The collected works of squeen - Page 4

vegac! You sly son-of-a-gun! You got the 24bit visual load working. I rebooted my machine in 8bit mode and iconbar starts up no prob (boy do I hate color map swapping :) )!
I'll have to see how you did it over the weekend.

Thanks and nice job!
foetz wrote:
As for glib2/gtk2 as far as I know they are 100% MIPSpro.


sure if you say it i believe it. it just shows some well known issues happening with gcc builds so
this was the first that came to my mind.
i guess you know that i know very well how much work it is to provide a complete suite like nekoware.
be sure i esteem this!


foetz -- you willey old hacker -- I've been hoping to get some feedback from you regarding the new nekoware. It is my belief, unless configure out-foxed me, that the gtk 2.4 on nekoware/beta is 100% MIPSpro. If you have run into some errors please tell me what they are to I can go back and try and fix the port. Also, if your builds don't exhibit the same errors I'd like to know that as well. You know that I'm just stumbling my way through this and its a big help if the elder statesmen like yourself help keep things on course. The goal is to have a stable foundation that most folks feels comfortable building against and using.
Where does one buy/download additional games for scummvm? I've play Beneath a Steel Sky.
Cool, I'll have to poke around ebay! Thanks.
lewis wrote: ...get a reliable 20 FPS playing a DVD ...


If you feel this is an improvement over the version on which you started, could you make a patch against the source and post it under the contrib section of the Nekochan downloads?
If you have freeware installed you might start with the freeware version of the gimp 1.2 .
There is a mac font conversion utility -- I'd have to dig a bit for that one.
Sodipodi is a vector illustration progam, also on freeware.
Xfig is another (older) vector drawing program.
running fixpath from the freeware site should set your paths for you. In general, the freeware executabales are in /usr/freeware/bin I think foetz puts his stuff in /usr/local/bin
If you are intent on gimp 2, you might try the Nekoware/beta version I just built. It installs with SoftwareManager (tardist) and will prompt you for the prereqs (i.e. gtk+ and libart). You can get both of those tardists from the same location. Each one will then prompt you for more prerequists just like SGI freeware. Eventually it should run. (You'll want /usr/nekoware/bin in your path, or just type /usr/nekoware/bin/gimp).
Sheer genius! -- I'm envious of your talent and time. :)

Thanks for sharing, there's more than a few good optimization tips there!

Also, don't let this good programming get lost! Build a super nekoware mplayer when you are through (patches and all)!
An alternative to Xinerama is Opengl Multipipe (which has the older sgi-Xinerama bundle in it as well). Take a look at the documentation regarding the differences.
I vote this the most exciting thread!

Maybe I can write a simple Motif front-end for the final product....hmmm
Not all of the Xscreensaver hacks are operational for me -- I'd be interested on what excatly you are doing that generates the Bad Match error. Maybe I can peek at the source code and come up with a work-around.

BTW I get following results when I run webcollage directly

Code:
/usr/nekoware/lib/xscreensaver/webcollage
Can't locate bytes.pm in @INC (@INC contains: /usr/share/lib/perl5/irix-n32/5.00405 /usr/share/lib/perl5 /usr/share/lib/perl5/site_perl/irix-n32 /usr/share/lib/perl5/site_perl /usr/share/lib/perl5/sgi_perl .) at /usr/nekoware/lib/xscreensaver/webcollage line 57.
BEGIN failed--compilation aborted at /usr/nekoware/lib/xscreensaver/webcollage line 57.



I assume it's because I haven't set it up correctly.
Hmm. I'll have to go back and make sure xscreensaver was finding the nekoware Perl during configure. Also, I tried running the xscreensaver-getimage command directly, giving it the root window id as an argument. Although I didn't know what to expect for results, it seemed to work, but if you could describe the steps for your trial image grab, I'd be happy to reproduce them to see if I get the BadMatch X error.
I have got to get one of those working on my Octane!
If you have the DM10 software installed, I suggest looking at the man page for libfw(3dm). There is also an example program for querying a firewire device. I'd guess some programming will be required.
Too cool! CVS away with one caution -- my parttially implimented of remote terminal tinting code is already in there and it needs a bit more work, Hopefully it won't interfer with what you have done.

Beautiful! I can't wait to try it.

BTW: I have also modified the tryRGBfile() routine not to discard alpha channel data. I've practiced using GIMP to convert png w/ alpha directly to RGBA (SGI format) and it went well. The next step is to alpha-blend + build an icon bitmask on the fly. This will allow users a method to construct nice (shaped) custom icons effortlessly from all the png images floating around on the web.
hamei wrote:
foetz wrote: hello,

$500 for a dm10?? by far cheaper as i thought!!
where do you get this price?


