[13:51:39] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TroubleShooting#Unexpected_realtime_delay_check_dmesg_for_details
Hmm.. i sort of figured I would take care of it if it would become a problem
_bugs_: as I said, you might not notice it on servos
_bugs_: intel board?
otherwize it would show up as inacuracys
Acer laptop :)
yeah don't worry.. I'm just testing
that's the most probable source of problems
you'll probably not see it on a real system..
with a bit of luck :)
I get thoose messages when I have a usb stick plugged in
Hmm... I was thinking of using an old K6-450... but I might neem some new hardware
450 is close :D
Aha... I need quite a few rs232 ports.. i might have an old 4-port ISA card somwhere
That would probably be the best solution right? :)
I have a PCI parport + 2 serial board plugged in right now
they have 16550's :)
OK... nothing to worry then
right now i'm looking more at getting the software side to work
so, any other problems with 2.0.5 except that warning?
well.. that's not just a warning.. it doesn't compile
MODULE_PARM is simply not defined
I did find a patch on the developers list.. but that applies to HEAD...
so i didn't try it
nm it then ;)
what distro is this?
hehehehe.. SLACKWARE :)
rtai 3.4 ?
JMK (another developer from here) just reported a problem with SMI & RT latency today
[13:59:54] <alex_joni> http://www.captain.at/rtai-smi-high-latency.php
System Maintenance Interrupt
it's above the RT system
so it doesn't know how to disable it
Hmmm. doesn't sound like something you would like to disable
sure you would
it does APM and ACPI stuff mostly
which are disabled for RT kernels
btw.. you _did_ turn off all ACPI and APM from the kernel.. right?
ah... that might explain my problems aswell :)
I was more concerned with compiling anyway...
03jepler 07HEAD * 10emc2/src/.swish_config: Update patterns for non-ASCII files, including ones generated by quartus and ise
anyway.. i suppose there is no work on an rs485 interface yet?
ok good :)
do you need RT on the serial interface?
probably you do
haven't figured out the details yet...
I'm just trying to find the best version to work on :)
but my guess is yes
there is an RT driver for rs232
included in rtai iirc
it runs at 19k2.. .. you need to send an decial 1.. and get a 7 byte return code within aprox 10 ms
ok, then definately not RT
well at 19k2 what can you expect :)
I've never worked with RT :)
steves_logging is now known as steve_stallings
19k2 is easily doable in user space
RT is 10 usec, not 10ms
yeah.. it's not a software UART :)
but it's acutally nice to turn of RTS after sending the byte quickly
otherwise you might mis the return data
one way of doing this is to hook the receiving interrupt and test for the data you've send
well, I guess you'll figure it out :P
sure.. never mind the details
euhm... so 2.1 branch it will be :)
remember it's not bugfree right now
I mean it's not production ready yet
yeah.. i mean.. what software isn't bug free :)
hehehehe... yeah but that doesn't compile :)
that's a feature
And i can't realy be bothered trying to fix it :)
if 2.1 is working moderatly well
i'm sure my efforts or better focused on that :)
2.1 has lots of more advanced features which you'll probably want anyways
please report back any problems you encounter
I still have a lot to learn about realtime and HAL :)
ok no problem.. my real name is Stefan Raaijmakers
Alex Joni here :D
I'm in the Netherlands, main intrest is robotwars
we got the CNC machine for scrap-metal value :)
but wat kan you expect for a 30 year old machine
yeah battebolts in the US
ahh.. ok, know them
* alex_joni works with different kind of robots
hehehehe.. actually.. i like the technology more that the actual fighting
but it's a good way to test your equipment :)
I built this one a while ago
[14:20:30] <alex_joni> http://dsplabs.cs.utt.ro/~juve/yanor/images.php
[14:20:38] <_bugs_> http://www.bugs.nl/media/album/view.php?&path=L3J3L3ByaW1hdGVjaDIwMDMvRFNDRjAyMTkuSlBH
_bugs_: one thing I always wondered about with robot wars, are electronic weapons not allowed ?
ah autonomous.. cool :)
that one's scary
nope.. they are not :)
_bugs_: figured, I always wondered why nobody had a raygun or 50Kv contacts
I used to work on small autonomous
yeah.. well the easy one would be to block the opponents transmitter
but that's not allowed either :)
_bugs_: I work with these usually: http://www.robcon.ro/301589/sIMG_8759.JPG
those are more deadly then ours :)
only if you are not nice to them
i'm always surprized at the precision you can get with those
although this is a welder..
