latency test question

when i run the latency test it give me a very high number but when i run
cd /usr/realtime*/testsuite/kern/latency; ./run

its stay bellow 15000 and no over run.

can someone help figure out why the latency trip out?

also i gave onboard video, can that be an issue?

its g41 intel chipset gigabyte board with dual core pentium, no bell and whistle, very bare system.

thanks.

actually it did give me a over run after 10 min of 160 000
i wonder why i cant get this to work,

i edit the files in etc/linuxcnc/rtapi.conf like its siad on the guide.

not sure what else to do.

buying a video card?

here is my bug file if someone can look at it.

Hi

There is a lot of conflicting stuff here and no real information

i edit the files in etc/linuxcnc/rtapi.conf like its siad on the guide.

This relates to SMI, what you describe does not fit the SMI problem, so remove the edits.

Please post the output from lspci -vv so we can see exactly what chipsets etc you have, not just 'I have onboard video'

here is my bug file if someone can look at it.

This is a log from when Linuxcnc failed to start - something you have not mentioned and not the problem you have described.

Without the full dmesg output cannot be sure but this appears to be the issue
insmod: error inserting '/usr/realtime-2.6.32-122-rtai/modules/rtai_hal.ko': -1 File exists

If you haven't screwed up module loading with your rtapi.conf edits, then it could be as simple as that rtai_hal.ko was not unloaded properly from a previous session.
If halrun -U does not work, then try rebooting and try again.

Have a read of these FAQs if you have not already
www.linuxcnc.org/index.php/english/forum...-the-latency-problem
www.linuxcnc.org/index.php/english/forum...me-latency-solutions

Then make sure that you have turned off all power management, multi-threading etc in BIOS
Set the isolcpus parameter on the boot line as described to isolate the second CPU, then try again.

The most important thing with latency diagnosis is to try to isolate what is happening at the time of the spike.

You have not mentioned any testing, what is happening when the spikes occur, heavy graphics use, HDD access, network use etc?

regards

i read those 2 link top to bottom.

so far what i did is:
-change the multithread
-stop the sound driver
-removed every usb accessories but keyboard and mouse
-removed the edit that i did on the .conf files.

now when i run the test ins pretty much at 4000 which is good until i start some heavy graphic 3d stuff it jump to 172 000 ,
i ran it for hrs doing nothing and it was smooth and soon that i did the 3d it jump so seem to be video related,
do you have an idea if i need to put a video card or change the driver?

my motherboard is gigabyte ga 41mt s2tp with coreduo 1.8ghz with 4gig ddr3 ram.

Hi

now when i run the test ins pretty much at 4000 which is good until i start some heavy graphic 3d stuff it jump to 172 000 ,
i ran it for hrs doing nothing and it was smooth and soon that i did the 3d it jump so seem to be video related,
do you have an idea if i need to put a video card or change the driver?

You seem to have done well, 4000 is a good figure, looks like video related

Unfortunately you haven't posted your output from lspci -vv so I have no idea what chip you have.

First try changing the driver to the vesa one
wiki.linuxcnc.org/cgi-bin/wiki.pl?Troubl...g#Using_Vesa_Drivers
I updated the wiki recently to give a bit more info, depends whether you are using 8.04 or 10.04 as to how easy it is.

Before you start make sure you know how to boot the recovery version into a shell and change the xorg.conf files back on the command line
ie save the old file to something easy before you start and if it all goes wrong and locks up, you can go back in via a root shell and copy the old one over the new one and get back to where you were.

regards

I tried installing the vesa but was unsuccesful. here is my ISPCI file.

i have question, is it better to tun the latency test from linux cnc or run it in the terminal,

when i run it in the terminal, its stay very low, around 3000 to 5000 and sometimes negative but when i run it in the gui i get some higher number like 30 000, i still get the very high number but only once in 10 hrs.

i also get this message when i try to make a xorg.conf file.


Fatal server error:
Server is already active for display 0
If this server is no longer running, remove /tmp/.X0-lock
and start again.


Please consult the The X.Org Foundation support
at wiki.x.org

ok,

i think i got it, installed the vesa driver.
not sure how to verify but when i do the latency test on the terminal i get good numbers,
so far so good after 30 min, numbers are at 3000 sometimes goes up to 30 000 and sometimes goes down to minus 500. is that normal?

thanks for your input.

my number seem better when i run the latency test in terminal thant the linuxcnc one.

is there a big diffrenec between the 2?