The collected works of SAQ - Page 10

Find a friend at SGI and ask - emphasize that you aren't going to be making any money.

If you don't have a connection they'll probably ignore you, though .

ianj wrote: ... Solaris to my knowledge is the only actively-developed "real UNIX" that offers modern desktop amenities.

Technically Apple Mac OS X is a "UNIX" according to the UNIX people, and in reality it has been moving closer and closer to a UNIX in feel (getting rid of NetInfo, X included,...)

The only downsides is that your choice of platforms is "officially limited" to overpriced unexpandable boxes based on last year's technology and a fancy case, unexpandable all-in-ones, and very expensive high-end machines. That and the capriciousness of Jobs.

I've heard Color Classic II was much improved. My favorite compact Mac case is the SE. Crisp and clean design, and of the innards for the compact Macs sold in the US it would be either the SE/30 or SE (for a 68000 Mac). Pity they didn't do a "Color SE/40" - that would have been a killer.

Do you suspect that Ian would have a race in "flight" or would he be trapped in "Battalion"?


Back on topic, the best way to troubleshoot is to check the link lights, then set up a BOOTP server for the SGI to try for and look at the logs. If you didn't want to do that you could sniff for the BOOTP packets somewhere on the network as well.

The PROM doesn't have any way of knowing about the presence or non-presence of the monitor. If you're getting a red bar with a keyboard plugged in to the keyboard port then there's a problem. Start out by cleaning and re-seating all of the boards (memory, processor, IP30, graphics) DON'T touch the flat connectors on the "front" of the boards because these can be easily damaged.

SGIs will go red-light and halt at the PROM screen if they do not sense a keyboard and they are set to graphical console. In this case the console is redirected to the first serial port.

If you didn't have a keyboard plugged in try starting it up with one. If you still see the red light you've got a problem.

VPro for the PC was nothing like VPro/ODYSSEY. Odyssey did connect via XIO rather than AGP, but the differences didn't end there.

ODYSSEY is a completely SGI design targeted at the high-end desktop market using two custom SGI ASICS (PC VPro used a single nVidia ASIC that had everything in it). The main chip was BUZZ and included the majority graphics pipeline, the second was called PB&J and while I haven't found information on it specifically I think it was primarily the DG portion (Display Generator), whereas BUZZ was geometry, raster and texture.

Odyssey blew the contemporary nVidia offerings away.

Dennis Nedry wrote: If that's one of those round keys with the (parallel) cuts down the outer side of the barrel...

I'm pretty sure you can defeat those by using a BIC pen with the cap removed from the end (or anything that fits snugly into the lock- so there's a fair amount of friction), by just jamming it in there and turning it in the direction the lock opens, then slowly pulling it out (while continuing to turn it).

I've defeated a few locks like that on other various keyless equipment.

If it's a bog standard pin/tumbler lock... You can defeat those easily enough with a bent jumbo safety pin (or anything else that retains it's shape under pressure) and a screwdriver... Google lock picking, it's easier then you might think. Paperclips might work, but they tend to bend before you can apply enough leverage to pop up the rear pins.

-DN

Spring steel is the way to go.

Cisco cables are probably 3-wire and there are a couple of machines that do not like three wire serial consoles.

fieldframe wrote: I've run a rack Onyx2 from a standard UK ring main wall socket quite comfortably. I really wouldn't worry about a deskside machine, it's no harder on the wiring than a 1KW fan heater. As long as you're pretty sure the wall socket is on a ring main you'll be fine.

Come to think of it, I ran a pair of desksides off a single wall socket during production of a recent-ish TV series.

3-phase (208V-or-thereabouts) is an unusual config, even for large SGI gear. Most common is single-phase 240V, although multiple racks can be run off of different legs of 3-phase service to equalize things out and make your local power company happy.

chancehooper wrote: ...as long as they got agreement from SGI/rackable. Which is feasible as they have no commercial interest in it, even to the point of ending MIPS/Irix support.