not to bad.. enought for welding
now to do some milling with it... :)
not at 15kg payload :D
hehehehe.. too bad :)
i'd be glad if we can get this 3-axis milling machine to work
We were actually planning on using mach3, but the linear scales we got only gave an rs485.. not an AB differential signal
so i was glad there's an opensource alternative :)
you still probably need something to output analog for the drives
yeah.. but electronics are no problem :)
for that you have 2 possibilities
1. motion card
2. parport + pwmgen
I'm going for option 3
DIY interface :)
in regards to diy interface
because i can
and i haven't found anything that can do analog out -10 to +10v
I thought about using the soundcard.. but that doesn't seem to work well with RT
I selected the DAC7744 chip because it seams to be the ideal DAC for our
+-10V outputs so you only need some amps to drive the outputs
16 bit resolution (dynamic on a analog drive is directly affected by the
resolution of the DAC)
that's from an email on the emc-users list recently
audio DACs aren't really good for industrial uses - they tend to have missing codes and such
availability is usually an issue aswell
and this thing kost 30 dollars
if you buy 1000 pices :)
just out of curiosity.. are there any other analog outputs for emc2?
[14:39:06] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?EMC2_Supported_Hardware
ah.. could have figured that one out myself :)
how would one go about calculating the ripple if you take 1MHz PDM (pwmgen in pdm mode but 'fpga fast') -> RC network with 100uS time constant -> suitable amplifier?
using emc2 to output an up/down pwm signal -(to run the h-bridge)
that's an h-bridge :)
jepler_, take the worst-case signal from the PWM (not sure what that is), and calculate the capacitive decay in that amount of time
* skunkworks should know this stuff
can't PDM have arbitrarily long period?
yes, bunt long periods don't create ripple
anyway.. i think i know enough for now
well, if you have the output on all the time, then you're trying to get to 100%, so the filter output approaches 100% over time
the worst case is the smallest or largest output, e.g., 1us on, 999us off
yeah, that's what I was thinking
low pass filter have an inherent intergration
actually, at 50%, you have a series of 1 uS ons and 1 uS offs, and that may be the worst case as far as ripple is concerned
you can use the rc-time to aproximate the ripple
steve_stallings is now known as steves_logging
capacitive charge / decay is e^-(t/tau)
(but you knew that :) )
i'm leaving aswell
the largest "t" is the worst case, then
Rugludallur_ is now known as Rugludallur
jepler_, yes and no. if you want 100% duty cycle, then there's no ripple, but you have the longest t
at 50%, you alternate quickly between 100% and 0%, and that's where you get ripple - the difference between the ideal flat 50% line and what actually comes out
with a time constant (tau) 100x the expected shortest t, you should get deviation of only e^-100, which is very small (and possibly wrong - I haven't looked at those books in a decade or more :) )
hmmm - I need more coffe. don't listen to me yet
it seems to me that (repeated) on/off/off would have more ripple than on/off, is my intuition wrong?
I think that's correct
if so, 50% isn't the worst case
that's why you use PDM instead of PWM - it keeps the switch frequency higher
this is PDM, 50% vs 33%
oh wait - I misread your comment ;)
(my gut is often wrong about this stuff)
I think you have to look at how much change in one direction can happen before a correcting pulse pulls it the other way
in the 999/1 scenario, there's only one pulse width of correction, so the filter is working with a 1 uS pulse (not the 999 per se)
in the 50/50 scenario it's also just a 1uS pulse, but more often
so are you suggesting that the ripple voltage P-P will be the same in both cases?
I'm not sure of the exact waveforms that will be output, which is why I mentioned finding the worst case
I think it will be, actually
but I could be caffeine-deprived and therefore wrong
(or just wrong anyway)
I think that proper PDM will always be of the "occasionally 1 cycle of the less frequent output" kind
yes - that's the reason for using PDM
but the ripple frequency varies from 1kHz (if 1:999 is the most extreme output) to 1MHz
(the most extreme, non-DC output)
so in that case, the output probably goes all the way to 100%, and is then pulled back down a little bit by the 1 uS 0 pulse
so the ripple is 100% - whatever happens in 1 uS, and it occurs every millisecond
ok - time for me to run off with the wife. bbl.
the mesa 7i33 (analog +-10v output) says it takes PWM input to analog output -- but it sounds like PDM would give much better results
maybe the results are perfectly good as-is, I dunno
I wonder if "interleaved PWM" is their word for PDM
The 7I33 is meant to operate with PWM rates from 100 KHz to 10 MHz. Operation at lower PWM rates will result in excessive output ripple . When used with SOFTDMC firmware, it is best to set the PWM rate to the maximum (192 KHz). If the motion control firmware is capable of generating interleaved PWM, that option should be enabled to minimize output ripple.
