The collected works of jsloan - Page 1

I would really, really love to put a SSD into each of my indys ... not only are they screaming fast, but they're silent as well.

And it would be very nice not to worry about disk crashes [1].

But I cannot find any scsi SSDs ... the only ones I see are these ridiculous external disk boxes that cost a million dollars, etc.

I use the Intel SSDSA2MH080G1C5 in SATA systems and it is great ... I really just need a scsi version of that to start plugging into my sgi/sun machines.

Do they exist ?


[1] I know all about the write-resiliance issues with flash media. I am aware of those risks.
Ok, they exist :)

Here are the scsi narrow, 50pin models (2.5"):

http://www.bitmicro.com/public_docs/pro ... 2S20BL.pdf

and the u320 model (3.5"):

http://www.bitmicro.com/public_docs/pro ... S320EL.pdf

So that's that. I have not looked up a price, but since they have relatively low capacity models (8 and 16 GB, etc.) I am thinking they are not outrageous... could probably be used in any desktop SGI/Sun application.
SAQ wrote:
Gray Fox wrote: Wont SWAP help end the SSD life span sooner?


Not if you disable it. Although Indy's 256MB memory limit doesn't make it the best machine to run without swap.



Hmmm... I am under the impression that modern SSDs deal with swap much better using their wear-routines ...

It's still a problem, but not nearly as acute as, say, a sandisk CF card sitting in an adaptor.

I'll look into it...
sybrfreq wrote: I had the exact same thing happen when I tried my octane with a dell P780. It does work fine on that display's bigger brother, the P990. I think the consensus is that you need a real designed-for-SGI 13w3 to vga adapter and not one made for a sun.


Why is this ? Just curious...
Sun makes two big 48 disk "thumpers" - the Sun x4500 (thumper) and the Sun x4540 (thor).

These are starting to show up here and there "bare", with no drives, and I could see buying one and supplying my own disks.

But I've never been inside of one, and further, have some specification clarifications ... if anyone can help:

- Are both of them SATA only on the inside ? As in, you cannot put a SAS disk into one of the 48 slots ?

Most important, however, is that I do not plan on running ZFS ... which means I need an actual hardware raid controller (not interested in vinum, etc., thank you). But how does this work, physically ?

Do each of the 48 drives have a sata port on the backplane, underneath, and I can plug my raid cards leads into each of the 48 slots ? Or are there 4 or 6 or 8 larger, SATA-aggregator ports somewhere on the backplane and I can go from the raid card to the backplane with some of those ?

Basically, how do I connect an add-in raid card to the internal drives of a thumper/thor ?

And finally, are there any cards that would even work ? I say that because the thumper has only 2 LP pci-x slots, and the thor has only 3 half height pci-e slots ... and all of the 16 or 24 port raid cards I have ever seen are full height ...

Looks like both 3ware and areca have only full height options for any more than 8 ports ... which is why I am curious about the ability to do internal SAS drives, since I wouldn't need a physically large card to do that ...

Comments ?

Thank you.
yeah, and there are some issues in our organization with tools and work flows, etc., that are all going to break with ZFS.

This isn't knocking ZFS - we're excited about it - but for this application we really do need physical raid controllers. I'm just not sure if it's possible to address 48 drives with physical controllers in a thumper, and am trying to find out how...
Ok.

Well, the thought was that since the x4540 (the "thor") has 3x pci-e, that there _might_ be some half height cards out there that had 16 ports each, but this does not seem to be the case - at least not with 3ware nor areca.

And the other thought was that there was some way to interface with the backplane, and those pics show that there _is_, but those:

http://www.flickr.com/photos/digitar/34 ... 455309524/

don't look like standard interconnects that I could get off of a raid card.

So it looks to be a bust from two different angles - at least if I want UFS2 on hardware raid.

Drifting a bit off topic ... isn't it surprising that in the 3+ years that the thumper has been out, nobody has cloned it ? When they were first introduced, I thought for sure we were only 6 months away from a 4u, 48 disk clone from supermicro ... and now all these years later, and nobody has brought one out. Isn't that odd ?
5u instead of 4u, but possibly a solution:

http://www.polywell.com/us/rackservers/ ... 048AIS.asp

still, I'd prefer a supermicro, as I trust their brand ... look at what they came up with:

http://www.supermicro.com/products/chassis/4U/?chs=847

24 in front, and 12 in back, for a total of 36 ... a little odd.
I have a nintendo 64 Indy development station, and I _believe_ I have all the software/SDK for it, but I am ALWAYS looking for more nintendo related SGI hardware and software.