${same_old_boilerplate_about_sanitizing_the_code_and_finding_out_SGIs_licensors_costing_money_that_Ive_said_before}

fieldframe wrote: Did anyone else see this already? Read the paragraph at the bottom.

http://www.theregister.co.uk/2011/02/25 ... page2.html


Simon

Too bad Hamei's gone - it would be nice to see the inside scoop. I don't think he was ever able to track down one of the Godson things, was he?

SGI was the major former MIPS maker. The association would be natural. All the same SGI is highly unlikely to consider it. It would require retooling of their ASICS, dusting off MIPSpro and bringing it up to speed with the new chips, and you still have the uncertainty of any Chinese logic in there - the Chinese don't really have the best reputation for overdesigning and extensive testing, and little bugs get expensive very fast in SGI's market. Their buyers don't buy the old "it will be fixed in a later firmware update which may or may not actually happen" or the intimation that it's the buyers fault and they need to buy a new machine. How many millions did the original R10k fiasco cost SGI again?

PymbleSoftware wrote: Not only that, but they act like they hate MIPS and IRIX.... Or at least the local (Australian) office seems find it an offensive topic to bring up.

R.

I imagine that it is probably something like a singer or sports player when someone keeps dwelling on the past "glory days" and they're wishing someone would notice what they were doing now. Especially because the "now" stuff is what's paying their salaries.

PymbleSoftware wrote: The local fleabay and other auction sites often have Indys and O2s, very rarely anything more interesting or with more CPUs, therefore there seems to be liquidity in the low end market which means high-end stuff is more successfully hoarded or they simply made and sold a lot more of the lower end machines. For every Tezro or o300 that turns up there seems to be at least 5 to 6 Indys, maybe that is my bias.

R.

There are several things that tilt the scales in favor of unis. First, SGIs heyday was when multis were generally physically large, and large systems take up space so they tend to be surplused rather then kept and are less likely to be grabbed and held by a hobbyist.

Secondly, multiproc SGIs (and Alphas, and RS/6ks, etc.) are significantly more expensive than unis. SMP really only came down in price with the intro of multicore CPUs (let's admit here that from a technical perspective there is little difference between a multiproc/single core and single-socket multi-core except in system board/logic complexity). Companies would tend to get a few fast machines and more smaller machines that people would use to get stuff ready to run on the Onyx or Origin. Combine this with point #1 and you have fewer out the door plus fewer kept on the big boxes.

For SGI you also have the smaller multis (O300, Tezro) coming later when customers were moving away from IRIX. Yes, O200 and Octane SMP did come at the end of the heyday, but uff-da, the pricing on those was steep and you didn't have the capability that a machine such as the SS20 had of plugging in a second CPU later, as the SGIs relied on dual CPU processor modules or PIMMS to get a second CPU. (exception would be IP19/IP21/IP25 and possibly IP27, but for at least IP25 SGI left out board logic for CPUs that were not installed, effectively eliminating the chance of a later upgrade that didn't require a board swap).

I like to have a SMP example of everything, but it took awhile for the non-Intel and non-Sun archs. I now have a Challenge, duallie Octane and two Origins (O200 and O2k) with SMP, but for a long time I was uni-only for Siggys (still am for VAX, POWER, Alpha, HPPA and a couple other oddies). Still have more uni SGIs than multis, too.

Curses, Red Baron.

So close and yet unattainable.

Here's a good one on RealityEngine:
http://citeseerx.ist.psu.edu/viewdoc/do ... 1&type=pdf

I've got a good one on the design decisions behind the Origin3000, but I'll need to drag out the reference.

Technically PCs have come a long way. Hypertransport, PCIe, and well-done SATA have made life much easier - though it's true that these are rarely done properly on the el-cheapo models.

I knew things were getting better when I could finally do something disk-intensive and have the computer still be responsive at the same time. Note that this doesn't mean that things can't be done better, just that the state-of-the-art in "cheap but good enough" has risen to where it is not bad at all.

Surprised it took this long to make the forums.