[15:46:31] <jepler_> http://groups.google.com/group/comp.arch.embedded/browse_thread/thread/58f0b3f563c91383/cc4e23312742a03b?lnk=st&q=%22interleaved+pwm%22&rnum=1&hl=en#cc4e23312742a03b
I don't quite follow that explanation of "interleaved PWM"
jepler_: which one do you generate on pluto?
... so, is it taboo to use the reiser fs now that the creator is in jail for murder?
one of the reasons to make software Free is so the software can survive various people-problems. so I think certainly not.
the other good thing about open software is that the ext has and will borrow stuff from reiserfs
so in the end it might not be called reiserfs but it will still contain some of the logic
cradek: traditional PWM
jepler_: I like that better because it doesn't "sing"
random question, what is the usual resonant speed for steppers?
yep -- it seems to be much more appropriate for directly driving a h-bridge
Lerneaen_Hydra: I think it depends heavily on the load on the stepper (as well as on the stepper itself) -- so it would be different for different table weights, leadscrew pitch, reduction (if any), etc
oh, that sounds sane
so no real "usual" speed
rpm at the shaft?
cradek: get your mix and match machine running last evening?
I did play with the lathe a bit last night
finished tuning (X)
Still running the pluto?
are you going to make a special board for it? (the lathe)
haven't filtered the spindle encoder yet, so it still doesn't work
I don't know what I'll do exactly - it's kind of a mishmash right now
with spindl;e encoder circuit
i instaled emc from cvs
[16:38:03] <jepler_> http://www.nt.ntnu.no/users/skoge/prost/proceedings/cdc03/pdffiles/papers/ThA08.3.pdf
that sounds like Ed's problem
Ed Niesley or something like that?
i installed ubuntu from cvs
jmkasunich: yeah that's what I was thinking
however I am not obtaining to use
Error in startup script: can't read "make_shortcut": no such variable
they could help me?
Guest329: start EMC2 from a shell, and capture all of the output that appears in the shell, not just the error message
then put it at pastebin.ca
i am comliled emc
Guest329: start EMC2 from a shell, and capture all of the output that appears in the shell, not just the error message, then put it at pastebin.ca
emc no start axis
pastebin.ca is a website: http://pastebin.ca
I do not know to use pastebin
I can place here?
it is not hard to figure out
go to http://pastebin.ca
with a web browser
copy stuff from the shell window
paste it in the pastebin window
click the submit button
it will give you a url
paste the url here
please one moment
I am in that place
with PWM, are you still restriced by the max voltage on the motor label ... or could you run 2x @ 50% duty cycle?
eholmgren: I run mine over their rated voltage. I think if they don't go too crazy, and they don't get hot, it's ok
if YOU don't go too crazy
Guest329: 10:52:32 <jmkasunich> paste it in the pastebin window
follow the next steps that jmkasunich said
ok i press
<jmkasunich> click the submit button
<jmkasunich> it will give you a url
<jmkasunich> paste the url here
[16:57:49] <Guest329> http://pastebin.ca/288823
yes and no
yes, you used pastebin correctly
no, you did NOT post everything that shows up when you run EMC, you only posted the error message
you can help me?
no, because we do not have enough information
it received my message?
I can read what you put in pastebin very well
but we don't know anything else
we don't know how you compiled EMC, etc
Alex_Joni this here?