I specifically need the breakout cables that go from the dev board to the controller port(s), and I'd love to have a backup dev board, and anything else nintendo / SN systems / Partner / etc. that you may have.

Thanks.
yes, definitely. I never post though, just lurk.
what were these used for ?
I just purchased:

http://cgi.ebay.com/ws/eBayISAPI.dll?Vi ... 0508982645

and the seller has the specs as dual-v12-DCD, including the special card carrier that is required for that.

Can anyone look at the rear-facing pic he has there and confirm that for me ? I've never seen that config in real life...
thanks! Looking forward to connecting it to a video wall...
I would be interested, and would even PAY the shipping, but _successfully_ shipping it without breakage/damage is the real issue...
Folks - remember that IDE SSDs _do_ exist, and are fairly cheap, etc.

So instead of SCSI -> IDE -> CF, you can just simplify SCSI -> IDE, and I suspect with a lot fewer compatibility issues.
Are you really in tokyo ? I may have an extra one of these, but I don't have the time or energy right now to properly package and ship, etc.

If you were in the SFBA, on the other hand ...
zizban wrote: I am the documentation lead for the CDE project. It's been a wild few months.

CDE runs on Linux. Here is a list of supported distributions:

https://sourceforge.net/p/cdesktopenv/w ... Platforms/

The code isn't from 1999. It's from 1995 :D

It's pretty stable on my Ubuntu 12.04 box. Somethings don't work but a lot does.


Out of curiosity, why was this released on sourceforge instead of github ?
Getting a r5k-180 indy all up and running with my n64 dev board.

My preferred shell is csh, and there is indeed a csh included with Irix 5.3, but it is a really bad csh ... no autocomplete, no up-arrow history ... nothing. What's the fastest, easiest way for me to get a more modern (say ... freebsd circa 2001 ?) csh on this 5.3 system ?

Second, what is a 'find' syntax that I can memorize and easily use ? I tried to piece together a good syntax from the man page, but when I search for files I know exist, I come up empty. On modern systems I do things like:

find / | grep myfile

fine /usr/local | grep myfile

what's a good syntax for irix 5.3 ?

Finally, there is a n64-development specific command "dbgif" whose function is described as:

"It establishes a link between the host workstation and the game target. Specifically, it provides a link between gvd, the debugger running on the host development workstation, and the game system development board. Note that the dbgif program must be running in order to use the gvd debugger (GameShop)."

Ok, fine ... but when I run the dbgif command, it fails with a very short gethostbyname error. My indy has a hostname set (not a FQDN, but a simple hostname) and that hostname is tied to the fixed IP in /etc/hosts. Further, dns resolution with external nameserver is functioning (I can nslookup external hosts, etc.).

So, while I am sure very few people have ever seen this 'dbgif' application, it seems to be a pretty general client/server operation, where dbgif is presenting a resource for the gvd debugger to hook to ... any ideas why it would fail with a gethostbyname error ?

Thanks.
Thanks - I guess I'll just use bash from TGCware - that will at least get me up arrow history and so on ...

Then I can just create a sane find alias and use that - thanks for your examples.
Ok, here is what I got from 'par'


