The collected works of SAQ - Page 1

Never dealt with one that old, but the 7012-3xx and 7030 series use proprietary IBM 40-bit wide SIMMS. Got a bunch from a dead 3CT (16 MB beasties, no they don't work reliably in anything else.)

Regarding the console question- it seems RS6ks need more than the standard 3-wire cable, per Usenet:

"c32 is the last thing you should see before your graphics display comes
alive so I can only assume your adapter is somehow incompatible with your
monitor. FYI on your 530 without graphics the last thing is c33
I suggest you try to get a serial console working. The trick with these
RS232 cables is that the RS6000 expects a signal on pin 8 (DCD), usually
from pin 20 on the terminal, not sure what this is on a 9 pin, to work as
a console at boot time, it's a lot less fussy the rest of the time and
will work with a normal 2, 3 & 7 pin cable with 2 & 3 crossed over. Plug
into S1 most of the time.

Below is the pinouts for what IBM called a Printer/Terminal interposer.
This was a little silver D25 way Male - Female device, about 2" long that
plugged onto the end of a straight through cable that IBM used to supply
with every RS6000 plus the interposer. The straight though cable therefor
used pins 1, 2, 3, 4, 5, 6, 7, 8, 20 & 22.

The interposer then :

Female Signal Male
1 Shield Ground Shell
2 TxD 3
3 RxD 2
4 RTS 5
5 CTS 4
6, 8 DSR, CD 20
7 Signal Ground 7
20 DTR 6, 8

You need to work out how this is applied to 9 way, or buy a commercially
available 9 - 25 way adapter then build a cable based on the interposer
above.

Try booting in normal mode but if this fails turn the key to service.

You don't say if you have boot media, I hope you do. A CD, Aix install
tape, mksysb tape or if they're very old Aix 3.2.5 systems boot floppies.
I also hope you have the key. Turn the key to Service and reboot. All
attached terminals should get a message with a "Press key 1 to make this
terminal console"

Depending on the version of Aix loaded you should get to an install menu,
all slightly different, which also has a maintenance shell option. From
here you are effectively in single user mode and you can set the console.

Hope this helps, an RS232 interfaker is invaluable for these types of
problems.

Assuming you get past this stage the next issue I suspect will be your
root password, but that's another news message for another day.

Kindest Regards & Good Luck

John McQue "

Scott Quinn
CD 1 is the bootstrap one, remember for IBM RS/6k stuff you'll probably
need an IBM CD-ROM, the machines expect IBM firmware.

With a machine of that age, you will probably want AIX v4.3.3. It's from the "standardized era", so it's not too weird. 4.3.3 doesn't have the "L", but GNU stuff does build O.K., and I have a "AIX toolbox for Linux applications on POWER systems" CD, indicating that IBM did support it somewhat. AIX 5.1 is even slow on a 3CT with 512MB RAM, so I don't know what it would be like on a 7011.
two classics:

FSN ("I know this ... it's a UNIX system") is available from ftp.sgi.com.
Get the COFF/IRIX4 version

TQ can be found in a COFF version.

AFAIK there were no official "Freeware" releases from SGI, Gerhard Lenerz might know of some stuff- check out sgistuff.g-lenerz.de

Source from 1990-~1994 probably will build, but IRIX 4.0 was from before the "great standardization" from the mid '90s. I think it was SGIs first Vr4 IRIX, which makes it esaier to build on than 3.3, though.

You'll definitely want the IDO. What version is the one on your tape?
(4.1.1 was the best for IRIX 4)
Why do you compile Mips3 on Mips3? Aren't you supposed to be able to use cc -mips3 (or -mips1 or -mips2, or -mips4) on any architecture that supports a compiler that will compile it? (e.g. -mips3 on MIPSpro 7.3.1 on a R10k, or even -mips4 on an Onyx R4k (but you won't be able to try the binaries out there...)

Just seems that it would be a bit faster than dragging out the ol' Indigo R4k
Obelix said
Now, is there anyone knowing (a) how to repair it (this info is available for the PSU of the Indy...), or (b) where to get another?
The only supplier I know of asks 550 US$ from a new one - more than I paid for the whole thing.