03jepler 07v2_1_branch * 10emc2/tcl/bin/pickconfig.tcl: fix error when user has no Desktop directory: can't read "make_shortcut": no such variable
03jepler 07HEAD * 10emc2/tcl/bin/pickconfig.tcl: merge rev 126.96.36.199: fix error when user has no Desktop directory: can't read "make_shortcut": no such variable
Guest329: you need to do a "cvs up" to get the fix that jepler just made
cvs up -dP
sudo make setuid
i go to try
jepler_ is now known as jepler
i I am making this
please one moment
now try running it again
it did not occur certain
same error as before, or something different?
before and now
when you did the "cvs up" did you see a line that started with "P"?
like this: "P tcl/bin/pickconfig.tcl"
(scroll back in your shell and check)
no not see
I go to make of ne
then paste that at pastebin.ca, and post the url here
mine looks like this: http://pastebin.ca/288842
please one time
in which directory
in which directory?
put it in pastebin please
cat: CVS/: Is a directory
I go to place in pastebin
[17:22:57] <Guest329> http://pastebin.ca/288854
ok, while we look at the pastebin, do this:
grep Revision tcl/bin/pickconfig.tcl
that will make one line of output, you can paste it here
(pastebin is for long things)
type "grep Revision tcl/bin/pickconfig.tcl" in your shell
grep Revision tcl/bin/pickconfig.tcl?
it will print something like:
# $Revision: 188.8.131.52 $
# $Revision: 1.48 $
you are running CVS HEAD
(I thought you were using version 2,1)
he should have 1.49
if the "cvs up" was successful, you would have either revision 184.108.40.206 or 1.49
i dont undestand
what i make?
jepler tried to fix your problem
he made a change to the code in CVS
when you do "cvs up" it should copy that change to your computer
but it did not
you still have version 1.48, jepler made version 1.49
we have to figure out how to get version 1.49 on your computer
I don't know - when I do "cvs up" it works
when you do "cvs up" it doesn't work
what directory were you in when you did "cvs up"?
should be /home/epica/emc2
"cvs up dP"
"sudo make setuid"
cvs up -dP
cradek is right, I made a mistake
"-dP", not "dP"
I am in this directory
what is "this" directory?!
you must be very precise when talking on IRC, or people get confused ;-)
and make this comand
cvs up -dP
that command should make many lines of output
cvs update: Updating tcl/bin
do you see the "P" line?
[17:38:19] <eholmgren> http://www.cs.tu-bs.de/rob/david.html
updating tcl/bin yes
type "grep Revision tcl/bin/pickconfig.tcl" in your shell
do not change directories unless we ask you to
# $Revision: 1.49 $
now we rebuild the software with the latest version
follow these steps exactly:
"sudo make setuid"
.. refers to the directory containing this directory, so it reverses the effect of "cd src"
takes you from /home/epica/emc2/src to /home/epica/emc2
\ is windows, / is linux
iam in linux
the programa in java
i use \
emc does not have java
eholmgren: I agree, human society are pretty stupid..
I go to explain
not this configured
therefore use \ instead of/
you type "/", but what do you see on the screen? "/" or "\" ?
anonimasu: ? :)
eholmgren: regarding the reiserfs issue
inlin linux I use/
in linux I use/
eholmgren: I find it retarded that people judge something good somone has made with regards to their personal matters..
anonimasu: I saw that in the headlines again somewhere and remembered when it first came out
I am in /hoem/epica/emc2
Guest329: ok ;-)
I found the whole situation odd, but even odder that certain distros were distancing themsevles from it
eholmgren: I wish people would consider how many USA has murdered in various war's the in last 100 years.. :D
good thing linux came from finland then ;)
ok this functioning
you are welcome
* jmkasunich has to go away now
eholmgren: I just think it's a bit funny :) it's typical american-ish thinking
jmkasunich: laters :)
by the way
merry christmas to you too Guest329
and God bless you
people in Brazil order hug for you
Brazilian hugs - that could be good :)
i from Brazil
3 cheers for jmk
therefore error very
debtor for everything
That the love of Jesus between in its hearts on this day
and that the favour of God always folloies them
Merry Christmas to you too.
Guest329, what language do you speak? Spanish? Portugese?
Spanish more ore less
merry christmas feliz navidad, and ??? in portugese
Spanish more or less
no Felis Natal
ahh! Felis Natal (happy birthday)
NO Feliz Natal is Happy Christmas
you speak spanish?
no, not at all
God bless you
Você trabalhou muito duramente e tem agora o sucesso. Aprecíe o programa e estude-o. Christmas Alegre. (rats, he left )
after all that work ;)
err - before all that work
i said, now enjoy & study
jmk is a saint
jepler is the one who fixed the problem
yeah, sure you did ;)
I'm the one who *wrote* the problem
SWPadnos: you shouldn't encourage people writing problems
keeps people in work though :)
jepler knows I'm only kidding :D
alex_joni: actually because of you I just quit the emc project
jepler: same here
i have a truetype-tracer question...
true-trace ( hydrapoint) or truetype ( adobe) ?
