The collected works of GL1zdA - Page 1

Since this is my first post, hello to everyone!

I'm trying to boot Windows NT 4.0 on a 43P. I already got it running on a 7024-E30, but the 7043 has Open Firmware instead of the 'classic' one (though it's still PReP). I tried booting the ARC disk that worked on the E30, but it wouldn't go. Recently I found out, that on Open Firmware (at least on Motorola Powerstack) I should use the VENEER.EXE to run the setup. This is what I get:

I tried changing load-base and real-base, but without luck. Does anyone know how to do it?

Is there someone who might try to run this on his machine (if not a 43P-140, then maybe on the 7025-F40, which is also PReP with OpenFirmware)? Just to make sure, that it isn't fault of the one that I use?

In the NT 3.1 Service Pack 3, there is a HALSGI.DLL. In the July 92 NT 3.1 beta there is a file called "SGIREX.SYS", hex-viewing it reveals a suspiciously looking string LG1Base - it seems it is an NT driver for Newport graphics (REX3 was the rasterizer chip, LG1 is the codename of one of the Newport cards). Also in the leaked NT sources are source files for the SGI HAL with comments, that they are for use with the Indigo. There were some demos of Indigos running NT. Isn't it possible to change the the endianness of Indigos?

This is still the 604e. CPU Cross Reference RS/6000 PCI Bus Systems contains the CPUs you can use in the 7043-43P.

Given manufacturers sell some better parts as low end parts because there is higher demand for them, it makes sense to do some money on unlocking them. Until now it was always a gamble - you could unlock some chips, but there was a possibility that they were locked because of QA, now you pay for it but you can be sure of the results. It's easy for the customer, because they don't even have to open the chassis to change the CPU, it's good for Intel, because customers don't buy cheap second-hand CPUs for upgrades (and I guess *THIS* is the target - Intel haven't got its share when people bought used CPUs).

Martin Steen wrote: So, theoretically, you could upgrade a cheap Geforce-card to an expensive Quadro card just by
changing the BIOS.

This worked for cards until at least GeForce 4, though I'm not sure if there might have been licensing issues if you softmodded a card.

theinonen wrote: Personally I am against these artificial upgrades. It just shows, that Intel could have sold the same features unlocked at the first place with a lower price. Either way it costs the same to manufacture, and actually the cheaper processor may cost more as parts of it have to be disabled.

But it's the same with software - you have all these Standard/Premium editions to be flexible - you buy features you need or can afford. Intel gives you the same - you might use it if you desire - if not, it won't cost you anything. It's nothing new - when buying networking hardware you often will be able to upgrade licenses to service addition workstations, even if the hardware was always capable of servicing them.

Judging by the comments on engadget people still think about computer hardware as a piece of metal instead of thinking in terms of processing power, time etc.

rusti wrote: 3) It will keep him from having a disk crash ever again - This is what I would like to comment. AFAIK these flash storage things have a maximum amount of write operations before thay, too, can very well fail. Is this number so big that you can rightfully assume that in a real life application the SSDs will last indefinitely. From what I read so far I got a different impression so I am curious why nobody ever commented on this assumption.

But there is one, very important difference - after it reaches the maximum number of writes it will become 'read only' (not only by Ontrack).

http://www.sgi.com/products/legacy/multilink/cards.html

VPro™ V3 (GeForce 256)
VPro™ VR3 (Quadro)
VPro™ V7 (Quadro2 MXR)
VPro™ VR7 (Quadro2 Pro)

There's probably also a V5 based on GeForce 2 (guessing by the VW230 drivers).

Except for IrisVision I don't know about any other SGI graphics ported to PC.

Voted uniprocessor. Have no multiprocessors. But I hope I will get one someday.

I'm looking for the SGI Technical Report papers. I found some, but I would like to get some newer ones (for some reason I can't find anything beyond Onyx2) and the very old ones. Here's what I have already:

Workstations:
The Personal IRIS - A Technical Report
Indigo Family Product - Technical Report (incomplete)
Indigo Family Product - Technical Report (scans)
Indigo2 and POWER Indigo2 Technical Report (2.0)
O2 Technical Report

Visualization/Servers:
POWERSeries Technical Report
SkyWriter Technical Report
The POWER CHALLENGE (TM) Technical Report
The POWER CHALLENGE (TM) Technical Report (2.0)
PowerChallenge Array Technical Report
Onyx2 Techical Report
Origin200/2000 Servers Technical Report
Spider: A High-Speed Network Interconnect ( Google Scholar search )

Graphics:
Elan Graphics - Technical Report (incomplete)
Elan Graphics - Technical Report (scans)
Infinite Reality Technical Report
RealityEngine graphics

Thanks - if anyone can recommend any papers from ACM or IEEE - my university has a subscription, but since I will get my degree soon I won't be able to access it anymore. I downloaded some papers already (mostly by searching for names like "Origin", "Onyx" or "Altix") but maybe I missed something important.

Could someone with admin rights add some citation templates to our wiki? They're really helpful for adding references.
Here's how to do it:
http://www.mediawiki.org/wiki/Help:Temp ... to_another

Citation seems to be a good all-rounder, more are here .

nekonoko wrote: Edit: False alarm; looks like it got it.

Thanks!

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

You mean this: The Stanford DASH multiprocessor ? This one is freely available, I will have to ask a friend to browse the ACM/IEEE papers, because my university already revoked my access (graduated today).

Java Enterprise developer for 5 years.

PymbleSoftware wrote:

fu wrote: you've been wikifying non-stop, got to rock

Pete has let me be admin, so its kinda my duty.
... after two weeks a few hours each day cleaning up red-links I get kinda grumpy when people introduce new ones.