Isn't that the official SGI price ? I thought it was ....

Quite a bit more than SIIG charges, at $50. Or unixmuseum paid, $5 :-)

Hey, does anyone remember where the url is for the SGI-recommended O2-cam replacement ? The one on the funky-looking little tripod ?I could swear i saw that exact same camera over at the electricity market for $15 today. If so, definitely should purchase one or two.


We got the DM10 (discounted) for $365 and the DM8 for $146.

The recommended camera for the DM10 is the GM051394-KD the 1394 - FireWire camera with legs .
Please let me know hwo this performs. I have the SCSI DVD-RAM drive from SGI and have had problems with the data integrity on some writes (i.e. I get a file of <nulls>). Other times, all is OK. I wonder if I got a bad drive or cable.
Inkscape looks like an offshoot of sodipodi. Of which I've put a beta version (no printing until the GNOME libs get built) in the nekoware beta directory. I'll try to build this when I have some time.
vegac : I just got your new border code from the CVS server and it works beautifully. I love the new look and am also very happy to see that you fixed the multi-row wrapping bugs. Great job ! The project would be "twisting in the wind" without you.

Also, for those interested : I added an optional feature for iconbar to manage the IRIX toolchest window. If you enable this feature via the Xresouce

Code: Select all

Iconbar*toolchestManage: true

iconbar will auto-raise the toolchest when it gets obscured by another window and also "snap" it to the root window coordinates specified by the resources *toolchestX and *toolchestY, which default to 1,1 (since 0,0 was causing the VisibilityNotify events to get lost). I know alot of you would prefer that iconbar leaves the toolchest alone, so I have this feature disabled by default . This is also covered in the iconbar man page update.
Oh heck, I got so excited I packaged it up as iconbar-pre0.2.tardist . I am a bit rushed, so hopefully I got everything correct. :D
edefault wrote: Squeen,

great work!

Now I have some suggestion:
What about the printer status panel icons... could these be integrated eventually?
It is very helpful to know about printer status by watching the icon color change.

And if I were able to choose to keep a liitle bit more of iconbar visible, i. e. not to
hide it completely...

Regards

I wouldn't might a little printer AND email notififaction icon tucked in somewhere, I'll put it on the TODO list. Although I have no idea how to check the printer status. Something I'll have to dig around for...if you have a clue---lend it to me. :)

How much more would you like to keep visible? The top border, the icons without names? Or just a number of pixels you select with a resource?If the latter, then there is already the *hideheight Xresource (see man page).

To all brave users : The iconbar-0.2pre.tardist I put together lacks two "fixes" I've put in CVS since the hurried release. One is an autodetection of Onyx2/UltimateVision graphics which has reversed RGB bitmasks from all other SGI visuals (which are BGR?).
Also, the new toolchest management auto-raise feature was not playing nice with fullscreen apps (again!) so I fixed that as well. I'll put together a pre0.2a sometime next week.
Mathworks definately have the "Microsoft" syndrome -- a near monopoly in their sector.
I am 100% certain that r14 and up will not be available under IRIX.
Hold on to to your r13!!

If you remember, Mathworks dropped (pre OS X) mac support at one time also.

Since December, we have been working on an in-house, native IRIX MATLAB replacement written from scratch in C (of course) as the front-end to our simulator. So far it's only has a fraction of the functionality, but a core lex/yacc based command parser is there. It reads M-scripts and writes MAT files. The plotting tool kicks-ass. I don't know if a public release is ever feasable, but maybe someday...

Meanwhile, there is Octave for which nekonoko built an IRIX binary back in the days of old.

http://www.nekochan.net/wiki/downloads/index ... Fnekonoko/


Meanwhile I plan to switch to Mathematica where possible.
Mathworks can go suck it for all I care - they drop IRIX, I drop them.
(They've gone all java/C++==Microsoft anyways!)
Arguing that Java is better than C++ is like arguing that grasshoppers taste better than tree bark.
- Thant Tessman

I invented the term Object-Oriented, and I can tell you I did not have C++ in mind. - Alan Kay
Here's a screenshot of Mathematica 5.0
[url=http://www.nekochan.net/wiki/albums/album36/mathematica.jpg]
Image
[/url]
click to enlarge
The IRIX OopenOffice 1.0.3 gets part of the word/power point documents sent to me in the office screwy (fonts & figures usually) about 50%.
Appropriately, I can not see the link (flash?) from an IRIX based browser. :razz:
I once attended an SGI/Intel presentation that talked about the Intel compilers on IA64, and how the GCC stuff would (because of some such optimization) run at best 75% the speed and how the AMD folks would not longer be able to cross compile. Sounded like Microsoft tactics to me at the time....
skywriter wrote:
i like the cabinet style...