I'm sure someone of lesser skill could finish the rest of emc in php
k, whats the q?
i thought that it is supposed to spit out g-code that emc understands.....
sounds like thats what the page sez, getting deb now
it's in the emc2 repository
sample gcode line---> G01 X [1590*#3] y [3008*#3]
what's wrong with that?
enc won't run it
i'll try it again
[19:07:29] <cradek> http://timeguy.com/cradek-files/emc/ttt-axis.png
near line 4 bad charachter used....(using hello world example
what's on that line?
as you can see in the image I just loaded and ran the default output (hello world)
wow it worked that time.....
* cradek squints at mdynac
this is too kewl......
yes, it works ( examples say 'ttt' but really it's 'truetype-tracer') fbrtfm ;)
edm nameplates for everyone.....
tomp: the packaged man page (man truetype-tracer) seems right
titanium edm nameplates for everyone ;)
so you wanted a job making licence plates? (joliet is not far )
tomp: when I packaged it for debian I expanded the name because ttt was already something else
no problem, it never installed a ttt file
well they would be awful expensive plates....
cradek: thanks, it is a nice app
tomp: thank you
it really is
for edm you would want to do something to join the letters. very easy in cad, but not so easy to hack the tracer to do it
the only problem is that it wasn't in the dapper repo
could just run a base across the bottom....
actually hal could handle a hydrapoint ( hydraulic tracer guiding a machine tool )
mdynac: yes, except for cutting out the centers
random question: you are completely fscked with EDM if you want to cut out the center of a solid, right?
just drill a hole?
mdunac: but you were adding the hole popper & auto threader anyways right ?
cradek: what about the wire?
i would just have to break wire and re-thread to cut out centers....kinda routine in edm, just pop start holes at the right locations.....no big deal...
or is it rewelded?
I think you just thread it
hey EDM is not just wire
backing away from the computer now
well tomp you could whip up an electrode from ttt.....
heh - you could do things kind of like the old days of printing, by sinking letter forms to make name tags (like a Guttenberg press)
EDW burns as EDM
tool has to be mirrored
you machine copper type, and burn it in - you could almost use existing type, actually (they're already mirrored)
you need a linotype machine to make electrodes
actually you use a cnc mill....
to make 'trodes
old wire edm machines had switches to reverse X and tyo reverse Y ( this allowed programs to be mirrored) i bet thats simple to chg the output of emc2
tomp ahh yes the "mirror" function.....
2d : get swapped, not 3rd
mux and multiply, with 1 and -1 as the inputs to the mux
can't you use carbon too?
very nice for making bookends.....
but there's a nasty spike if you're not at HOME when you flip the switch ;)
usa: carbon europe & asia : copper
err - at zero
oh, I see
backing away from the computer now (again)
I assume they work very differently?
depends on your material = different trode type
oh, what can you sink into with copper vs carbon?
desired finish also
and cutting polarity....
paarung: there's pairs opf tool materials and worpiece materials that work well together ( and pairs that dont)
that's german :)
german is the language of edm
well er swiss???
the nicest accent :D
yes nice to hear
alimenation = power supply?
puissance = ?
that's french for power
not if you read that as english it isn't
yes, thats from reading Charmilles stuff
coming right off of my Charmilles schemos
a-l-p-h-a would misinterpret it :p
the 'other suisse'
SWPadnos: re: nasty spike... is it safe with G92?
I don't know, but I suspect it isn't
this is a HAL reversal, not a G-code reversal
is what safe?
negating one or more axis positions in HAL, for mirroring
oh, i see, so a math function, a gcode translation is needed
oh, that can't be a good idea
g1 x[#1 * 1.234]
you'd probably want an X0-X thing anyway
since you want to usethe same work envelope, but backwards
better would be to have two inis, one with negative scales
then you can set home, limits, etc correctly
that's probably true
reboot between jobs? well it's cleaner than all that added code in the file
start emc with a different config, not reboot
yes, bad wording reboot, re-run
anything else is playing with fire
yes, screws up tp
if you really need that ability in emc [and I think you don't] it belongs in the interpreter - a gcode to reverse
for what its worth ...
