Re: [time-nuts] DPC Latency
This was a good excuse to fix some driver issues. My mobo had an issue with USB3. I also switched from IDE to ACHI on the Sata. [There is a registry trick to do this. You can't just change the bios. At least for windows.] After all the tweaks, I think the D525 is always going to have 2ms jumps. It is not a fast processor. I tried disabling drivers and it made no difference. Incidentally if you really want to get the latest drivers, you need to run some hardware probing program to see what chips are on your mobo, the get the driver from the chip manufacturer. The motherboard manufacturer isn't always up to date. I flashed the bios, but using PNP OS still gives better results. On 6/18/2011 2:25 PM, Bob Camp wrote: Hi Like any basic tool it's just a start. Simply knowing that the system has this or that latency compared to another system is a useful piece of data. Anything under 100 us isn't going to be easy to find. Stuff that's into the many ms range probably should be tracked down. Bob On Jun 18, 2011, at 3:35 PM, gary wrote: Doing a bit of bios hacking, DPC shows better results with Plug and Play enabled. It is my understanding that it is better to let the bios handle plug and play rather than the OS, but the results are better with Plug and Play handled by win7. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] DPC Latency
Those of us who run SDRs are all too familiar with the DPC issue... You're right, this is a useful little tool. Bill On Sat, Jun 18, 2011 at 9:42 AM, Bob Camp li...@rtty.us wrote: Hi DPC Latency = Deferred Procedure Call Latency = how long it takes Windows to get around to doing what it should do real time. Put another way - how lazy is your computer? It may not be a big issue, but it's probably worth checking for free: http://www.thesycon.de/deu/latency_check.shtml We don't do audio or video since we're serious Time Nuts:)... The same issue would mess up our stuff though. It certainly would be useful to know when playing with NTP. Usual disclaimer - if your PC explodes don't blame me. There are lots of other programs out there that do the same thing. I have no connection with these guys. It's just the first one I found. Bob ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] DPC Latency
Hi I did indeed originally come across it in the SDR context. Bob On Jun 18, 2011, at 9:46 AM, William H. Fite wrote: Those of us who run SDRs are all too familiar with the DPC issue... You're right, this is a useful little tool. Bill On Sat, Jun 18, 2011 at 9:42 AM, Bob Camp li...@rtty.us wrote: Hi DPC Latency = Deferred Procedure Call Latency = how long it takes Windows to get around to doing what it should do real time. Put another way - how lazy is your computer? It may not be a big issue, but it's probably worth checking for free: http://www.thesycon.de/deu/latency_check.shtml We don't do audio or video since we're serious Time Nuts:)... The same issue would mess up our stuff though. It certainly would be useful to know when playing with NTP. Usual disclaimer - if your PC explodes don't blame me. There are lots of other programs out there that do the same thing. I have no connection with these guys. It's just the first one I found. Bob ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] DPC Latency
Hi DPC Latency = Deferred Procedure Call Latency = how long it takes Windows to get around to doing what it should do real time. Put another way - how lazy is your computer? It may not be a big issue, but it's probably worth checking for free: http://www.thesycon.de/deu/latency_check.shtml We don't do audio or video since we're serious Time Nuts:)... The same issue would mess up our stuff though. It certainly would be useful to know when playing with NTP. Usual disclaimer - if your PC explodes don't blame me. There are lots of other programs out there that do the same thing. I have no connection with these guys. It's just the first one I found. Bob Bob, Yes, that's a program which produces some interesting results, and has been useful for detecting bad behaviour by some poorly written drivers (e.g. when receiving masses of data over a satellite DVB link). For NTP, I think you may find that all the time-critical stuff is done at high-priority, and it's only the non-time critical stuff which is done at DPC time. This is how the DPC is intended to be used, of course. VMS has a similar mechanism, ASTs, if I recall correctly, but that was a long time ago. On Windows with NTP, you can get within a couple of hundred microseconds (XP) or perhaps 50 microseconds (Windows-7) if my graphs here are any guide: http://www.satsignal.eu/mrtg/performance_ntp.php The best I can offer is within 10 microseconds or so on an Intel Atom system running FreeBSD. I'm sure that others can do better! Cheers, David -- SatSignal software - quality software written to your requirements Web: http://www.satsignal.eu Email: david-tay...@blueyonder.co.uk ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] DPC Latency