Aahh- the SMPSU. Very implementation specific, and a bit of a black art.
Are you familiar at all with the theory of operations? The site at http://www.smps.us/ has much semi-technical and technical stuff, the sci.electronics.repair faq is probably the best start, http://www.repairfaq.org/sam/smpsfaq.htm

Basic thing to keep in mind: on most cheap SMPSUs (the Fuel's is probably in this category, all PC supplies are), the voltage regulation reference is taken from one line and the other supplies are expected to track along with it, which they do mostly. Are the other lines also off on voltage? (often the 5v line is the reference). N.B. - DO NOT LEAVE IT PLUGGED INTO THE MAINBOARD FOR THIS TEST! it's already too high, you don't want to blow something. Jimmie the PSU on, usually there's a line that you short to ground to turn it on. Who knows? It might be a common ATX PSU- you could get a good one (e.g. Antec) and put it in. Make sure though, they aren't always the same.
Not that big a problem if you are aware of it - Dell is also notorious for changing the PSU pinout (but they don't have reliably nice PSUs on all equipment). Sit down with a VOM and a dental pick, trace out the voltages on the SGI and then, using a ATX pinout (see http://www.hardwarebook.net ) pull apart the connector using the dental pick and reassemble. Done the rearrange a few times with various computer plugs.
It seems that there are two seperate issues here with Fuel - (1) the environment monitoring can go south and (2) the PSU can fail with overvoltage. In Obelix's case, it seems to be the PSU since he has tested it with a VOM. Perhaps this could go somewhere in a Fuel FAQ or a SGI PSU FAQ (since they seem to have problems...)
along with the enviromental monitoring outcome (definitely recommend the use of a good DMM to verify PSU accuracy before disregarding EnvMon warnings).

I don't have a Fuel available, but it seems like a good thing for hobbyists to have would be a pinout of the Fuel power connector so it can be replaced by generally-available sources. (provided the ratings are correct). Anyone with a Fuel, VOM, and bit of time want to oblige? :wink:
SGI now offers SiteManager and Cosmo for free download with free licenses (when applicable). Other than that, I don't know of any commercial creation or management systems (other than Netscape Composer). Other than that, I've seen Bluefish, which looks like it might be decent when version 1 comes out (the pre-1.0 I last tested was a bit unstable), and OpenOffice which saves into HTML but doesn't have a complete HTML editor set of features.
So is the general consensus that a user with .22 should go directly to .30 and not pass go provided their system will run it, or is there a better intermediate release?
Not at my Siggy now, but don't you get the MIPS backend with IDF/IDL et al and the frontend is the $$$$$$ licensed part??

Hmmm.... possibilities
8)
Don't know if he has the whole computer, but the gentleman who's helped me immensely with my Motorola IRIS has some boards for a Pixar Computer, they were out when I went over the last time.
Especially considering that with IP24 you're limited to IRIX 6.5.22 and down, and no new security patches are being built for those releases (patches are only verified on current release c, c-1, c-2, c-3).

Another good reason is that you are experienced building firewalls in xBSD. You could have some problems if you're learning on something that security-critical :shock: . Not will, just could, and there's no reason to risk it. Have some fun with the Indy instead.
I'd advise sticking with UPA framebuffers- better bandwidth. Creator3D isn't particularly great for complex 3D stuff, but you've got a SGI. Elite3D is O.K., not worth what Sun charged but you don't have to pay that :wink: .
UPA framebuffers keep all the graphics data off the PCI bus- bit like XIO. It's a Good Thing
SGI pulled the IP19,20,21,22,25,26,28 specific kernel files and the LG1, Newport, Express, etc. graphics files from IRIX starting at 6.5.23. This is not like some other systems where it is not "supported"- here it doesn't work unless you want to attempt major hacking (which would probably look something like: install on an Octane or other 64-bit architecture, move disk to IP27, move over at least /unix, /usr/sysgen and /usr/gfx, attempt boot whilst praying, and most likely move over some more files.
jan-jaap wrote:
SAQ wrote:
major hacking (which would probably look something like: install on an Octane or other 64-bit architecture, move disk to IP27, move over at least /unix, /usr/sysgen and /usr/gfx, attempt boot whilst praying, and most likely move over some more files.

An Octane (IP30) kernel wouldn't contain hardware support for IP28 hardware such as the GIO64 and EISA busses, SCSI, seeq ethernet etc. etc. so wouldn't boot.


