter of a few millseconds almost 
immediately. ALthough both offset and jitter jump up by about 20 msec 
depending on what I run but then settle down again as I expect under 
Windows without -M switch


Overnight (without the -M option) with the system idle and only running 
NTPD, both offset and jitter on all configured server clocks dropped to 
about 1 msec or less. Again even if the system is "idle" the time drift 
rate rockets as soon as -M is turned on.


Anything else I might look at?

Martin Burnicki wrote:

Alan wrote:

I downloaded the latest prebuilt binary from Meinberg (4.2.4p4)and
installed on XP Service Pack 3. NTPD was not only completely unable to
keep the clock in synch but actually made things far worse with time
drifiting backwards by up to a second every few minutes and NTPD
continually stepping the clock forward with the "exceeds 500ppm message".

I then downloaded the pre-built 4.2.6 from Dave Harrt's site and it did
exactly the same. Then I tried turning off the "-M" option and
restarting (both versions). The clock quickly achieved synch to within a
few milliseconds. The drift frequency was calculated at 4.764 (rather
than above 500) and the jitter reduced to a few milliseconds. Now when I
run certain applications the clock jumps by a few millseconds (which the
-M option is supposed to cure but I can live with this under windows)
but my question  is why is NTPD broken in my environment when installed
withe the default Meinberg installation option of setting the timer to
highest resolution? Anyone else seen this?

This on a dual core AMD-64 with WIndows XP Professional fully patched.


By default Windows has a time resolution of 1 timer tick interval only, i.e
about 16 ms with Win XP.

The reference implementation of NTP uses the Windows QueryPerformanceCounter
(QPC) API call to interpolate the time between 2 timer ticks. 


The QPC API is implemented in the Windows Hardware Abstraction Layer (HAL)
and uses one of the timers provided by the mainboard and/or the CPU.

If the QPC call works based on the CPU's timestamp counter (TSC) registers
then the value of the current timestamp taken depends on the CPU core on
which the code is executed.

On many types of AMD dual core CPUs the timestamp counters in the individual
cores are not synchronized, so if a piece of code runs alternating on
either CPU core the resulting timestamps may not be consistent, and
computations based on those time stamps yield faulty results.

Also, changes in the CPU clock frequency by AMD's Cool'n'Quiet or Intel's
Speedstep can mess up the QPC results.

A workaround can be to add the USEPMTIMER switch in the boot.ini file. See:

Explanation for the USEPMTIMER switch in the boot.ini
http://blogs.technet.com/perfguru/archive/2008/02/18/explanation-for-the-usepmtimer-switch-in-the-boot-ini.aspx

This forces the QPC API to work using the power management (PM) timer
provided by the mainboard's chipset instead of the CPU's TSC registers.

AFAIK this boot switch should have been set by some XP service pack, but you
may want to check whether this is really the case.

I've written a little tool which can reports clock frequency for the QPC
API:
http://www.meinberg.de/download/utils/windows/wclkres-1.2.zip

Unpack the ZIP archive and run the command

wclkres -p

in a command line window. Please let us know which Performance Counter
Frequency is reported by the command. If the PM timer is used then the
frequency should be about 3.58 MHz. If the frequency matches the CPU's
clock frequency then the CPU's TSCs are used which may cause such bad
behaviour.

Martin


___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Martin Burnicki]

Alan wrote:
> I downloaded the latest prebuilt binary from Meinberg (4.2.4p4)and
> installed on XP Service Pack 3. NTPD was not only completely unable to
> keep the clock in synch but actually made things far worse with time
> drifiting backwards by up to a second every few minutes and NTPD
> continually stepping the clock forward with the "exceeds 500ppm message".
> 
> I then downloaded the pre-built 4.2.6 from Dave Harrt's site and it did
> exactly the same. Then I tried turning off the "-M" option and
> restarting (both versions). The clock quickly achieved synch to within a
> few milliseconds. The drift frequency was calculated at 4.764 (rather
> than above 500) and the jitter reduced to a few milliseconds. Now when I
> run certain applications the clock jumps by a few millseconds (which the
> -M option is supposed to cure but I can live with this under windows)
> but my question  is why is NTPD broken in my environment when installed
> withe the default Meinberg installation option of setting the timer to
> highest resolution? Anyone else seen this?
> 
> This on a dual core AMD-64 with WIndows XP Professional fully patched.

By default Windows has a time resolution of 1 timer tick interval only, i.e
about 16 ms with Win XP.

The reference implementation of NTP uses the Windows QueryPerformanceCounter
(QPC) API call to interpolate the time between 2 timer ticks. 

The QPC API is implemented in the Windows Hardware Abstraction Layer (HAL)
and uses one of the timers provided by the mainboard and/or the CPU.

If the QPC call works based on the CPU's timestamp counter (TSC) registers
then the value of the current timestamp taken depends on the CPU core on
which the code is executed.

On many types of AMD dual core CPUs the timestamp counters in the individual
cores are not synchronized, so if a piece of code runs alternating on
either CPU core the resulting timestamps may not be consistent, and
computations based on those time stamps yield faulty results.

Also, changes in the CPU clock frequency by AMD's Cool'n'Quiet or Intel's
Speedstep can mess up the QPC results.

A workaround can be to add the USEPMTIMER switch in the boot.ini file. See:

Explanation for the USEPMTIMER switch in the boot.ini
http://blogs.technet.com/perfguru/archive/2008/02/18/explanation-for-the-usepmtimer-switch-in-the-boot-ini.aspx

This forces the QPC API to work using the power management (PM) timer
provided by the mainboard's chipset instead of the CPU's TSC registers.

AFAIK this boot switch should have been set by some XP service pack, but you
may want to check whether this is really the case.

I've written a little tool which can reports clock frequency for the QPC
API:
http://www.meinberg.de/download/utils/windows/wclkres-1.2.zip

Unpack the ZIP archive and run the command

wclkres -p

in a command line window. Please let us know which Performance Counter
Frequency is reported by the command. If the PM timer is used then the
frequency should be about 3.58 MHz. If the frequency matches the CPU's
clock frequency then the CPU's TSCs are used which may cause such bad
behaviour.

Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

Re: [ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Alan]

Alan wrote:
I downloaded the latest prebuilt binary from Meinberg (4.2.4p4)and 

I meant 4.2.4p8

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions

[ntp:questions] Timekeeping broken on Windows XP with multimedia timer enabled (-M option)

[Alan]

I downloaded the latest prebuilt binary from Meinberg (4.2.4p4)and 
installed on XP Service Pack 3. NTPD was not only completely unable to 
keep the clock in synch but actually made things far worse with time 
drifiting backwards by up to a second every few minutes and NTPD 
continually stepping the clock forward with the "exceeds 500ppm message".


I then downloaded the pre-built 4.2.6 from Dave Harrt's site and it did 
exactly the same. Then I tried turning off the "-M" option and 
restarting (both versions). The clock quickly achieved synch to within a 
few milliseconds. The drift frequency was calculated at 4.764 (rather 
than above 500) and the jitter reduced to a few milliseconds. Now when I 
run certain applications the clock jumps by a few millseconds (which the 
-M option is supposed to cure but I can live with this under windows) 
but my question  is why is NTPD broken in my environment when installed 
withe the default Meinberg installation option of setting the timer to 
highest resolution? Anyone else seen this?


This on a dual core AMD-64 with WIndows XP Professional fully patched.

___
questions mailing list
questions@lists.ntp.org
http://lists.ntp.org/listinfo/questions