the trumpf laser I have does this on its controller
robin_sz: where does 0,0 go when you flip the switch?
a program consists of a main block that calls subroutines for each part to be cut
when you call a subroutine for the part, its always in relative mode, and the part can be rotated or flipped
rotation is an interesting problem
thus a nest containg the same part in many orientations consists of 1 main routine with say 100 subroutine calls, and just one part subroutine
you could do that with emc, but the subroutine would have to rotate itself based on a parameter
(which is not that hard, especially if it's generated by a program)
but really, this is the 'virtual C axis' stuff we've talked about before - maybe one day we'll do it
boy, are you getting close to tilting planes ( i'd love to be able to tilt a program onto a new surface )
ive seen other controlers where you can temproarily rotate the frame of reference
tomp: yep, that's "just" virtual A and B too
[19:40:34] <Lerneaen_Hydra> http://wtfbutterscotch.ytmnd.com/
this is a very common feature in laser/plasma/waterjet controllers, for obvious reasons
i'm trying to list the obvious reasons... like i wanted the cutout on this side or that side is one... mating lefts & rights, ups/downs...
Ive used it make a female electode to burn a solid heeled punch
I saw on a cenotroid where they did a offset
and machined the same part that way instead of straight..
in truetype-trace, how do i change the ttf font?
tomp: some commandline option
man tells everything
% No manual entry for truetype-tracer
? SYNTAX ERROR
i got it, but i just installed via pkg manager
tomp, on laser and plasma etc there is one very obvious reason ... you have just placed a 2mx4m sheet of 20mm steel on the bed. now, would you rather try and shift it into the exact position for 0,0 and straight, or type a few commands on the console?
offsets are simple enough, the rotation is the nicest touch
I called a friend to see how he was doing
he were pondering drowning himself in dough
wtf? dough as in?
robin_sz: you're right ( i just cut my hands trying to turn it!)
I find that hard to believe
20mm sheet doesn't usually cut :D
you have a nice shear
and you have a nice year too :)
Lerneaen_Hydra: stuff you make breads off..
Lerneaen_Hydra: err bread..
you mean literally drown himself in dough?
yeah, as in suicide..
he were bored at work..
most lasers can figure out the sheet poition (angle and offset) all on thier own
what is he, a baker?
Lerneaen_Hydra: more like a bread factory worker..
Lerneaen_Hydra: bread passing by on a conveyor into a oven
they assume you have got the sheet within +-50mm
robin_sz: what do they sense? reflectivity?
lower head to 2mm above sheet, 50mm from where the edge should be
turn on height sensor
disable height feedback to Z
what is that based on?
(the height sensor)
move towards sheet edge, watch height signal for when it drops off
you found the edge.
the usual capacative height sensing from the torch tip
[20:30:11] <anonimasu> http://www.gogglemarks.net/index.php?action=display&tag=fightswitch
wow, i had this crazy plotter from israel, it found the edge of the sheet that way, oriented itself, plotted,parked itself back in it's garage
(searched for edge, not capacitve , was optical )
our hp plotter does that.
finds all the edges of the sheet
anonimasu: i love the idea of a switch that works when he 'threatens' the machine
i built such a thing back in college, an architecture course wanted us to design a random motion machine
the jury was a room full of junk spinning and ping pong balls flying
I think im going to build one..
i came in with a campbells soup can & set it down
it screamed when you got cclose
there was a crowd around it balancing chairs near it, throwinf pens past it
and the professor leaned over and asked me, where's the random motion?
everywhere around it ?
does youtube work for you?
but it's slow
very very slow
[20:42:26] <alex_joni> http://www.youtube.com/watch?v=bXqI9wDLXMc&NR
meh, would have been better if they had told the pulling force
very impressive ( but in the US, the sales were hurt by the name -- sounds like toe-rag )
it won the darpa grand challenge though :)
well, the electronics did
im more impressed by the bike that almost finished
they had a motorcycle in the race
[20:59:57] <anonimasu> http://www.ghostriderrobot.com/
[21:01:11] <anonimasu> http://www.scipy.org/Cookbook/Matplotlib/Maps
cool stuff :D
cradek or jepler: axis uses opengl doesn't it?
jmkasunich: yes, why?
it just occurred to me that the 1-2 frame per second results I saw in glgears mean the axis rendering will also be slow
(matrox millennium, no hardware GL support)
so I don't want crappy slow display
I run without hardware GL on my nvidia too
I guess I can always upgrade the video
get a different video card ;)
glxgears is also 3-4 frames while full screen
but AXIS is more than enough on the rendering
skunkworks: what else?
