Real time error on fast PC & install PCI Parport?

Hi I got two noob questions. One is about why real time error is so bad on my super PC and also how to install PCI parallel port?

(1)

I used to use a crappy stone age computer that was doing fine with EMC2. It had jitter of around 7000. That PC is now gone. I decided to use my super gaming PC that handles Crysis 2 max config without a hitch, but I am getting all sorts of problems from real time error to jitter in the 20000s. Talk about david vs goliath.

I do have the so called dreaded Nvidia graphics card but there is no way for me to change this unless I sink another few hundrew to get another brand of cards. I also dont have native graphics card on the motherboard. For now Im stuck.

Other than that what could be the reason? I even tried the SMI thing but doing so makes EMC2 not capable of starting/


(2)

At the same time, since I am now using a new computer without a parallel port on the motherboard, I bought a startech PCI parallel port.

However I have trouble installing that. More so, I am not too sure how to install it.

Firstly, is the PCI parallel port plug and play? There is an 'installation guide' on the CD for linux but that is generic linux and also some of the commands in the manual dont work for ubuntu.

I suppose I need to find the address using lspci -v. I did this and I do see what seems to be the PCI parport card. There are like 5 addresses for it? Anyway one seems to be df01.

So in stepconfig, do I write in 0xdf01?

I am super noob so expect all of this to be wrong. So please help me!

Hi.

Yea its a quad core. Unfortunately its the only one I have. The next will be a laptop and that I heard is definately not the way to go or is worse.

Im not next to my PC at the moment but are you saying I should type in isolcpus=1 in the terminal to only utilize 1 core then it will become ok? I can try this later.

About the PCI card part. I am stuck about that. Unfortunately Im not near my comp so I cant paste the details, but when I type in lspci -v, I remember getting a bunch of ouput addresses. It is a startech parport PCI card with 2 physical parports.

So say the first output from lspci -v was df01. In stepconfig, should I type the parallel port address as 1xdf01?

I should mention that when I type in 0xdf01, EMC2 gets and error and cannot be started.

Cheers

Hi,

isolcpus is a kernel parameter, not a command, syntax etc below.

www.kernel.org/doc/Documentation/kernel-parameters.txt

I would not go rushing down that road immediately, the first priority is probably the nVidia card.
Have you tried uninstalling the restricted nVidia driver and using the bog standard open source nv driver?
That sometimes gives much better results for EMC (but not for gaming!).

It does not have to cost much to put another video card in, a cheap PCI card with older Matrox or ATI chipsets are said to work well.

but when I type in lspci -v, I remember getting a bunch of output addresses. It is a startech parport PCI card with 2 physical parports.

Ref the PCI parport card you need to include the output in a post (just the bit relating to your card) so we can see what it says, use lspci -vv for greater detail
There are usually but not always 2 addresses shown for each port, 2 for the hard wired port and 2 for the fly lead one, sometimes it is a case of experimentation to see if 1 & 3 are the ports or 2 & 4 for instance, and which port is which ( hard wired or fly)

The address is used in your hal file, if you already have a configuration it is easier to just edit it than keep using stepconf

The relevant line is loadrt hal_parport cfg="0xXXXX out" where XXXX is the hex address of the port.

regards

Have you looked here?

wiki.linuxcnc.org/emcinfo.pl?TroubleShooting

www.linuxcnc.org/component/option,com_ku...id,6310/lang,german/

Rick G

Have you tried both 8.04 and 10.04? 10.04 may be better for newer machine.

also look here...

wiki.linuxcnc.org/emcinfo.pl?action=hist..._Parameter_And_GRUB2

Rick G

questi wrote:

I do have the so called dreaded Nvidia graphics card but there is no way for me to change this unless I sink another few hundrew to get another brand of cards. I also dont have native graphics card on the motherboard. For now Im stuck.

You don't need a high-end gaming graphics card, really crummy graphics cards out of scrapped computers will probably work to handle the Axis 3-D preview. I keep a few of these out of dumpster computers around.

Firstly, is the PCI parallel port plug and play? There is an 'installation guide' on the CD for linux but that is generic linux and also some of the commands in the manual dont work for ubuntu.

Yes, all PCI cards must have a PnP function to enumerate and assign addresses on the bus.
That's why PCI cards don't have DIP switches, and will change address when other cards are
plugged in.

I suppose I need to find the address using lspci -v. I did this and I do see what seems to be the PCI parport card. There are like 5 addresses for it? Anyway one seems to be df01.

So in stepconfig, do I write in 0xdf01?

Well, you have to try them out one at a time and see which one works. So, yes, try each of the addresses.
Usually, they will end in either 0 or 8 and not an odd number like 1.

Jon

I dont wana buy another computer if I dont have to. Maybe I can get my hands on a pentium 1.6GHz single core, 500mb ram. Is that enough? Its worse than my old one.

I tried all addresses and none of them work. THen I went into the .hal file and manually deleted the 'out' and the spaces after the ports in the double quote after the address. Then it started to load. Its wierd. I dont know why. Then if I tried to add two addresses in step config, for both cards, it doesnt work again e.g. loadrt hal_parport cfg="0xdf01 out 0xdf02 in". I cant put two addresses. I can only do one in that quote. I also added the extra lines to read the extra port.

The EMC I am using is the LiveCD uinstallation.

So for isolcpus I set the highest core so in my case 3? I am unsure how to add the parameter to the kernel?

it doesnt work again e.g. loadrt hal_parport cfg="0xdf01 out 0xdf02 in".

Please post your lspci -vv print for your PCI card as requested.

You can't possibly have 2 ports at addresses 1 byte apart and as Jon stated ports will usually end in 0 or 8.

I have 2 PCs with 2 additional PCI cards in each and all the port addresses end with 0 and are either 0x400 or 0x800 apart.

I would forget about isolcpus for now, until you can connect your computer to a machine, the source of any realtime errors is rather irrelevant.

isolcpus is unlikely to be the magic bullet anyway unless you can eliminate the nVidia card from the equation first.
How did it run with the basic nv driver?

regards

It might be better to look at just one problem at a time.
I would start with the latency. As pointed out if you have access to a different video card try that.
If not look at the troubleshooting section mentioned and try the basic video driver. Do any of these help at all?
Can you turn any of the cores off with your bios?
Not sure if you mentioned if you are using 8.04 or 10.04.
There is an example of the isolcpus statement here...

wiki.linuxcnc.org/cgi-bin/emcinfo.pl?EMC_Ubuntu91


Rick G