Yep, That's why I talked about moving over the 6.5.22 /unix, /usr/sysgen and /usr/gfx parts. Considering that IRIX was being depreciated at the time of the 6.5.22 release, it's highly unlikely they did major changes. The one advantage I can see is that you'd pick up the security updates that don't officially exist on 6.5.22 (although it might be possible to force installation of these without any issues).
Supposedly, GCC 3.4 went a long way towards resolving the "can't link with MIPSpro objects" issue, haven't tried it yet, though. Nekoware is built with MIPSpro, so if the issue is incompletely resolved there can be problems. DSOs were never an issue in recent memory, though (GCC programs link just fine against the system C libraries et al). Can you try it with a hand-hacked Makefile/configure/whatever to use MIPSpro?
Yep- dialup+ 4GB is no fun here :(

Before you run the disks, could you check to make sure there aren't any incipient "great advances" that another week would add?
Another trip into the past:

I vaguely recall having come across the DWB packages at some point either on one of the IRIX or IDL/IDF/Compiler CDs (not a seperate DWB CD), or on the SGI site as a download. Is/was this true, and if so can anyone remember the CD or URL (since SGI has completely changed their Devel. pages)- or was I imagining this?
Groff it is, then. I just thought I might have seen it on that page in the devel. progam with all of the downloadable tools that's disappeared, but I guess not.

I'd heard that it was depreciated with the release of IRIX 6.5 anyway. What does DWB have other than nroff?
So, do you have the whole dev toolbox, or just some of it?
list init 128c000 12ff654 73654
mrboot
source : bootp()192.168.53.171:/30/sa(mr)
dest : dksc(0,1,1)


What's the problem here? Perhaps I'm reading it wrong, but it looks like it's copying the miniroot over to partition 1-which is what it should be doing.
Following this, if everything goes the way it should (which it isn't), your IRIS should then commence to boot from the installed miniroot and start inst to install IRIX to partition 0. The miniroot is a complete filesystem unto itself that just needs to be copied raw into what used to be the swap partition, so having a filesystem doesn't matter (it would just be clobbered by the miniroot).

Not skilled at IRIX netinstalls, but doesn't what happens next involve fetching the kernel from /dist/miniroot/unix.IPxx and integrating it in with the miniroot? Perhaps it's falling down here at the remote end. Of course, this would depend on whether the PROM has successfully copied the common miniroot yet.
porter wrote:
SAQ wrote: Perhaps I'm reading it wrong, but it looks like it's copying the miniroot over to partition 1-which is what it should be doing.


But if the disk did not have a label/volume-header it wouldn't know where the swap partition was supposed to be....


He said he labeled it (or at least booted fx.64, and why else use standalone fx?)

Perhaps I was confused by the use of the term "raw disk" - I interpreted raw disk to be the whole partition (10 in IRIX) rather than swap (1 in IRIX).
So- did Intel's practice of taking SGI trademarks and applying them to low-end technologies come out of the Microsoft "cross-licensing" fiasco? Just thinking- "Intel Extreme graphics" "Intel Vpro".
recondas wrote:
starfoxacefox wrote: which should have boot to a graphic prom


It shouldn't be an issue with an Octane2, but it is also possible you don't have a PROM revision new enough to support VPro graphics.

If you need to check you can connect via serial terminal and issue the command "version" at the PROM command line - while you're in there, you might consider executing a "resetenv".


Perhaps also a 'hinv' and 'printenv' just to see what comes out.

Checking the XBOW lights is also a good idea (which I think is what was meant by "do any lights come on") to make sure that the XIO connection is good and something didn't slip. When you shipped it, what protection was used for the compression connectors?
nekonoko wrote:
I installed a new captcha on the wiki last night since we've been getting more spam then legitimate edits over the past month or so. Let me know if you encounter any problems.


Not captcha related, but it's a bit odd that there isn't a link back to the nekochan home page on the wiki.
Two possibilities for fixing the bug so you can get around to downloading all of your nice open-source and public domain stuff:

look at ./configure --help, as there might be a variable that needs to be set (like CC, CXX or the like). As another possibility look at the makefile and see what is calling c99

(quick hack that might work) Alias or link gcc to c99.
deBug wrote: Tried it on my new fuel 600MHz with V12 and it was actually OK.
Around 15 fps according to the built in fps counter.


O.K- what's going on here? must be huge textures, since that is the only difference between V10 and V12.

Perhaps a "texture trimmed" edition would work better for SGIs?
Remember you can usually run ldd on the (ELF) binary and get an idea of what it needs.

Invaluable when you come across one of those sites who think that "Linux 2.6.14" is a sufficient descriptor of the system requirements :roll: .
kramlq wrote:
- Having root's home directory as "/". As you use editors, and in some cases install applications (e.g. adobe), you tend
to get silly .<something> file and log files left in "/". Why not use "/root"?
- Why are home directories in /usr/people as a default - it makes things like like a "find /usr -name <something>" search through even more directories and files than it should have to, compared with a "/home" layout (e.g. when searching to locate a binary or library that you have forgotten the path to).

