The collected works of mapesdhs - Page 9

Yes, that's most likely bad TRAM(s). Swapping boards is a pain - with 2 TRAMs per board, it can be hard to isolate the fault.

I'm not sure if I have any spare, had a request for some from the OZ airforce. I'll know next week.

Ian.

Good point, try a TRAM reseat, ya never know.

Ian.

Yes, it's on the middle PCB.

Careful how you extract the 2nd TRAM. Remove the end plate first (the one that goes over the monitor
and stereo socket), then remove the top 4 locking screws. When you bend the board upwads, don't
force it all the way to 90 degrees. Just keep it at around a 50 to 60 degree angle. Use a pair of tweezers
or needle-nose pliers to extract the plastic locking pins from each end of the TRAM. If you use anything
like a flat-bladed screwdriver to lift up the TRAM away from the PCB, be careful to do so at points on
the PCB where there are no surface mount components that could be harmed.

When you're testing a Max set, do so with either no TRAMs installed, or both. Don't try it with just one
TRAM. This is why I tried very hard to find a HighIMPACT set so that I could test TRAMs one at a time,
then move pairs onto a Max set once I had 2 TRAMs that are ok.

Also, do not test IMPACT boards with TRAM installed when the top-lid for the Indigo2 is not fitted.
Alternatively, use a separate fan to help cool the TRAM on the top board if the top-lid is not in place.
I use a fan connected to the molex cable of a disk sled in the upper disk bay.

In all the above, observe antistatic precautions and make sure the bottom PCB does not scrape
against any surface or object which might result in other surface mounted components failling off.


The most common fault with TRAMs is the Alpha channel. Use the tex_cube demo to check initial
Alpha functionality. The image should appear consistent whether the window is small or large. If it
works ok, check more complex Alpha features with the Macau demo - watch how trees and lamp
posts appear, look for artefacts around the billboard edges (press T to show the billboard shapes),
and also watch how building textures in the distance appear - they should not show any dithering
or 'rolling' pixel effects if the TRAM is working ok.

If the Alpha is working ok, I then test more basic texturing using the Mirrors demo, Matterhorn, Huge
Engine Model (including transparency mode, and test shading/wireframe with textures off), A10, F18,
and of course try an IrisGL task aswell (I use Beethoven model from buttonfly).

Other symptoms:

- If running a 3D app causes the display to lock up and then resets the X server back to the login
screen, this is normally a fault with the GE board. Often this is the case when the display looks ok
after initial power on and after the system has fully booted and logged in, ie. the DG and raster
elements are ok, but a bad GE isn't exercised until a 3D app is executed. This fault results in gfx
timeout errors in SYSLOG.

- If the display is already bad after initial power-on, ie. when the Press ESC for Maintenance message
appears (typically visible as lots of jagged horiztonal line effects), then this is normally a bad DG board.

- If the system makes a normal power on tune and sounds like it's booting ok, but the screen is black,
this can be the result of one of the TRAMs not being seated properly. Check by removing both TRAMs,
ensure the base Max set is ok, then refit the TRAMs, ensuring they are correctly seated.

If you decide to swap boards between Max sets, be very careful. Don't use the tiny levers to disconnect
the ribbon cables, they are too fragile; I use a small flat-bladed screwdriver to just lift up the connector
either side of the socket. Take the opportunity to remove any dust with a soft brush and/or inert air duster.

Don't bother using the PROM ide for testing, it's not accurate, often reporting false positives.

If you can, use the lower three slots to test Max sets. This gives better cooling for the TRAM at the top.

Ian.

Thanks!

I've added some extra info. Thought I may as well since it's fresh in my mind atm having just
sorted out two Max sets and four extra TRAMs for a simulation company.

Ian.

mia wrote: I would really, really like the following available as a nekoware tardist.

c-ray: http://www.futuretech.blinkenlights.nl/c-ray.html

I'd rather c-ray not be a tardist, because it's best used after being compiled natively
on each test machine, rather than using a predefined binary for all compatible systems.

Btw, my main site is www.sgidepot.co.uk/sgi.html (Blinkenlights is just a mirror).

Ian.

