Author Topic: PPS for timing on MIPS hardware running BSDs and Linux?  (Read 1866 times)

0 Members and 1 Guest are viewing this topic.

Offline cdevTopic starter

  • Super Contributor
  • ***
  • !
  • Posts: 7350
  • Country: 00
PPS for timing on MIPS hardware running BSDs and Linux?
« on: December 17, 2018, 05:42:30 pm »
Does anybody know how the interrupt needed for precise timing is implemented on older MIPS architecture machines under Linux and BSDs? I'd like to use an old Cobalt machine I have as a headless time server which requires the ability to provide it with the PPS pulse to a hardware pin.

A brief high whose leading edge asserts itself at the exact instant the second changes, provided by a GPS or similar.

My server has a QED RM5231 CPU which ust predates the rise in interest in precise timing.

Ive found documentation on the CPU chip and can identify the places where the interrupts likely are but would like to know a bit more about how other MIPS hardware running BSDs and Linux handled PPS and GPIOs.

"What the large print giveth, the small print taketh away."
 

Offline LapTop006

  • Supporter
  • ****
  • Posts: 467
  • Country: au
Re: PPS for timing on MIPS hardware running BSDs and Linux?
« Reply #1 on: December 18, 2018, 04:42:01 am »
Given those MIPS cobalt machines are coming up on 20 years old I'd consider ripping the guts out and use a beagleboard for which all the timing bits are documented.
 

Offline cdevTopic starter

  • Super Contributor
  • ***
  • !
  • Posts: 7350
  • Country: 00
Re: PPS for timing on MIPS hardware running BSDs and Linux?
« Reply #2 on: December 18, 2018, 01:52:45 pm »
I've thought about that.

Or maybe some other SBC, one with two NICs, because the Beagle Board only has a single NIC.

But I also like the idea of just keeping it as it is and adding the PPS to it, by putting a GPS inside it. Or maybe using it to house my GPSDO?

It has a 250 MHz 64 bit processor, and although I am fairly sure that the current OS's which run on it don't use the 64 bit aspect of the processor, its possible that some of the renewed interest in MIPS will lead to more options there.

It was very much ahead of its time when it came out (I'm its second owner) but now its very old. Even when I bought it it was old and outdated.

However its built like a tank, ran for years without a hiccup, takes little power, has an RTC, real hardware interrupts and non-USB NICs, its RTC keeps excellent time, has an interesting CPU architecture, and its a really pretty color.
« Last Edit: December 18, 2018, 02:35:04 pm by cdev »
"What the large print giveth, the small print taketh away."
 

Online ebastler

  • Super Contributor
  • ***
  • Posts: 6202
  • Country: de
Re: PPS for timing on MIPS hardware running BSDs and Linux?
« Reply #3 on: December 19, 2018, 02:44:08 pm »
Given those MIPS cobalt machines are coming up on 20 years old I'd consider ripping the guts out and use a beagleboard for which all the timing bits are documented.

Why bother gutting it? Just throw the whole thing into the trash and mount your Beagleboard in a nice, small modern enclosure.  ???

Heck, this is the "Vintage Computing" forum! Sacrilege!! :scared: ;)
 

Offline cdevTopic starter

  • Super Contributor
  • ***
  • !
  • Posts: 7350
  • Country: 00
Re: PPS for timing on MIPS hardware running BSDs and Linux?
« Reply #4 on: December 19, 2018, 07:11:49 pm »
Scandalosa!

----

Actually, I have a very good electronics-based reason to want to use it. Its a server class, incredibly robust, 64 bit machine that I think has real hardware GPIOs close to the metal. I suspect that if i can use it with the NTP PPS it would make a very decent machine for NTP for that reason.

It's also a 3.5 volt logic machine, so the GPS that I have will work perfectly with it. No level shifting required. It has a real NIC card, so less latency than my Raspberry Pi.

And its a really pretty looking machine, its an almost iridescent blue colored cube.

Its by far the prettiest looking machine I own.  Also, when it was in storage its onboard RTC kept amazingly good time for 10 years, while it was sitting in a box, and had no power. When I recently booted it up to see if it still worked, it had the time within just a few hours. When I subtracted the 3 hour time difference the error was even smaller.

All the time that I had it it worked without a hitch. the reason I stopped using it was the untimely death of its power supply.
"What the large print giveth, the small print taketh away."
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf