03cradek 07TRUNK * 10emc2/src/emc/task/emccanon.cc:
fix a corner case, in tolerance mode, for two consecutive moves where
the second one ends at the same point as the first one starts (like a
drill down and back up).
03cradek 07v2_1_branch * 10emc2/src/emc/task/emccanon.cc: backport rev 1.95 fix
cradek: that was quick :)
I came, I saw, I read, I got confused, and finally I wiki'd
[02:55:42] <a-l-p-h-a> http://www.simonpanrucker.com/beans.html
a-l-p-h-a is now known as a-l-p-h-a_
Ziegler: I haven't done anything more with that program since I posted it. the source code is there for you to download if you care to improve it.
welcome back jepler!
thanks and goodnight
wow - many Chinese items just jumped in price
chinese items? like machinery or fine china?
CHICOM .gov just cut subsidies from 17% to 5%
machinery was top of the list
July 1st was the change date
Syil just bumped there machines $1000 each
isn't that like a 50% increase?
and there is a re-valuation of the RMB coming - so another price jump is expected
the SX3 and C6 went up $1000
SX3 went from $3850 to $4895
not sure of the C6 price
that's almost 25% increase
guess that will make Emco stuff a lot more attractive
Petrol prices have given the dollar a beating - and now the chinese are finaly reacting to the loss of dollar value
any issue using xubuntu 7.04 + emc ?
Jymmmm: only missing a RT-kernel, which you probably need to build yourself
alex_joni: is that any different from doing an apt-get install emc?
or are you saying the 7.x part of it?
we have a kernel that works on 6.06
I suspect it might work on 7.04, but wouldn't be so sure
when I ran the 5.10 kernel on 6.06 it didn't quite work as it should
it ghought you werw talking about Xubuntu for a moment
nah, xubuntu is a non-issue
alex_joni: you think xubuntu would give better ovl max on RT test?
hmm.. you can test a non-X version first
see if it helps
console only you mean?
if that doesn't improve things, then xubuntu is likely no help either
is there a console based client?
and you can always run the GUI remote
Hmmm, how do you get rid of X in ubuntu?
alex_joni: oh, true
alex_joni: defaults to gui login
Jymmmm: ctrl-alt-F1 to switch to a console
then sudo /etc/init.d/gdm stop
alex_joni: ah, ok
alex_joni: is that temp?
a reboot will cure it
or sudo /etc/init.d/gdm start
alex_joni: all these fscking daemons, three schedulers,etc... such a pita
remember thing all that is =)
Jymmmm: I know..
that's why I like ubuntu.. things just happen to work
heh, I prefer debian, but, well, you know
* alex_joni has a new favourite tune
same sh*t really :P
I have debian on my servers
but ubuntu is great for desktop
alex_joni: all the dependancies being linked. I apt-get removed totem, it took all of gnome with it
and for computational challenged people
Jymmmm: that sucks ;)
alex_joni: Yeah, just a test to see what would heppen, glad it wasn't the real deal.
ever heard of Lissie?
Lissie - Bright Side
sounds like an actress
is there some simple way of making G61 default?
Vq^: you can put it in the ini
alex_joni: not my cup of tea
under startup code
Vq^: the exact ini name is somewhere in the manual
RS274NGC_STARTUP_CODE or something like that
seems I got the name right
it's under [RS274NGC]
hmm.. I'm not sure under which section.. there are examples with both :P
* Vq^ grep(1)s for it
it should be in [EMC]
demo_mazak and dallur-thc apparently
Alex got that page up - took awhile for me to get past stupid mistakes, but I got it - http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Servomate
good for you :P
looks good to me
just hope it makes sense to others who read it.
Guest884 is now known as richarda
Skullworks-PGAB: just read your wiki page on the servomate with interest
semms to work the same as my Digiplan amps on my KMB
have you had any success with your set up?
'lo, anyone here?
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-07-22.txt
cradek, when you come by later, I've found some very strange behaviou
r in backlash comp
I think I heard about it before, but that it was supposedly fixed
it appears that (this only happens when backlash comp is activated, ie backlash value !=0 in ini), that when moving an axis in it's positive direction all is well, and backlash performs as it should, but when moving the axis in the negative direction a movement of 0.1 mm moves the table by less, in my case by 0.098125mm
in my case backlash was set to 0.1mm and I was taking steps of 0.1mm
it appears to be related to the amount of backlash set, as another axis that had backlash 0.08 had much less deviation, around 70% of the deviation
it could very well have been 80% of the deviation too
alex joni:r u here
ler_hydra: if you use halscope to view the motor-pos-cmd, does it change by the expected amount when you move in each direction? I tried in sim while looking at axis.0.motor-pos-cmd and it seemed to behave as I expected.
jlmjvm: I am now
alex_joni: I am getting a limits message when I run a program,but the program and limits are right,is there any way I can turn this message off?
is the program small enough to put on pastebin? (or can you make a smaller program that demonstrates the problem)
it's the message shown by AXIS that says "Program exceeds limits on axis Z, run anyway?"?
there's not a way to disable that without changing the inifile limits or editing the axis source code
it's known that you can erroneously get this message in programs that use tool length offsets -- if you aren't using tool length offsets, I'd be interested in seeing the code that makes it happen
alex_joni: There are 2 messages,1 is the warning,the 2nd tells you the line number and which axis,I only get the first
i am using tool offsets
the programs run correctly,just have to hit run anyway every time
i can pastebin a program
the tool offset problem is known but difficult to solve
If you build emc from source I can tell you what to modify to get rid of the warning altogether though
if you use the package you could also modify a line in the file /usr/bin/axis with a text editor, but this will be overwritten when you upgrade to a new version of emc
[15:49:30] <jlmjvm> http://pastebin.ca/629677
in emc 2.1.6 change line 2251 in src/emc/usr_intf/axis/scripts/axis.py or /usr/bin/axis to read "if 0:" instead of "if warnings:"
make sure to maintain the spaces at the beginning of the line exactly the same as before
in the development version, the line 'if warnings:' is around line 2093
jepler, here now, I'll test that
sorry,let me roboot into linux right quick
jepler, yes it shows the correct value, ie multiples of 0.1mm
some very minor jitter in the last digit, but I assume that's always there
anyway, it's nothing I'd be able to see with a dial
a couple nm
if the motor-pos-cmd is right, it seems to me it's either mechanical or maybe due to the finite size of a step or encoder count
what is the SCALE?
I don't think it can be becuase when I turn backlash comp off the problem dissapears
you can't accurately represent 0.1 with 625 steps per unit (you'd need 62.5 steps)
yes but the error isn't a jitter
it could be a jitter but that I haven't gone past the "step" when it hops down to the next level
if you get what I mean
unfortunately my dial doesn't have that much throw
hmm, no, wait, the problem dissapears when backlash is disabled
servo or stepper?
perhaps I don't understand what you're measuring
200 spr, 5x microstepping
ler_hydra: you are measuring actual physical travel, right?
a dial with magnetic base attached to the table with the end touching the spindle
your step resolution is 0.0016 mm, and that assumes that microsteps are perfect
the jitter/error is on the range of 0.01mm when traveling 0.8mm
note that 0.08 is an even multiple of 0.0016, but 0.1 is not
0.015 or so
you're reporting that when you asked for 0.1 you got 0.098125, an error of 0.001875 if I did the math right
jmkasunich, that's true, though I did the measurement over a travel of 0.8mm to get a more accurate reading
when you say jitter, do you mean that the machine is bouncing around, it doesn't stop at one particular position
no, jitter probably isn't the right term
heh - jitter is what you get when you have too much coffee
I'm sorry - I didn't understand the last part of that sentence :)
(specifically, the last 3 words)
what should be an equal travel distance in both directions wasn't, in one dir. the machine moves as commanded, but in the other it's as though the scale is off a bit, when I command a movement of 0.8mm I get 0.785 or so
though when I think about it it could be that the acme screw is badly machined
and that the error isn't accumulative and just happens to look like this right where I am on it right now
to measure in the positive direction, do you do this: 1) move some short positive distance 2) read the dial 3) move the desired positive distance 4) read the dial
yes I move 0.1mm 8 times, and read the dial between each
what happens if you move in one direction, then reverse direction, then measure between successive moves in the reverse direction?
ie, ignore the reversing move
if you are making 8 moves in the same direction, backlash comp should have no effect after the first one
put a halmeter on joint.<whatever-axis>.motor-pos-cmd
it seems to be accurate to 0.005mm or so when moving in the positive direction, when going in the opposite there seems to be an accumulative error (like if scale was set to a wrong value), there's about the same amount of high-freq deviation (deviation betwwen each move) though over 0.8mm of movement there is a total accumulative error of 0.015mm or so
motor-pos-cmd is after the lash calculation
what is your screw pitch?
probably cheap crap
can your dial travel 2mm?
do you have a 2mm gage block?
.9 or so
wow, thats a tiny range
it's one of those with a arm that moves angularly
gage block = very precisely ground block exactly 2.0000mm thick
oh right, don't think so, not here
the angular ones can't really be used to measure distance
because there is a cosine term in there as the angle changes
jepler: got my linux booted now
oh, found a linear one now
they're normally used to compare things, such as "is this part centered"
yeah I've used them before
you can use them with gage blocks to measure distance - zero dial, command 2mm move, insert 2mm block between dial and object, read dial
this shop isn't as well stocked
I found a linear dial now though
it would be interesting to measure over multiples of 2mm
that would eliminate any periodic error in the screw
or screw mounts
how do you get scale 625 with a 2mm screw, 1.8 degree motor, 5x microstepping?
jepler, not 1:1
oh, belts or something?
theres a timing belt too, not that much ratio though
it would appear that I was just in a bad region of the screw
keep in mind that the belt may have some small hystersis or stretching
you are looking at very small deviations here
as over a longer travel it averages out
the screw was rather bad, deviations of 0.03mm or so :/
ah well, it was cheap
depending on how critical this mahine is, and how much you care about it, you might want to do screw comp
it's not that important
this is just for a electronics-club at college, 0.05-0.1mm accuracy is all thats needed as of yet
maybe they'll get ballscrews for it later
..and servos and a better motor and and and...
where do i make the change to turn off the warning?
jlmjvm: what warning?
i lost what you said when i rebooted
jlmjvm: oh, jas
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-07-22.txt
jlmjvm: load the log and scroll down to 15:49:40
that was cool,i tried to copy and paste to my thumb drive but it wouldnt paste
jmkasunich, jepler thanks for helping to sort that out
Guest904 is now known as richarda
jepler: i found the "if warnings" in usr/bin/axis,do i just change that to "if 0" ?
jlmjvm: yes. make sure to keep the space at the beginning exactly the same, and keep the ":" at the end
not wanting to let me save change for some reason
jlmjvm: if you are editing /usr/bin/axis, you must do it as root (administrator)
in a terminal, you can run something like 'sudo gedit /usr/bin/axis' to get a graphical text editor that is not too likely to mess up the whole file
seems to be working
hmm, funny, could someone test running emc with backlash, relatively low accel (say 70) and hammering the jog button for that axis (in one direction only)
I seem to be able to get a speed higher than maxvel
(continous jog mode)
sometimes it goes over maxvel
I am wanting to put a us digital 5v encoder on my pacsi stepper,can the a,b,and z inputs be hooked up to 3 inputs on a parallel port breakout board
without any other hardware
input number 12,13,and 15
jlmjvm: yes, but the max count rate of the prallel port might not be too good
what kind of rate will it read?
it will be a 500 cpr encoder with index
the absolute maximum is 1/(2*base_period)
not over 1/base_period
in practice it might be a lot lower than that
awallin, aren't they quadrature?
my base period is 10000
2000 counts per rev
awallin, where's the 2x from?
if you sample at a frequency f then the highest you are able to measure is 0.5*f
jlmjvm: so if you are generating steps also with the parport you will be able to measure encoder counts at roughly the rate at which you are generating steps
oh, yeah, but one step on the encoder is half a period if you look at it like a wave
i can generate way more steps than my motors will run
awallin, only if he's running 10x microstepping
if he's running full stepping then he can only sample at 1/10th of generation speed (assuming 200spr)
can generate steps to run 250 ipm,but can only run motors at 100 ipm
im 10x microstepping
and 200 steps per rev?
with the 10x microstepping included I take it
if so then that should be ok
im wanting to do this so it will give me a ferror if it goes out of position
03cradek 07TRUNK * 10emc2/src/emc/task/emccanon.cc: fix problems with very short linear moves (1e-8) combined with rotary moves
03cradek 07TRUNK * 10emc2/src/emc/kinematics/ (tp.c tc.c): fix problems with very short linear moves (1e-8) combined with rotary moves
Skullworks-PGAB: on your servomate you could use index-only homing
it involves jogging the axis to a certain marked location (mark is between two index pulses) and the homing does only a search for index from that location
no switches are needed and no limits are hit
it's new (only in trunk)
cradek: will the index only homing be available in the next release?
jlmjvm: I don't think it will be added to any 2.1 version, but it will be in 2.2.0
this homing method is axis independant - can do only 1 per command?
if you'd have all your axes on their marks, you could use "home all" and have them all search together
So run everything into the limit switches, use the MPG to back each off just enough so the prox switches are not tripped, then issue "home all".
thats kinda like how a Fadal homes out,line up the pointers,then home
Skullworks-PGAB: I'd stay away from switches - just line up some marks like jlmjvm says
it finds where it misses
no way to mark the Z axis
* cradek hands Skullworks-PGAB a sharpie
its a quill machine - kinda like the Mazak
is the quill filly retracted @ z=0
yeah - its up flush in a seal/wiper
can you add a home switch for z?
the movement is only about .05" between limit and home
good enuf... pullback.. mack proxy.. pu forward SLOOOWLY... monitor prox til t loses signal
i dislike proxies ALOT... ive seen fadals home 5MM out
x/y use prox - Z uses rollers micro switches
with the pointers, if you make your final home position be the one where the pointers line up, it's easy to verify the home afterward
that seems kind of nice
make a button type switch that togle ove as the quill disappers into the headstock
sorry i seem to be missing fingurs today
on most machines without scales its all hard switches - 3 axes
The Mori Seiki (monster) has lines like that - because you have to recalibrate the home position if you let the backup battery die in the control unit. (lots of fun!)
* anonimasu hugs linear scales
Fanuc 18 (pre i)
dead baterries in Fanuc's can be a PAIN...
ESPECIALLY if its in the pseudo axis drive... called the tool changer...
the i's arent alot beter as far as im corncerned
yeah I'm lucky that way - the tool changer still just does 1 360 to find pot 1 and a switch zeros it out - its not part of the battery backup
the newer ones...
some are run as axes... pot positions are at co-ords... you know...
siemens840d looks at EVERYTING that moves ang give it axis data ie.. pos,kinematics,max/mins limits..etc MONSTER CNC PC
this one is almost 10 years old - and its still scary fast 1250IPM rapids, 9.2 HP servos, 30hp 10K spindle
nice size machine
thats a rigidtapping machine
did a 1.25"-7 in a die plate -
try to peck tap with it... ; )
ive blown guys away doing that
and with the proper parameter set you can rigid tap using the Z MPG
never seen that before
toshiba tosnuc 888 does it well also
will emc do that?
1"-8 x 6" ful thread in cast BOILERplate
but it would require the spindle to be controlled like a servo
36 holes per part 1 tap per part for safety
so it would be a hardware limitation for most
I threadmill everything
it has to have angular resover or encoder on it
1"- 8 tpi x 6" dp... ???
love that macro for NPT threadmilling
not that - too deep for that sized pitch
i thd mill alot of stuff...
when I get my little machine done and running EMC - I will start a user macro library
ive used single pointers... full thd npt cutters....one turn... drill & tap 1 turn.. was cool 5 mm
m5 has a 4 mm tap dr... for shallow thru holes.... 4mm tool with dr at lower prtion and tap form above.... dr thru to thd for... 4mm dia... then interpolate a 1mm dia circle and pitch up 1 turn
I ruin Haas at work
and I run Okumas
get outta dodge... m5 x 8 dp in 5 seconds per hole
okk , wfl , vdf boeringer , Tachii , la la. la
drilling holes in 3/8 plate - 22mm dia - 3sec per hole
how are your burrs??
.032" IPR feed
that's a scary cut
has a perfect reamed finish
slight exit burr - just hit it on a C-sink
of course your punching them with a rotary die...
drills are like $730 each from Kennametal - solid carbide
drilling in the Mori
just keep sharpening the dril back.. get a few spares cause itll wear the margin up in a hurry
coolant thru tool at 800 psi
and they cutt almost silently
all you hear is chips hitting the enclouser
sandvick are good in drilling too
do they sound like nickles hitting the floor yet???
owner knows several old time Kennametal brass - they use the shop for testing stuff
i had a boss loved to get TOO close to some equipment ... Cuase he liked the sound of the nickles hitting the floor
had an insert drill show up
like whats this for
COOL.. i used to do alot of ISCAR tesing while at toshiba
muco gracias... is all id say
kennametal rep shows up later with a 6 foot section of rail road rail
lets make some holes???
wants me to drill the holes the use to rivit sections
so I do and tell him his drill sucks
insert only lasted 27 holes
he is just grinning
says "That just fine, cause they aer only getting 4 holes with our competitors inserts>"
yeah... yhat was work hardened rail he brought you wasnt it
seems that the iron workhardens after having trains run over it for years
its meant to
and all these rails were being recycled - welded into mile long segments
only partially h/t at forming... to allow the workers to bend corners
cuts down on noise and safer - less chance of derail
they are called back ins
skullworks:I mentioned my KMB1x earlier, I can home okay, are you having problems?
cause the train backs into the steel mill and rail is laid on the flat bed and the train moves out until the rail is a mile long - then the train backs in and does another
Place I used to work - still do some programing for them - has a KM3 that died about 2001
I ran that machine from the day they put it on the floor in 88 until it died
the iron is mint
but its dead
and the owner just don't care
it sits there in the shop being used as a place to store extra Kurt vises
I've made cash offeres and the guy just ignores me
won't say yes or no
I'm still tryin
machine powers up and can jog
but both displays are toast
kyb is marginal now - its gotten gummed up
it begs for an EMC conversion
I have the facory manuals and wiring diagram
but I'm not going to touch it unless I have clear title
so the issues you described with the servo amps are theoretical at the the moments?
I got quotes of about $8000 to repair it using OEM parts
yes - sort of - I was talking with someone else who ways trying to convert a KM3P
they were having the issues - this was shortly after 2.0 came out
although my amps are not the same,
KM3P had a servo which moved the varidrive to set spindle speed
I think he just pulled that and put a handwheel on it
the function you describe applies to them and
bet its the same electrocraft servos
I have it working
used also on BMC20's
no, by a company called Digiplan, take over by Parker Hanifin some years ago
thats the amps - or amps and motors?
amps. motors are Electrocraft
expensive they are - real proud of those motors
for homing you have to keep to the 'live' side of the limit swich as you describe,
but if you keep the speed down'
you can keeo the ferror quite small
I can keep min_ferror to 0.002 thou,
for me quite workable
well for now its a moot point
keep on at him,
I just wanted to get issues sorted out should I ever get the owner to budge on the sale
it is almost scrap with the cost of hurco's stuff
it was appraised @ $1200
was the vari-speed servo or just motor driven
and using oem parts, the parts cost alone was $8000
the KM3 is just like a Bridgeport except the front of the quil is all inside the casting
the varidrive is hand cranked
KM3P had a motor which turned the varidrive crank
need a pic?
same as mine, tho there is controller board to give some software speed control
I think the KM3 just a bigger brother to mine
here is a pic of a KM3P head http://images.machineryvalues.com/pict/133537h.jpg
looks identical to mine
this was from an ebay auction http://cgi.ebay.com/24-CNC-VERTICAL-MILL-Hurco-KM3P-Ultimax-II_W0QQitemZ150141687384QQihZ005QQcategoryZ12584QQrdZ1QQcmdZViewItem
there seem to be some nice improvements coming up for indexing but,
even now, should you get it,
I think you can make a workable set-up.
I know I can
but I have to finishe the little machine first
then build another building to put everything in
yeah I'm converting a Grizley/SEIG X3
kinda a micro bedmill
Macht_Tot is now known as Fencer
why does stepgen.c bother with a "old_pos_scale" variable? Are runtime changes to INPUT_SCALE permitted?
for the most part, all parameters can be changed at runtime
runtime changes to stepgen.N.scale are permitted
stepgen is NOT constrained to use by EMC only
one could have a variable pitch screw
and have scale based on position ;)
that would require scale to be an pin, not param..
or an online setup application, with the system still "running" in the background ...
'variable pitch screw'?
is that code for 'broken screw'?
alex_joni: I know you were only joking, but that wouldn't work anyway
since the scale is absolute, not incremental
This seems weird: "ShipIt Unfortunately, unlike the other Ubuntu derivatives, Xubuntu does not yet have free cds available for shipping due to lack of funding." Sounds like it's not even of the same project as ubuntu.
it's not. it's a derivative project
SWPadnos: So, not even the same backers as ubuntu?
they have some support from canonical, I think (there is after all a link on the ubuntu.com website)
ah, never realized seeing the Kubuntu has some $ as well
pressing of cd's is cheap enough, suspect it's the shipping
SWPadnos: have you ever seen how they make cd's? Very cool
I think I read about it shortly after the CD was invented :)
CD's are pressed, not hit with a laser or anything
sure - I knew that
similar to a record, only higher resolution
good night all
Making the master is a different story
and then there's the sandwiching of the foil layer between the plastic layers
there was a presser on some site recently
20k$ or so
I've seen them on eBay
you could do about 10k CDs daily
alex_joni: thanks for the tips on the init stuff
alex_joni: almost 50%
17000 vs 10000
that means emc with remote display might work way better/faster
alex_joni: but I'm not sure how to tax the cpu via cli
SWPadnos: I'll try to get you the video sometime
hmmm. where are these init tips? :)
SWPadnos: oh, disabling X /etc/init.d/gdm stop
sure, I know that one :)
I've done some testing with and without X as well
that was it, in respect to RT test
on the NForce 6150, it's crap either way
with CLI, you can switch to another terminal and do something intensive, like a large compile
SWPadnos: like I compile anything =)
flip to another terminal and do something like find / -name "*fred*"
SWPadnos: can't you disable the integrated graphics on the nforce4?
you can, by installing a video card
which I didn't feel like doing
which results did you get with the integrated vga?
SWPadnos: Got bandwidth?
this is my file server, with integrated DVI video
JymmmEMC, not really - only 2 Mbits/sec
and why is it controlling a cnc machine?
SWPadnos: is that up or down?
lerneaen_hydra, it isn't, I was testing hte chipset :)
I need to put together 3 or so RT computers soon (like next week)
SWPadnos: that should be fine (you can just pause your p0rn downloads)
so I'm trying to find out what might be usable
SWPadnos: aint that a bitch. I've been trying to find the good RT test mobo's for a long time.
JymmmEMC: haha, see how he paused there after you said that
oh - missed that one :)
lerneaen_hydra: Yep =)
looks like you got him
I'm actually downloading ubuntu updates on another machine :)
I'm married, I don't need pr0n
SWPadnos: Oh, is that what you kids are calling it these days
or so they say :)
SWPadnos: You're married, you DEFINATELY need p0rn
keeps things spicy
depends on who's watching
and who's watching the watcher
[22:24:20] <lerneaen_hydra> http://ceilingcat.istheshit.net/
SWPadnos: Hey, well, if exabishionisim is what keeps it spicy, go for it!
anyway - what have you found regarding RT and any motherboards you've tested?
I never understood lolcats.
preferably recent motherboards
maybe I'm just dead inside.
SWPadnos: None of my mobo's are "recent", all P3's
[22:25:26] <a-l-p-h-a_> http://lolcats.com/
ok, nevermind then
a-l-p-h-a_: oh, those
I want a intel E6750, with an MSI P35 platinium mobo, and 2 gigs of ram. :/
SWPadnos: I have found that if you disable things in the bios, it helps. serial ports, secondary IDE controller, etc
yeah. I'm trying for an industrial kind of mobo, since this is going inside a large power supply
there's actually a budget, but not enough time
SWPadnos: what about the itx mobo's ?
SWPadnos: without needing a discrete graphics board?
ITX is OK, but there are issues with changes
SWPadnos: phyiscal changes?
like, order the same part number and the EMC disk fails to boot on the new MB
SWPadnos, how about a metal shielding around the mobo? like a good computer case?
needing a discrete board would make the entire setup not nearly as discreet
discrete graphics is fine, the controller doesn't need any local UI
SWPadnos: Just curious, have you tried burning a emc iso at 4x and see if you get the same results?
shielding will be there, in the form of a PC case :)
Ray had that problem
got great results on an ITX board, then couldn't boot on the one he ordered ~4 months later
same part number, different layout/chipset
Jetway, as I recall
(just to know to avoid them)
SWPadnos: tried shuttle?
well, I think that'll be more common than consumer manufactureres that bump rev numbers consistently
I'm testing out a Shuttle XPC right now, actually - that's what's downloading updates :)
SWPadnos: we REALLY need to have a page of mobo's and test results.
it seems pretty good - ~13000 max latency in X (nvidia card, I haven't checked the driver yet)
I know - I've thought about it a lot
SWPadnos: onboard video?
I had intended to try out a bunch of MBs at ESC when I was there, but it didnt work out
nope, 7600GS I think
SWPadnos: we need a criteria list though, so we have an even playing field (w/o tweaking)
bios set at defaults, etc
well, I'd say that BIOS at default and then "best found" would be good
BIOS version, on/off board video, etc
board revision <---- BIGGY
I'll be doing some more rigorous testing during the week. I just need a "short list" of boards to try so I can get them bought for me :)
too bad the wiki doens't have photos.
SWPadnos: I have right now two tyan mobos, one has Promise raid, the other doesn't. both are S2518, there is no designator on the mobo P/N wise other than the missing chipset and IDC connector
nothing on the side of the last slot?
there's often a sticker there
gimme a sec to pull the old mobo
No, I even looked at the stickers on the chips - no difference
Even the bios still has the "enable SCSI" option in it, but neither mobo has SCSI.
scsi i, ii, or iii?
a-l-p-h-a P3 server mobo
SCSI still stopms all over IDE and SATA for servers
SWPadnos, raid 1, raid 5 scsi iii vs raid 1 or 5, SATAII hmm... I'd like to see a price/performance comparison.
I have shitloads of dual P3 mobo's around here. This one supports up to 4GB
how many P3 cpu's do you have?
price/performance isn't so obvious, but in terms of performance, IDE/SATA can't compare
a-l-p-h-a I got spares
what's the transfer speed of scsi? burst and sustained.
U320 is 320 MB/sec
I think FC is higher
[22:40:12] <SWPadnos> http://storagereview.com/php/benchmark/bench_sort.php
Hmmm, where can I upload a 180MB video?
select IOMark with some relatively high number, like 32, 64, or 128
gack - hold on a sec
JymmmEMC,, split it to be less than 100mb / file... google.
[22:41:00] <SWPadnos> http://storagereview.com/Testbed4Compare.sr
a-l-p-h-a no tools to split it
on this machien
the 128 I/O benchmark shows the WD Raptors as the best performing IDE/SATA drives, at 222 IOs / sec
JymmmEMC, ask SWPadnos for an shell account on his dreamhost.
the highest performing SCSI drives are 411 IOs/sec
or close to twice as fast
[22:42:49] <a-l-p-h-a_> http://www.tomshardware.com/2003/11/14/raidcore_unleashes_sata_to_take_out_scsi/
SWPadnos: got ftp upload?
SWPadnos: gimme da credials baby!
I got a SATA RAID card for my server, just because SCSI is still quite expensive
how much downloading do you expect? :)
SWPadnos: No, I just wna tto upload this video to you
SWPadnos, these are from 2003!!!
I don't think so
SWPadnos: was that to me?
if you upload it to him... arnt others going to download it?
it would have been hard to do Office DriveMark 2006 testing in 2003
lol, good point
Ziegler: no idea
what's this video of, anyway?
SWPadnos, tom's http://www.tomshardware.com/2003/11/14/raidcore_unleashes_sata_to_take_out_scsi/page37.html
review of scsi vs sata, in raid situations.
SWPadnos: p0rn of course (kidding) pressing of cd's and how the master is made
you have kiddy porn!
you sick bastard!
a-l-p-h-a you gave it to me
do you still have the one with the goat?
you have transcode?
Ziegler: I have whatever ubuntu comes with
but no sound card
you might be able to shrink the vid a bit
you know, I'm not sure I'm inclined to believe a review with this caption:
I just want to start the upload so I can start getting ready for work
I need to choose a pizza. My regulars:
A.) no tomato sauce, bbq sauce instead. pepperoni, grilled chicken breast, bacon strips.
B.) spinach, feta, grilled chicken breast
C.) open to suggestions
"Too large, too slow: The hard drives from Seagate (ST3146807LW) held up our tests because of their large capacity of 146 GB. This situation prompted us to switch without delay to models running at 15,000 rpm (see below)."
a-l-p-h-a chicken with garlic sauce
ok, time for dinner. see you guys later
a-l-p-h-a_: pineapple, mushroom, black olive, banana or jalapeno peppers
SWPadnos: ftp info?
cradek, that sounds... odd.
maybe it's regional - yours sound odd to me
JymmmEMC: go to terminal and see if you have ffmpeg installed
cradek, fair enough
Lemon Pepper Chicken pizza with garlic sauce, red onions, cilatro, zuccini, carrots
carrots?? are they soft?
or crunchy on top?
a-l-p-h-a_: do we both agree that Jymm's sounds more odd?
maybe it's best not to ask for help
I could go for a delux... but that's just boring.
pep, green pep, mushrooms.
cradek: a-l-p-h-a: http://strawhatpizza.com/menu-pizza.htm
It does sound weird, especially the zuccini, but tastes really good. especially the cilatro
wth is Canadian Bacon??
something like potato onion cilantro sounds nice - very indian tasting
I think it's a thick ham
with cornmeal around it?
don't think so
does anyone else like pineapple on pizza? it's common enough around here, and good
yeah, I like pineapples, bacon, and extra cheese.
we call that a hawian.
pineapples, bell pepper, canadian bacon
that sounds good...
do european patents restrict north americans ? ?
Unit41: I patent has to be field in each conutry for it to be effective in theat country.
what if its filed after you've allready sold a bunch of it
I'll keep my SCSI thank you.
hang on... WIPO...
Unit41: No problem, as long as you are not infornging on someone else's patent
Unit41, I'd look into patents, instead of getting it from a bunch of jar heads like us... as we have differing info.
I wana start building those vortex hydro dams
JymmmEMC: type "man avisplit"
weren't those part of some government study?
safe for the fishies
Unit41: Though, here's the (reality) catch... Yo MUST fight for your patent, meaning that if I imfringed on your patent you have to take me to court, if you don't you will lose it
low speed vortex hydro damns.
if you are still lookig to see if ubuntu can split vid for ya
Unit41: surely you should ask your lawyer about such an important thing
I'm filing for a patent as we speak.
on what ?
Unit41: So, unless you have the legal financial backing, patents only make the lawyers rich.
patents are as good as the money you have to defend it. So if BIG COMPANY wants to take your patent... you're kinda SOL... you should just sell for millions, and get a job there too. :)
Unit41: Figure a lawyer and lawsuit cost $20,000 each time you sue someone, you better make sure that you got the backing.
I'll just email them asking if they want royalty's
Unit41: do you already have the patent?
Unit41: just because you made it first, doesn't mean someone else can't get the patent.
JymmmEMC, yes, it does
SWPadnos: It's who field the patent first
if I make a widget, but don't patent it, it shows that there's prior art
not in the US
everywehere else, it is first-to-file
SWPadnos: Ah, I can't keep up with them anymore
here it's "first-to-invent", so if you can show that you developed the thing (using dated lab notebooks or the like), you can litigate
SWPadnos: Does it still have to be worked, or can you just sit on it?
in the US, you can sit on it as long as you like, but you do weaken your case if you don't bother filing for years
but you can still prevent someone else from getting a patent on the same thing, since you can show prior art
it has to be "public" though - ie, if you're selling widgets or have published source code / algorithms or the like
SWPadnos: Gotcha. Dated/notorized notes should take care of that.
they should, but they won't give you patent protection, only patent prevention ;)
my pizza place is closed. :'(
anyway, it's now dinnertime for real. bbl
SWPadnos: I meant to proven WHEN. hasta!
Saw in article in the paper today about pizza flavor beer
that sounds gross.
pizza + beer, but not pizzabeer.
thats what I said... guess its made with tomatoes and a few other ingredents during the brew process
beer has been introduced to dough for centuries
supposedly its award winning
and tastes great with a pizza
frak. got noz fooodz
hate the franciased crap...
but craving pizza.
there is another guy other there making it that actually puts bits of pizza in the brew... thats really gross
[23:11:44] <Ziegler> http://cbs2chicago.com/local/local_story_134181507.html
pre vomt piza beer
if a cad system could describe a contour in a lot of tiny straight lines that could be read & executed in a time period equal to a typical servocycle, wouldnt that drastically reduce the workload of a motion controller? ( circumvent the tp, describe the exact blended path on the cad side, not the nc side )
yeh, its basicly run the code thru emc as it stands now, and record all the tc motions
then send those trajectory cycles (tc) to the motion controller, not gcode
tomp: if you think about the throughput you would need for that.. I think it's not really a great deal
i think thats what i see sent to mills tho, a zillion tiny line segments, like the users are avoiding the cnc in the machine by feeding it data that it cannot reduce
tiny = < .025mm ( never saw one that large ) recently on some tests between Selca and Fanuc
tomp: you end up with blending calculations for thoose segments.
0.025mm is pretty big :/
tomp: I think sending preprocessed segments were exhausted with the TP trouble we had about a year ago.. or two(cant remember)