Hi My concern would be a really bad driver that spikes into the 1,000's of us every so often. Every once in a great while there could be a collision. That would really drive you crazy trying to find it. Bob On Jun 18, 2011, at 10:04 AM, David J Taylor wrote: Hi DPC Latency = Deferred Procedure Call Latency = how long it takes Windows to get around to doing what it should do real time. Put another way - how lazy is your computer? It may not be a big issue, but it's probably worth checking for free: http://www.thesycon.de/deu/latency_check.shtml We don't do audio or video since we're serious Time Nuts:)... The same issue would mess up our stuff though. It certainly would be useful to know when playing with NTP. Usual disclaimer - if your PC explodes don't blame me. There are lots of other programs out there that do the same thing. I have no connection with these guys. It's just the first one I found. Bob Bob, Yes, that's a program which produces some interesting results, and has been useful for detecting bad behaviour by some poorly written drivers (e.g. when receiving masses of data over a satellite DVB link). For NTP, I think you may find that all the time-critical stuff is done at high-priority, and it's only the non-time critical stuff which is done at DPC time. This is how the DPC is intended to be used, of course. VMS has a similar mechanism, ASTs, if I recall correctly, but that was a long time ago. On Windows with NTP, you can get within a couple of hundred microseconds (XP) or perhaps 50 microseconds (Windows-7) if my graphs here are any guide: http://www.satsignal.eu/mrtg/performance_ntp.php The best I can offer is within 10 microseconds or so on an Intel Atom system running FreeBSD. I'm sure that others can do better! Cheers, David -- SatSignal software - quality software written to your requirements Web: http://www.satsignal.eu Email: david-tay...@blueyonder.co.uk ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] DPC Latency
Hi My concern would be a really bad driver that spikes into the 1,000's of us every so often. Every once in a great while there could be a collision. That would really drive you crazy trying to find it. Bob Indeed, Bob. On a critical system, it might be best to establish a baseline for DPC when you first install the OS, and then note any changes as extra devices are installed. You would be unlucky to find a bad driver in the base OS install, although it's possible. Cheers, David -- SatSignal software - quality software written to your requirements Web: http://www.satsignal.eu Email: david-tay...@blueyonder.co.uk ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] DPC Latency
I tried two PCs and my D525 system has 2mS hits. Unfortunately this program doesn't really aid in finding the glitch. I turned off the sound and not much difference. My AMD quad core has peaks of 200us. Regarding linux, I've never tried them, but there are hacks to reduce the latency. Google will dig up all sorts of things to try. http://www.mixxx.org/wiki/doku.php/troubleshooting has tips for windows and linux. On 6/18/2011 8:26 AM, David J Taylor wrote: Hi My concern would be a really bad driver that spikes into the 1,000's of us every so often. Every once in a great while there could be a collision. That would really drive you crazy trying to find it. Bob Indeed, Bob. On a critical system, it might be best to establish a baseline for DPC when you first install the OS, and then note any changes as extra devices are installed. You would be unlucky to find a bad driver in the base OS install, although it's possible. Cheers, David ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] DPC Latency
Doing a bit of bios hacking, DPC shows better results with Plug and Play enabled. It is my understanding that it is better to let the bios handle plug and play rather than the OS, but the results are better with Plug and Play handled by win7. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] DPC Latency
Hi Like any basic tool it's just a start. Simply knowing that the system has this or that latency compared to another system is a useful piece of data. Anything under 100 us isn't going to be easy to find. Stuff that's into the many ms range probably should be tracked down. Bob On Jun 18, 2011, at 3:35 PM, gary wrote: Doing a bit of bios hacking, DPC shows better results with Plug and Play enabled. It is my understanding that it is better to let the bios handle plug and play rather than the OS, but the results are better with Plug and Play handled by win7. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