Perhaps I got too used to Linux... but these are some places where I think Linux makes more sense than IRIX.
Will anything break by changing root's homedir to "/root", or users to use "/home"?


Root's home in "/" and user's home directories in /usr/nnn are (new-)old-UNIX-standard ['80s]. Pretty much every UNIX will use some variant of this until the mid '90s and the rise of GNU (with the exception of Sun, since they were so network/nfs based home directories are/were in /export/home. Root's home is still / in SunOS4 and Solaris, though).

Having a separate home for root does keep things looking nice, I'll admit, but you really shouldn't be doing that much as root anyway. That said, between basic administration, software install, and patches I build up a fair number of files owned by root.
There shouldn't be an issue with changing root's home, just make sure it's on the root filesystem just in case.

The bigger issue with IRIX is that in the mid '90s SGI broke with UNIX tradition and put required binaries in the /usr tree. Granted, disks are big enough to where it isn't that big of an issue, but it still was bad form.

_________________
Damn the torpedoes, full speed ahead!

There are those who say I'm a bit of a curmudgeon. To them I reply: "GET OFF MY LAWN!"

:Indigo: :Octane: :Indigo2: :Indigo2IMP: :Indy: :PI: :O3x0: :ChallengeL: :O2000R: (single-CM)
fvador59 wrote: yes and ?


I think Foetz' "yes, and" was an and of support- since you can't get support from $ORIG_COMPANY it's a good thing that support is available elsewhere in an informal manner. My PI/Indigo/IRIS 3k is unsupported by the vendor, so I'm very glad to have any support elsewhere, even if it is "unofficial" :mrgreen: Another point in it's favor- unofficial support is usually a little bit less expensive :lol: .
"Brakes??? What Brakes???"

:Indigo: :Octane: :Indigo2: :Indigo2IMP: :Indy: :PI: :O3x0: :ChallengeL: :O2000R: (single-CM)
No, what the thread was saying is that the PSU is a simple ATX+12v model, but the cable to the mainboard has a different pinout (and different color code, so it's a bit harder). What we need is someone to trace the connector and post it (along with the SGI color-code)- changing the connector is relatively simple, or you can get a box of crimp butt connectors and crimp on the old SGI cable (but you still need to know the pinout).

If someone will donate the Fuel, I'd be happy do do it :lol: , otherwise it's a 5-minute exercise with a voltmeter.
"Brakes??? What Brakes???"

:Indigo: :Octane: :Indigo2: :Indigo2IMP: :Indy: :PI: :O3x0: :ChallengeL: :O2000R: (single-CM)
regan_russell wrote:
Can I just stick in any crap PCI graphics card or does it have to be IBM special...?


It has to be RS/6000 supported, but would you expect different? You can't dump just any card in a SPARC, AXP, SGI, HP9k, etc.

http://pci.unsupported.info/ has some compatible cards.

Looks like http://www.flinkmann.de/linux-auf-ibm-6 ... .10.1.html has some more info:

The IBM S15 is a Viper Pro PCI/VP from Diamond Computer Systems Inc.
Its using a Weitek P9100 Videochipset and an IBM Ramdac (IBM RGB525L).

The IBM E15 is a S3 Vision864 Videographic Card with an S3 Ramdac.

The IBM GXT 120P is a Matrox Millenium I PCI Card.

The IBM GXT 130P is a Matrox G200 PCI Card.

The latter two may or may not have console/firmware display capability (but do have AIX capability)

_________________
Damn the torpedoes, full speed ahead!

There are those who say I'm a bit of a curmudgeon. To them I reply: "GET OFF MY LAWN!"

:Indigo: :Octane: :Indigo2: :Indigo2IMP: :Indy: :PI: :O3x0: :ChallengeL: :O2000R: (single-CM)
Been talked about- the 500MHz Indy is the result of a bug in the PROM, and isn't really running that fast.