0mS par( 536): was sent signal SIGUSR1
1mS par( 536): END-pause() errno = 4 (Interrupted system call)
1mS par( 536): received signal SIGUSR1
1mS par( 536): sigreturn(0x7fff2aa8) OK
1mS dbgif( 536): execve(/usr/sbin/dbgif, 0x7fff2f50, 0x7fff2f58) OK
4mS dbgif( 536): open(/lib/rld, O_RDONLY, 04) = 3
4mS dbgif( 536): read(3, <7f 45 4c 46 01 02 01 00 00 00 00 00 00 00 00 00>..., 52) = 52
4mS dbgif( 536): lseek(3, 52, SEEK_SET) = 52
5mS dbgif( 536): read(3, <70 00 00 00 00 00 00 a0 0f b6 00 a0 0f b6 00 a0>..., 96) = 96
5mS dbgif( 536): elfmap(3, 0x7fff225c, 2) = 0xfb60000
5mS dbgif( 536): close(3) OK
5mS dbgif( 536): getpagesize() = 4096
7mS dbgif( 536): getpid() = 536 ppid=535
7mS dbgif( 536): syssgi(SGI_TOSSTSAVE) OK
8mS dbgif( 536): open(/dev/u64_debug, O_RDWR, 02000004764) = 3
8mS dbgif( 536): socket(PF_INET, SOCK_STREAM, 0) = 4
8mS dbgif( 536): gethostname(indy64, 64) OK
8mS dbgif( 536): open(/etc/resolv.conf, O_RDONLY, 0666) = 5
8mS dbgif( 536): fstat(5, 0x7fff1c90) OK
8mS dbgif( 536): getpagesize() = 4096
8mS dbgif( 536): brk(0x10137000) OK
9mS dbgif( 536): ioctl(5, TCGETA, 0x7fff1c38) errno = 25 (Not a typewriter)
9mS dbgif( 536): read(5, "nameserver\t192.168.1.1\n", 4096) = 23
9mS dbgif( 536): read(5, 0x10134b18, 4096) = 0
9mS dbgif( 536): close(5) OK
9mS dbgif( 536): gethostname(indy64, 256) OK
9mS dbgif( 536): getpid() = 536 ppid=535
10mS dbgif( 536): getpid() = 536 ppid=535
10mS dbgif( 536): time() = 1355791715
10mS dbgif( 536): getdomainname((null), 256) OK
10mS dbgif( 536): socket(PF_INET, SOCK_DGRAM, 0) = 5
10mS dbgif( 536): connect(5, <00 02 00 35 c0 a8 01 01 00 00 00 00 00 00 00 00>, 16) OK
10mS dbgif( 536): send(5, <00 01 01 00 00 01 00 00 00 00 00 00 06 69 6e 64>..., 24, 0) = 24
11mS dbgif( 536): select(6, IN:set=5 OUT:set=5, 0, 0, sec=5 usec=0) = 1
27mS dbgif( 536): recv(5, <00 01 81 83 00 01 00 00 00 01 00 00 06 69 6e 64>..., 1024, 0) = 99
28mS dbgif( 536): close(5) OK
28mS dbgif( 536): open(/usr/lib/locale/C/LC_MESSAGES/uxsyserr, O_RDONLY, 01752601414) errno = 2 (No such file or directory)
28mS dbgif( 536): write(2, "gethostbyname call failed", 25) = 25
29mS dbgif( 536): write(2, ": ", 2) = 2
29mS dbgif( 536): write(2, "Error 0", 7) = 7
29mS dbgif( 536): write(2, "\n", 1) = 1
29mS dbgif( 536): prctl(PR_GETNSHARE) = 0
29mS dbgif( 536): exit(0)


So it finds hostname just fine, and then correctly returns null for domainname, since there is none.

But it seems to fail on:

open(/usr/lib/locale/C/LC_MESSAGES/uxsyserr, O_RDONLY, 01752601414) errno = 2 (No such file or directory)

and indeed, I do not have a:

/usr/lib/locale/C/LC_MESSAGES

directory at all ... should I ?
robespierre wrote:
Irix 5.3 includes both AT&T csh(1) and tcsh(1). Only tcsh(1) has line editing and arrow keys.
ksh(1) also has line editing, but with the vi command set only. So you would use ESC k instead of up arrow.
[Linux, BSD, and Darwin don't have a real csh; if you check you'll see that csh is hardlinked to tcsh on these systems, and it mostly acts like tcsh.]



Is it safe to change roots shell to /usr/bin/tcsh ?

My /usr is a different filesystem, so I don't want to do something dumb...
I needed to reseat my RAM so I removed my cpu carrier module from my octane2 ...

Now that I have reseated it, I cannot reinsert the module.

I retract the two slider bars all the way, I slide the module in until it goes almost all the way in (about 1cm or less is left to go) then I insert the two slider bars as far as they go ... but that's it.

I cannot get the module to insert that last 1cm (slightly less than 1cm).

I really, really hate to force things, but I have put a fair amount of pressure on it, and cannot get it to go that last little bit.

Tips ?
Yes, those did "sag" a bit and I was unable to remove the module until I bent them back up. So that problem was solved.

But now, this is simply a failure of the clips/locks to engage and seat. The carrier is all the way inserted *except* for about 1cm or less... and I cannot get it to budge past that...
Thanks. I just got it to go - the manual makes it sound like you *don't* slide it all the way in before you lock in the long bars, but eventually I tried seating it all the way in before I slid the bars in, and then it worked.

So in summary, pull the slider rails all the way out, slide the carrier ALL THE WAY in, then slide in the bars and lock.
Vagabondo wrote:
Wouldnt it be possible to write a vfo with 1280X1200 and then use the dcd for dual input into the 30" since 1280 = 2560/2 and 1280/64 = 20??

sorry...just reread above and looks like you want 1600 vertical but you say the limit is 1200 with DVI? or is that just when using horz of 1920?