Typing this with FF3. It has slightly improved the visual layout of Yahoo Mail (date column is now correctly
in view), but alas the buttons are still missing (I can see where they're supposed to be, but there's no image/text).

However, this FF3 vastly improves all sorts of other things, eg. sites like tomshardware, eBay, overclock.net, etc., all
work much better and faster, so I'll stick with it for sure. Mega congrats to all those who've worked on this build!


hamei writes:
> Ian, the final solution is to set up a real mail client then get the stinking yahoo mail via pop. ...

Hmm, I suppose. It's just that the webmail interface has always been fine for me, and it's handy when I'm away, etc.


> Webmail is useful when you go on a trip : otherwise it's awful.

I never found it awful at all. Been fine for ages, and my account goes back many years (long before Yahoo UK was
set up). Thing is, I rarely use the webmail interface to type replies (except for very short replies of which I don't feel
a need to keep a record). Normally I use the mouse to copy a mail into jot; I have (had) some nice little jot macros
which process the headers to tidy them up (one press of F9). Then I CTRL+A and middle-button the text into a
different jot reply window, press F4 to process the headers again to leave just the source email address and subject
line. It's very fast & it means I have a normal text copy of just about every email I've ever received or sent. I don't
use webmail as a place to store emails or other info long term.

But with the new setup and the way an email is displayed, I can't use the mouse anymore to highlight the whole
text of an email (including headers) so my macro setup is void (might add I've been using Yahoo Mail this way for
fifteen years ; infact it'll be exactly 15 years tomorrow, my account was opened 15/Jun/1998).


> The other solution of course would be to tap directly into the NSA's feed .....

You do realise of course that having said that you'll have concrete shoes by morning.

Ian.

Does anyone know where on a single-CPU module for Octane resides the NIC chip?
It's this which reports back the info given by hinv -vm for serial no. and part no.

I have a CPU which works 100% ok (ide diags, every benchmark, etc.), but it halts
on power-up because the serial number NIC on the module is bad. If I can locate
the NIC, perhaps I can swap it for a NIC from a lesser CPU. It'll report a different
SN/PN of course, but at least it will then boot and function ok. Anyone know? Is it
perhaps the DS2505 IC?

Ian.

Solved!!

There is a surface-mount 6-pin IC inside the module on the PCB, marked DS2505. This is the NIC. I swapped
it for the same IC found in an old R10K/175; bingo! No more halt at power-on, PROM hinv is fine, CPU working
as normal (performance checks with c-ray, etc.), IRIX hinv is fine. Only the PROM command 'system' and IRIX
'hinv -vm' show a different PN, and a serial that doesn't match the sticker on the CPU.

Makes a chance... fixing a PITA problem in just a couple of hours.

Ian.

ajw99uk writes:
> So here I am back with FF3 - but still asking Yahoo to "remind me later" about the new webmail UI!

You'll be forced to switch over about 1 month after you received the original notification email. You do
have the option of using a simplified version of their new setup, but it's still a mess compared to the
old system. Only saving grace is it's fast enough to use with FF3 with javascript left on, and attaching
files now works correctly, if kinda slow.

Ian.

I access my mail from a lot of different systems as I'm doing my stuff. I'd rather not have to farf
around with such things on so many different systems.

Anyway, doesn't matter now, I've put a post-it on my monitor border showing the button map
of Yahoo's mail layout, so I've gotten used to clicking on what/where already even though I
can't see the buttons. :}

Btw, does anyone else see this FF3 build displaying certain characters rather strangely,
especially quote marks?

Ian.

Indeed, good point! And yes it's an EPROM.

For anyone reading this in the future, note that the part used in O2 looks completely different; located on
the PCI tray PCB, it looks just like an ordinary transistor, has only 3 pins and is not surface-mount.
Thus, if you want to change an R5K O2 to R10K (or vice versa), then swap over this component
between the two PCI trays and this will retain the serial no. to match the sticker on the back of the O2.

Ian.

recondas wrote: That's logic at a level beyond my comprehension

Here's a rather dark photo overlaid with a partial screenshot...




Ah yes, it's all super advanced technology around my office, la la la... :}

I'd be doomed without postit notes...