PymbleSoftware wrote:

Geoman wrote: f***!! it is exploded

Apparently they don't use heavy water reactors , so it is not as bad as a Chernobyl type event... Still pretty worrying. But then just going on what I am reading about it...


R.

??? Chernobyl was RBMK gas-cooled graphite moderated reactors, probably leveraged from Pu-production reactors. Heavy water/light water doesn't make a difference, really.

I'm glad they weren't gas or sodium-cooled (sodium cooled for obvious reasons: leak+flooding=very bad problem, considering that the sodium is usually radioactive, and gas because the fuel cladding is usually not as good as is in place at a water-cooled reactor).

Here's the Origin 3000 one - it's really on NUMAflex, but it's by John Mashey and is pretty interesting.

http://homepages.cwi.nl/~robertl/mash/numaflex

Do you have access to any of the write-ups on Stanford DASH?

skywriter wrote: Unfortunately, it's not all a radioisotope of iodine, radioisotope of cesium has also be detected and it's half life is considerably longer than merely 8 days.

Does Cs accumulate biologically the same way? I know I and Sr are hazardous because they are either identical (I) or very similar (Sr/Ca) to elements that are accumulated.

Looks like there was a failure of the cladding, but less of the Cs will have been emitted than would have been possible in a Chernobyl-type failure because Cs is solid and there wasn"t an explosion.

I hate Windows' habit of munging up the shift key if you keep it pressed for a while while thinking

ianj wrote: I run an O300 and appreciate this information. I'd say it's worth waiting for the system to show up before buying any, though - I have a feeling many O300s were run as nodes in multi-unit setups, never had hard drives put in them, and hit the recyclers with their factory drive sled/air baffle inserts untouched (like mine did).

So SGI used "real" drive sleds as baffles on the O3x series? Good to know- on O2k and other systems they came up with placeholders that wouldn't hold a drive so you'd be more likely to buy a drive from SGI rather than add an off-the-shelf model.

Yep, clone it.

Have one build for each major CPU/graphics version and clone em.

IP22/R4k/Express (IP22 Indigo2 XZ/Extreme and all R4x00 Indy with XZ)
IP22/R4k/Newport (IP22 Indigo2 XL and Indy R4x000/XL)
IP22/R5k/Newport (Indy R5k XL, you probably don't have any Indy R5k XZs)
IP28/Express (Indigo2 R10k XZ/Extreme)
IP28/IMPACT (Indigo2 R10k all IMPACT)

should do it, as long as none of your Octanes are working (if they are add IP30/whatever graphics it is)

You can do RoboInst to automate installs, but the decompression/install/setup still takes a very long time. Choose the fastest CPU in the class to set up your image with (e.g. R4k/250 for IP22), and patch it before cloning. Set up your image on the smallest disk you have (make sure to give yourself enough swap, native swap is a bit faster than fs-based swap)(*if you're doing a dd-based clone, xfsdump/tar doesn't need this), then use xfs_growfs to expand onto bigger drives. Be sure to change everything you need to change.

200MHz is still usable under 6.5 for many things (they're a bit slow for what should be silly little tasks like webbrowsing, but overblown modern software such as Firefox and OpenOffice is to blame), but Octane or Fuel do have noticeably more kick.

Don't worry about the aftermarket sled - they work just fine except for being harder to install. I suppose you could look at the jumper on the Fujitsu as a plus, but I always looked on it as "our disks can't figure out the bus width on their own so you have to tell them". Then Indigo2's 10MB/sec SCSI bus does get a bit slow for big transfers, but for contemporary (mid '90s) stuff it's just fine. IRIX 5.3+ is new enough to where it's usable and can mostly be figured out by someone used to modern SysV-derived systems (for fun get something really old such as Xenix or SunOS 3 and see what inspired the UNIX Hater's Handbook). 256MB is OK for playing around, I'd stay stick with it and if you want to get something a bit faster later go for a newer system.

SGI freeware is old enough to be of limited use, go for Nekoware-MIPS3 when possible.