I am resurrecting this dead topic because I would like very much to do this ...

Is it indeed possible to drive a 2560x1600 monitor using both ports of a v12 DCD ?

I have a loaded octane2 with two v12 dcds in it and would like to try driving two of my dell 30" screens @ 2560x1600...

Thanks.

edit: my dell u3011 screens have two dvi inputs, which is why I'm hopeful there is a way to drive half the screen with one input... or perhaps there is a dongle/adaptor that exists that will combine dvi links into a single....
It does exist - if you look at firms that make matrix switchers and other video distribution components, it's a feature that exists ...

The problem is these devices are quite expensive and even more expensive if you add 2560x1600 support.
I have a maxed out octane2 (2x v12 DCD) that I am going to connect four monitors to.

Irix load is fresh ... nothing there yet.

What are some tools and/or software packages I could put on this to have some really visually appealing and impressive graphics demonstrations ?

Preferably something that can be scaled up to all four screens, etc.

VR simulations / demos ?

oil and gas visualization tools ?

What's floating around out there ?
I've been using google image search, etc., and I cannot find a picture of the interior of a desktop prism.

I got one on ebay ome time ago, and am just getting around to firing it up ... and before doing anything with it, I am looking at the interior and don't see anything that look like cpus ...

I see two fairly small chips on the motherboard with very thin heatsinks on them that look a lot like chipset chips ... I can't believe those are the cpus ...

If you have a desktop prism that is populated with cpus, could you take an interior picture ?

Thanks.
wait ... I just thought of something ... they're on the other side of the motherboard, aren't they ?

I won't be at my office to check for a few days, but the ram and cpus must be on the other side of the system, yes ?
great - thanks. I did indeed only open the left side of the system...

LOTS of pci slots in a prism, huh ? I should fill those up with modern FPGA cards and turn the old prism into a modern supercomputer :)
I just did 20-30 mins of searching around and did not find OpenGL Multipipe (or any software like that that could be downloaded from *.sgi.com...

Found plenty of documentation, but not the software.
Octane2 with 2 DCD v12 cards ...

I have two 24" 1920x1200 displays plugged into the two DVI ports on card "0".

All is well - I have a nice two-screen desktop with both panels driven @ 1920x1200.

However, I'd like to rotate both panels to vertical orientation... how hard is that going to be ?

Is this a custom VFO file kind of job, or is there some reasonable way to do this ?

Thanks.
I will turn them back horizontal :)
got it - thanks.
Looking for a SS20 with very, very good skins and spec'd decently.

So, dual cpus, but not so fast that they overheat and run the fans all the time, etc. - I think there were some problems with dual ross 200s ?

Max ram.

qfe would be nice.

TGX+ graphics ?

I am in California.

Thanks.
thanks - very helpful!

And yes, I do think I want a quad cpu configuration, as long as we're already in there maxing things out.

The key with the graphics is that I would like to try to achieve triple-head, somehow... possible ?
jpstewart wrote:
jsloan wrote:
The key with the graphics is that I would like to try to achieve triple-head, somehow... possible ?

Sure should be, in any of (at least) 3 different ways:

1) 3 x TurboGX SBus cards would likely be easiest to find and cheapest

2) 1 x VSIMM (for onboard SX) + 2 x TuboGX SBus cards

3) 2 x VSIMM + auxilliary video board (AVB, Sun part 501-2488, which is rare and expensive) (for 2 SX heads) + 1 TurboGX SBus


I think I am going the first route, with three tgx+ cards ... I have a spare one sitting around already...
Do not use raid5 for any purpose whatsoever.

Drives are too big to survive a rebuild these days. raid5 is useless, and beyond 4TB drives, raid6 is getting close as well...

If it's four physical disks, just do raid10, since raid6 gives you the same amount of space, with better performance.

If it is 6 disks or more, use raid6.

If you are from the future and reading this with disks larger than 6-8 TB each, raid6 is useless. Use raidz3 on ZFS or ... whatever exists in 2016 :)
quantity 20 of these are for sale:

http://www.ebay.com/itm/Sun-Sparcstatio ... 4ac045eea4

and:

http://www.ebay.com/itm/Sun-Sparcstatio ... 589ab166e0

I guess I have found the cosmetically perfect ss20 I was looking for ?

edit: they are aurora cases with 8mb vsimms installed. Amazing.
Yes, the 8mb vsimm is the real value there ... those are not so easy to find.