:D :lol:

me too (damn!)
SGI still talks about a five year IRIX commitment.

Also the 1GHZ R16K is due out "soon".

After that.....?
I agree with Neko that Softwindows 95 was nicest, but I couldn't seem to update IE past version 4 and it wouldn't run AutoCAD correctly (But you can load the latest Office).

I now use rdesktop regularly when I need a windows interface (~ 1/month). The SGI freeware sharity package is what I use to mount the windows drives on the IRIX box. You can do it the other way too (IRIX drives on Windoze) using samba -- but I don't know how. (I'd love to see a nekochaner write a HOW-TO-SAMBA on IRIX article for the news site).

Lastly, there is VNC or TightVNC (freeware) client/servers which sends a rasterized desktop across the network. Not as fast as rdesktop, but it works with more OS's. I use the x2vnc component to control the Wintel PC using my IRIX box's keyboard and mouse as an alternate for rdesktop.
Thanks Diego -- I must have missed this one!
DVDauthor or equivalent/better.
Nice. Thanks!
Thanks bcasavan for the clarifications. I'm curious as to you background on IRIX pthreads (current/former SGI programmer?).

It's only your last point I would disagree with: I've never seen PROCESS scope threads appear across more than one CPU. I understand that the function call

Code: Select all

C SYNOPSIS
#include <pthread.h>

int pthread_setconcurrency(int level);

int pthread_getconcurrency(void);

DESCRIPTION
Threads which are created with the PTHREAD_SCOPE_PROCESS attribute (which
is the default) [see pthread_attr_setscope()], are scheduled on a number
of kernel execution vehicles.  By default the number of execution
vehicles used is adjusted by the library as the application runs and is
called the concurrency level.  This is different from the traditional
notion of concurrency because it includes any threads blocked by the
application in the kernel (for example to do IO).  The library raises or
lowers the level to maintain a balance between user context switches and
CPU bandwidth.

is a hint for doing this, and supports your point. However, in practical application I always end up with just one pegged CPU.

Also, the default priority of SYSTEM pthreads seems to be at the bottom of the real-time priority band, which is above all normal user processes.

Anyway, practically speaking, it would be great to realize some sort of dual processor speed-up of mplayer. Any suggestions? :)

EDIT : one last question : When did pthread_barriers appear in IRIX? I just stumbled across them in the past month. I'm pretty sure they weren't there pre-6.5.21.
bcasavan wrote: For almost four years I've been the primary Pthreads engineer at SGI.

It's an honor! Welcome to Nekochan, were old IRIXer go to fade away :) :( .
I'm pthreading right now (with an without REACT) and loving it. Thanks! All has been well since 6.5.21.

I'll try out your sample program shortly, but I have one pressing question -- would you expect the pthread_barrier call to have microsecond latency or ten's of microsecond latency on an Onyx350, where each thread is SYSTEM scope and running on an isolated processor, non-premptively? I am looking to speed up non-real-time performance, and when I used semaphores I was disappointed by a speed decrease when I doubled the threads (BOUND_NP, preemptive). The threads take about 500 usec each (2x= 1 millisec sequentially)), but need to sync a couple of times during the execution (say every 4x125 usec). What I'd like to see is a 2x speed up running them on 2 CPUs, versus the sequentially result.

Lastly, anything new on the REACT horizon? (IRIX or Linux) :)
I've never claimed to have a clue.
My SGI is Holmes, Mycroft---not I!

Thanks for the info Brent. Let me know what the real-time fella says.
We just got the BG Electronics flybox up and running for our sim. I'll have to see if I can get it to work with flightgear on the Onyx350!
Thanks skybound. I figured I would have to mod the code some -- the beauty of open-source!
We are preapring to buy 5 new Tezros as hardware-in-the-loop because of (in order):

a) non-preemptive multi-processing (i.e. REACT real-time)
b) scabable symmetric multiprocessing
c) stability
d) 64-bit extended precision
I'm in a position to buy a DVD-R for an IRIX box at work. Can someone suggest a manufacturer + model number?

If I can get that working I promise to dilegently port DVD authoring tools. :)
Wow. SGI aught to impliment it! Have you contacted them?