But that's the whole point of a wiki - if you see a red link you know you can write an article. Because it is red it doesn't mean it is bad (maybe the CSS should be changed to make green links ). http://en.wikipedia.org/wiki/Wikipedia:Red_link

PymbleSoftware wrote: Go back through the history of the wiki... You will find that for two weeks, years ago, I spent about 2 or 3 hours a day at least, putting every article in a category and every category in a category and generally every article had a path to it, there was probably hundreds of edits for that... Then for another 2 week period I went through again several hours each day and took red links out of articles, or created new articles.. at one point I made about 10 new articles a minute clicking red links..There was again probably hundreds of edits. I really think I don't need a lecture from you.

I was talking about my edits which never introduced non SGI red links.

Which EMU card is it? I assume it's EMU10K2 based (Audigy 2 ZS like)?

Ryan Fox wrote: Hmm I think your best bet would be the ACARD SCSI to IDE bridge, then a CF to IDE, one the IDE side of the ACARD
Got this working on an Apple 6100/66 , so it should, without a doubt work on a SGI indigo , or Indy

Does anyone know a cheap source of these ACARD SCSI to IDE/SATA bridges? Because the one that would be best for an SSD drive in the Indigo, the ARS-2000SUP , would cost me as much as I payed for my Fuel.

noth wrote: Has anyone got a suggestion on what I may need to change to be able to install this quite curious OS ?

Have you tried at the betaarchive - it seems it can be done with older versions VirtualPC.

UnixHD wrote: I have a machine that requires AIX 1.3 and I but lost the media...help!!!

Look on 9595 - Ardent Tool of Capitalism .

zafunk wrote:

GL1zdA wrote: Does anyone know a cheap source of these ACARD SCSI to IDE/SATA bridges? Because the one that would be best for an SSD drive in the Indigo, the ARS-2000SUP , would cost me as much as I payed for my Fuel.

I too have been looking at the ARS-2000SUP for an Indy, but the $250- price tag + $X- for a hard disk is quite prohibitive.

If you don't mind using IDE, the ACARD 7720U does show up on Ebay from time to time. Ian has a writeup about installing one in an Indy .

I decided to get a SCA drive and a SCA->SCSI Ultra converter. Unfortunately a simple 80->50 won't fit physically into the drive sleds in the Indigo, I will have to try a SCA->68/50 converter, which has the 50 pin connector a bit higher.

zafunk wrote: I've also tried the Yamaha V769970 in the Indy and was able to install IRIX onto the drive. I don't remember why I didn't keep that setup, aside from it not really fitting inside the chassis in any sensible way.

I've seen that one, and it's designed for a 5.25 bay, so it's not very useful for me.

A while ago every time I google for Origin or Altix, a website with a annually comparison of supercomputers showed up in the results - talking about the architecture of basically all produced high performance computers - Origin, Altix, Superdome, SX-6, Crays etc. The problem is, I can't find it now and I haven't bookmarked it, so I don't even now the link for archive.org, in case it was taken down. I think it was done by a person working at a university but I'm not sure, there were no pictures, only text on a beige background. Does anyone remember, what site it was and maybe knows the link?

Martin Steen wrote: http://www.top500.org

You mean this site?

Unfortunately no, this would be to obvious
It was not just a list comparing FLOPS etc, it was more about the architecture of the "mass" produced high performance computers like those I enumerated in my first post. It was about these manufactured in the year when the report was written and about those which disappeared during the last year. I think there were reports for most years during the 2000-2010 decade. I'm not sure if I remember correctly, but I think the last time I visited that website, there was a note, that the man who have done it, retired from the university and there were no new results published during last 2-3 years.

smj wrote: Hope somebody with a clue responds, it sounds like it would be a good thing to harvest and instantiate elsewhere...

I spent about 10 hours searching (it seem to be gone), I looked through bookmarks on my computers and haven't found it. I never found a link to this site, it always appeared in search results, so unless someone also bookmarked it will be hard to find a link and recover it from archive.org... I hate the volatility of the Web.

Finally, found it buried in my bookmark collection!
http://www.euroben.nl/reports.php - it's down, but the reports in PDF were archived !
So a quick peek at one - it was Aad J. van der Steen, his website at the university was deleted:
http://www.phys.uu.nl/~steen - but again, it was archived
And fortunately, the link from the "farewell"page - http://www.hpcresearch.nl - is up! And there are the reports from 93 to 2011 .
Now I know why Google didn't work. The HTML version at http://www.phys.uu.nl/~euroben/reports/ ... erview.php was deleted and the files (the directories web00 through web11) are on the new server http://www.hpcresearch.nl/euroben/reports/ but can't be viewed (403).
Some of them were archived:
2000 @euroben.nl
2001 @euroben.nl
2002 @phys.uu.nl
2003 @phys.uu.nl
2004 @phys.uu.nl
2005 @phys.uu.nl
2006 @euroben.nl
2007 @phys.uu.nl
2008-2011 are missing (it seems pre 2000 were not available in HTML), but everything is present at EuroBen Benchmark reports and Overview of recent supercomputers . I like them, because you can read about the machines at a specific moment of time and see how the whole HPC industry evolved, not only a particular architecture of one manufacturer.

PS.
Even TOP500 has a mirror of 2006/2007 reports, though I like the layout of the original page more: http://www.top500.org/resources/orsc
Now I also know, why it was so hard to find the page - Google seems to rank low these PDFs at hpcresearch.nl - and no, there is no ROBOTS.TXT at that website.

Use a pointer int*, pass inner dimension as argument and access it like array[i * width + j]. (the compiler will complain about passing int[][] as int*, I'm not sure how portable it is).


This:


does exactly nothing.