I think I rand a matrox for a while
opengl cards are always $$$
it's not a matter of opengl cards
I have a quadro at work
but getting the driver to work while there's a RT kernel installed is a whole different problem
I have run quite a few different cards. maybe I don't know how fast it _could_ be
I am normally basing the speed of the system on base period. The pluto will be a different experience ;)
skunkworks: jmk asked about graphics performance
unrelated to base_period (well sorta anyways)
it's too bad that software rendering is so slow
Lerneaen_Hydra: why is it bad?
graphic cards manufacturers need to make a living :D
how do you run glgears?
should be 10-15 or so :p
type glxgears in a terminal
my portable does 9 seconds between each move full screen :)
and here I was thinking 3-4 fps was bad
0.11 is really bad
dang... you think thats bad
I just went back to the shoptask system to try running glxgears again....
I'm used to 30-40-ish with hardware acceleration
oh, glx should have several thousand
with hardware accel
it ran very nicely (20+ frames/sec) for about 3 seconds, then it went to 1-2 frames a second for 10 seconds, then X restarted
not so good
jmk: same here, but x didnt restart
jmkasunich: opengl crashes X? ouch.
does glxinfo say direct rendering: yes?
I think thats what happened
went back to the graphical login screen, after some flickering and such
nothing in dmesg
where does the xserver keep its logs?
my glxinfo sez no
jmkasunich: /var/log/Xorg.0.log.old may be the one you want
no here also
in reality you probably want "no" for a realtime system
(usually you really want yes)
have you seen it as yes on a rt system?
[21:30:24] <jmk-st> http://pastebin.ca/289109
yes my machine has it
if I run one opengl app, it's fine, two gives latency errors
jmkasunich: pretty useless...
so many words, so little meaning
why is it trying to set up a tablet, and stylus, and eraser, and all that crap?
in case you have a wacom tablet
you could disable all that crap, but it's not the problem
does glxinfo say direct?
the machine that I use emc2 on does a frame every 12 seconds. axis runs just fine though.
p III 600
direct rendering no
...and here I was thinking I had issues
skunkworks: yeah I've seen that too
I think it might be a glxgears bug actually
glxinfo result: http://pastebin.ca/289118
jmk-st: in that case I'm very surprised and you should try to reproduce it.
98.3% of that I don't understand
btw, eariler I ran 3 glxgears at once
very slow, but they didn't crash
you ought to change your depth to 16 - it will be generally faster
thats in xorg.conf?
cradek: i tried unicode in truetype-tracer: Fatal error in FT_Get_Char_Index: no error (0) at line:194 can I get an example?
tomp: maybe there is no unicode map in your ttf file
if I comment out "Section "InputDevice"
Driver "wacom" (the whole section) that should be safe, right?
jmk-st: check ServerLayout too
k, i think i ran into that before, some ttf font dont have them & there were tools to generate it
what is section "DRI"?
that's something about direct rendering - don't think it matters in your case
a hardware 3d extension
that's the bad spelling for three
direct rendering interface
the only item there is Mode 0666
the number of the beast
leaving that alone
good thing guest isn't here
under ubuntu, whats the easiest way to restart X?
the beast is RW
that one :D
at 16 color depth I get a frame every 5 seconds ;)
I suppose I should decide which channel I want to talk on and stay there...
(that was on my protable - 9 - 5 seconds
the machine computer went from 12 to 9 seconds - it did spin the gears really fast for a bout a second.
Today is a very good day, I got my embedded gentoo linux running with xorg, 20.5MB size on disk for the whole thing :D
This thing with EMC sounds like a whole lot of fun for embedded controllers :D
alex: thanks, I will share the steps in a day or two when I have sorted and cleaned my notes up
no hurry.. I can still sleep in peace without knowing that :D
alex: hehe yup
good night all
so ubuntu is controlling my fans. good
skunk: yup, 6.10 is the first linux disto ever where fan control for my laptop is better done by the OS than BIOS, they also do great cpu throttling based on the heat
I am wondering if the smi? patch will effect it.