Getting Unexpected Realtime Delay on New Install

Hi Guys,

I'm stumped. I'm trying to set up LCNC on a new computer, but keep getting the above mentioned error message. Could someone take a look at the attached error file and let me know if something jumps out at you? Thanks,

Mr. Technical

[ 0.000000] 24 Processors exceeds NR_CPUS limit of 8
[ 0.000000] SMP: Allowing 8 CPUs, 6 hotplug CPUs

[ 0.000000] setup_percpu: NR_CPUS:8 nr_cpumask_bits:8 nr_cpu_ids:8 nr_node_ids:1

[ 0.192874] Total of 2 processors activated (13599.23 BogoMIPS).


[ 8.606054] ACPI Warning: 0x00000b00-0x00000b07 SystemIO conflicts with Region \SOR1 1 (20120320/utaddress-251)
[ 8.606061] ACPI: If an ACPI driver is available for this device, you should use it instead of the native driver

[ 8.710643] [drm] radeon: power management initialized

All sorts of potential problems, what computer is it and what were the latency test results?

regards

It's an MSI 760GMA-P34FX board with onboard Radeon video. 4G of RAM. CPU is an AMD Sempron 145, 2.8 GHz. BTW, there was an error message when I did the Wheezy install about the RTAI component if memory serves.

Mr.Technical

What result do you get from a latency-test?

Sorry, forgot to mention that. A short test puts it at 22458 ns for the servo thread and 22206ns for the base thread.

That sounds OK.

What are the thread speed settings in your .ini file

eg

[EMCMOT]
BASE_PERIOD = ?
SERVO_PERIOD = ?

You probably don't have the former, looks like you are using a mesa board

You're right, I am using Mesa boards (5i25, and a 7i76). They are not necessarily matched to each other though, since they came from different orders.

The servo period is: 1000000
The INI file does not show a base period under EMCMOT.

There are lots of crazy things in the dmesg, This also stands out:

[ 31.463837] In recent history there were
[ 31.463838] 463051, 24210495, 234240995, 27159464, and 2443651
[ 31.463839] elapsed clocks between calls to the motion controller.
[ 31.463842] This time, there were 6381947 which is so anomalously
[ 31.463842] large that it probably signifies a problem with your
[ 31.463843] realtime configuration. For the rest of this run of
[ 31.463844] EMC, this message will be suppressed.

Have you turned off all power management and C states in the BIOS?
It looks like the CPU clock rate is changing.

The wiki has a list of things to check:

wiki.linuxcnc.org/cgi-bin/wiki.pl?Troubl...ck_dmesg_for_details

Some computers seem to have a latency spike when loading LCNC which is never repeated, but this appears possibly to be connected to the Mesa boards

Every time you start the RTAI error is immediately after the 5i25 initialisation, which causes the 5i25 watchdog to activate and it is dropped.

I will leave PCW to advise on this, my knowledge of Mesa boards is limited

regards

The clock rate changing is interesting. Looking at the box I just noticed that it says it does auto overclocking with OC Genie. I'm going to read up on how to turn that off. Thanks for the lead. I'll keep you posted.