The R5k/180 -> R5k/200 upgrade seems to work pretty well, though, based on reports I've heard, and you might be able to boost up a 175 or 200MHz R4k CPU module to something faster (the 100 and 150MHz R4000/4400 modules are 5v CPUs, the 175 and 200MHz modules are 3.3V CPUs. The 250MHz R4400 as used in the PM5 for Indigo2 is also a 3.3V part). The main question, apart from voltages, is how the CPU clock is made. On Indigo2, 100, 150, 200 and 250MHz modules use a straight .5x the internal frequency can oscillator. Not sure what the Indy R4k modules used. Also unsure about the PROM capabilities.
canavan wrote:
Does anyone have the eprom image or burned eproms that would be necessary to overclock a R5k 150 to 200MHz? What's the minimum cache speed required for this to work?


PROMS are the standard R5k PROMS, cache is (AFAIK) the same (the modification is changing the external oscillator from 45MHz to 50MHz, and the 150MHz processor module is also a 50MHz external clock. R5k doesn't have the backside cache bus, so cache access is at the system bus speed (45 or 50 MHz)- it's very likely that the 150SC and 180 modules have the same cache RAMs.
chatuser wrote: It's easier to buy an IBM Power System on eBay to enjoy AIX.


NB that AIX/6000 (v3 and above) are a bit different from AIX v1.x (and internally substantially different from AIX/RT v2).

If you really want to run AIX v1 get a PS/2- you're just asking for trouble to try to build a almost-exact-clone of a PS/2 from third-party parts. A little m80 or above is pretty cheap now.
"Brakes??? What Brakes???"

:Indigo: :Octane: :Indigo2: :Indigo2IMP: :Indy: :PI: :O3x0: :ChallengeL: :O2000R: (single-CM)
To be fair, compare with Linux support for MCA.


I'm not complaining- no one would expect to pick up a PlayStation or other MIPS-based system and run IRIX on it, now would you? Why should IBM go out of their way to support their competitor's machines?

PS/2s are pretty common and bulletproof (just make sure you get a SCSI model- a few of them were ESDI and that can be difficult unless you have a well-stocked garage).
"Brakes??? What Brakes???"

:Indigo: :Octane: :Indigo2: :Indigo2IMP: :Indy: :PI: :O3x0: :ChallengeL: :O2000R: (single-CM)
I wouldn't demand complete proficiency in assembler, but understanding assembler is understanding what the computer is doing at the instruction level, and that is valuable in teaching you to think like a computer.

At the same time, I don't think I would necessarily do assembler on an OS, especially not one like UNIX, but something like a Z80 or 6800-family system (or emulator) would give a good foundation in how computers work on the software level.

My uni started you off on Ada (back when they had a Sun, don't know what they use now), and I didn't quite understand that.
Damn the torpedoes, full speed ahead!

Living proof that you can't keep a blithering idiot down.

:Indigo: :Octane: :Indigo2: :Indigo2IMP: :Indy: :PI: :O3x0: :ChallengeL: :O2000R: (single-CM)
Probably not for a while, if ever. The lack of Classic is a bit of a dealbreaker.
"Brakes??? What Brakes???"

:Indigo: :Octane: :Indigo2: :Indigo2IMP: :Indy: :PI: :O3x0: :ChallengeL: :O2000R: (single-CM)
So what's hinv saying? Fx when you view the partitions?
Damn the torpedoes, full speed ahead!

Living proof that you can't keep a blithering idiot down.

:Indigo: :Octane: :Indigo2: :Indigo2IMP: :Indy: :PI: :O3x0: :ChallengeL: :O2000R: (single-CM)

Code: Select all

# hinv
1 20 MHZ IP6 Processor
FPU: MIPS R2010A/R3010 VLSI Floating Point Chip Revision: 4.0
CPU: MIPS R2000A/R3000 Processor Chip Revision: 2.0
Data cache size: 32Kbytes
Instruction cache size: 64 Kbytes
Main memory size: 64Mbytes
Integral Ethernet controller: Version 0
Graphics board: GR1.2 Bit-plane, Z-buffer options installed
Integral SCSI controller 0: Version WD33C93A
Tape drive: unit 4 on SCSI controller 0: QIC 150
Disk drive: unit 1 on SCSI controller 0
#


It came with the Turbo option, but I was having problems with graphics lockups, so that's sitting in an antistat bag :( .

Code: Select all

#uname -a
DouglasI DouglasI 3.3.2 12031609 IP6


Not sure if IRIX 4D1-3.3.2 has a bug or something, but the machine slowed down noticeably after going from 32MB to 64MB. Guess I'll back down to 48 and see what happens.
"Brakes??? What Brakes???"

:Indigo: :Octane: :Indigo2: :Indigo2IMP: :Indy: :PI: :O3x0: :ChallengeL: :O2000R: (single-CM)