Ian.

Notice how the 'Spam' button is shown correctly, though it's kinda blended into the 'Search Mail' button somewhat.


Never mind though, I'm getting used to it, and the extra speed of this build really does make a difference.

I'd say the only thing that bugs me atm is the browser not displaying certain characters correctly,
especially the various kinds of quote marks.

Ian.

I've edited my post so it makes more sense.

Ian.

Nah, I'd rather it left as it is for the moment.

Ian.

Minor point I'd like to mention - the 36GB 15K drives I fit in Indigo2 are one of the quietest
drives I've come across (particular model of HP-branded Seagate, BF01885A34, Seagate
PN 9W7006-038), far better than the vast majority of 10Ks and way better than any original
disk for Indigo2 (it's almost as good as the Fujitsu MAS series). Only thing better in an Indigo2
is an SSD. If he can hear the disk now then indeed the new fans really must be quiet.

And yes, I totally clean every system before shipping.

Ian.

PS. Keep a careful watch on your R10K for a while. When I tested new fans on O2 CPU
modules, using ones with slighly lower air flow didn't work that well. In the end I found
a thicker fan with higher flow and less noise, whcih I fit to custom R12K/300 modules.

PPS. foetz, on Thursday I upgraded an Indy at a textile company with two OCZ Vertex4
64GB SSDs, R4K/200, max RAM. Since it has a Sony PSU, it now makes no noise at all .
Beats the Indigo2, hehe.

I also replaced their dying external floppy and DDS2 with another 64GB Vertex4,
stored in the case which used to hold the DAT. It's cut the time for their full
backup from 30 mins to 5 mins, and the guy loves the random access nature &
flexibility of the SSD vs. how one must deal with DAT. Textile patterns are a
perfect fit for SSDs - lots of very small files. Their whole archive is currently only
75MB, but it has more than 22000 files.

Sorry, I digress... Indigo2 roolz!

Ian.

vishnu wrote: This is fascinating to me, do they have a website? Are they just doing this just for file storage or is there some industrial control that they're also running on the Indy?...

No site that refers to the Indy. The Indy is called a Stoll/SIRIX system; they were sold without any SGI markings. It's used to
control more than a dozen large knitting machines, each between 2m and 4m long, big warehouse style factory. Scarves, gloves,
jumpers, hats, etc. The Indy connects to each knitting machine via a Selan, a sort of network box connected to the serial port.

Ian.

foetz wrote: erm, how do you get an ssd in an indy?
converter or something?

Acard ARS-2160 SCSI/SATA bridge box (SCA version) + direct low-profile 50/80pin adapter.
The bridge box has the same 3.5" 1/3rd height form factor as a normal HDD. Here's a pic of the
system with the two internal units fitted:



So far I've tested the folllowing SSDs & confirmed they will work ok in Indy/Indigo2/Fuel/Tezro, so
they should be ok with all other SGIs: OCZ Vertex2E/3/4, Vector, Samsung 830 & Samsung 840 Pro.
Hence, one could have an Indigo2 with a 512GB 840 Pro.

For reference, the Acard bridge boxes cost about 155 to 195 UKP each depending on the source.

I had recommended the company use 128GB 840 Pro SSDs, but ended up going with 64GB Vertex4s
to reduce the cost a bit.

Ian.

I have plenty; just bought 10.

Indyman007, are you looking for a unit with skins that have very few scratches? That could
be more difficult, though I've been able to work wonders with T-cut.

Ian.

Geoman wrote: my guess is: fastest Fuel: 0.5 per CPU

Sorry to disappoint you Geoman, a Fuel/900 gives about 0.4, though as I type this it's wavering
between 0.33 and 0.36 because of Firefox 3.0.19 and other stuff running at the same time.

Hmm, I might shut FF down, see what it does...

Ian.

---

EDIT: without FF running, it gives 0.41.

Bitcoin mining is a complete and utter waste of time for a home user. It ceased being financially viable many months ago,
as the power costs now far exceed the mining returns, all as a result of ASIC/FPGA systems, etc. Rising energy prices
have made it worse, as have rising component prices for PC builds.

Litecoin uses a different algorithm for which there aren't any ASIC solutions as yet, so it's still viable atm, which has
resulted in a sharp increase in prices for AMD GPUs such as the 280X/290X/etc. What launched as good price/performance
alternatives to the 780/780Ti, especially with aftermarket coolers, are now far too expensive, and also top-end PSUs have
either gone way up in price, or they're impossible to find.

As for running minerd on a Fuel, I have my Fuel turned on a lot of the time, so it may as well be doing something useful,
good for a laugh.

Ian.

PS. Bitcoin background and return analysis:

http://www.tomshardware.com/reviews/bit ... ,3514.html

Curiosity was burning a mental hole, so I tried out my 5GHz 2700K, which gives an average of 67.28 khash/s for CPU-mining.

Ian.

*sigh* Seems like SGI really didn't think ahead with respect to having a cleaner way of fixing these issues.
I've been supplying parts to companies globally who are still using SGIs after more than 20 years, back to
Personal IRIS. There are still loads in use out there (mainly medical, industrial control, defense and infrastructure
setups, though a few for performance-related tasks where sw license issues preclude a move to newer kit); Fuel
is used for medical scanners. Atm it's just power supplies and the odd RAM module problem cropping up (I sent
a PSU to a hospital in India last month), but eventually such institutions are going to be in a right old bind trying
to sort these things out when they the kind of problems jan-jaap is going through.

IMO SGI should back-release some kind of fix for this, eg. a program to sw-override the lic code setting, etc.,
and then declare the old IRIX systems as defunct and 'open' or somesuch. I'm sure it's possible.

In some cases companies can eventually move to newer PC-based systems, despite the signicant
expense, but others just can't for other reasons, eg. original sw vendor is gone, or support hw isn't
compatible, etc.

Ian.

FREE (Edinburgh, UK, personal collection only):

Sagitta Dual-Channel U160 SCSI 12-bay JBOD storage units, more than a dozen available. They look like this:

http://www.sgidepot.co.uk/sgidepot/pics/raidtest2.jpg

More pics:

http://www.sgidepot.co.uk/misc/jbodunits1.jpg
http://www.sgidepot.co.uk/misc/jbodunits2.jpg
http://www.sgidepot.co.uk/misc/jbodunits3.jpg

Each holds 12x 3.5" SCA SCSI disks (max height 1.6") in removable cannisters. Host connections are via dual U160
VHDCI ports, with two more ports for daisy-chaining. The units are key-switch controlled and redundant power can
be fitted (each unit has 2 PSU modules, with space for a 3rd). An instruction sheet, long power cable, Y-splitter
cable, keys set, etc. all included with every unit.

I've been told the PSUs are compatible with other SCSI/FC units, eg. SGI TP9100, etc.

Ian.

PS. I have disks available cheap if required, eg. Seagate 73GB 10K 1.6" for 5 UKP each, or Maxtor 147GB 15K 1"
for 40 UKP each.

Bump! No takers? 16 units to give away! Otherwise they're off to the scrap heap.

Ian.

Grey top cover... is it actually an O2+ unit you have? Normal O2s have a black base.
I have a number of skins in good condition, both orig and later logo, but nothing for
O2+ except a PSU cover.

Ian.

Hmm, I could ship by pallet. Not until September though.

Ian.

With no disks installed (ie. just the sled cannisters fitted), each unit is 27.5Kg, ie. about the
same as a typical good-spec Octane.

Ian.

Could well be! The label on one of the PSUs says:



and of course a SN.

Btw, I used these units to get almost 600MB/sec with an Octane via three XIO channels. Not sure if
anyone's beaten that yet. Wanted to try with four units but never got round to it.

Ian.

TeamBlackFox writes:
> Damn if only I was rich enough to have one shipped...

Indeed, it would cost about 110 UKP + packaging costs to send one unit to the US,
but alas it's not worth time/effort. I couldn't even sell them way back for 25 each.


> ... Good luck with the sale.

Not a sale; free!

Ian.

miod in another thread pointed to this one as a place to post a pic of the horror that is moi.

Here's me visiting the Bovington tank museum while on holiday last month,
and the coolest road-sign I saw while in the area.

Ian.