SGI: hinv

O2 with dual-channel Adaptec SCSI card (modified kernel)

My thanks to Jirka Tolvårige (omelett) for helping with this (he supplied the modified kenel file).

Here is an O2 with a dual-channel Adaptec 3940UW SCSI card, made possible by modifying the
adp78.o kernel file in /var/sysgen/boot. The replacement file is now available on my Depot Resources
page for download (backup the original somewhere before installing the new file).

Code: Select all

CPU: MIPS R5000 Processor Chip Revision: 10.0
FPU: MIPS R5000 Floating Point Coprocessor Revision: 10.0
1 300 MHZ IP32 Processor
Main memory size: 256 Mbytes
Secondary unified instruction/data cache size: 1 Mbyte on Processor 0
Instruction cache size: 32 Kbytes
Data cache size: 32 Kbytes
FLASH PROM version 4.18
Integral SCSI controller 0: Version ADAPTEC 7880
Disk drive: unit 1 on SCSI controller 0 (unit 1)
CDROM: unit 4 on SCSI controller 0
Integral SCSI controller 1: Version ADAPTEC 7880
PCI SCSI controller 3: Version ADAPTEC 7880
PCI SCSI controller 4: Version ADAPTEC 7880
On-board serial ports: tty1
On-board serial ports: tty2
On-board EPP/ECP parallel port
CRM graphics installed
Integral Ethernet: ec0, version 1
Iris Audio Processor: version A3 revision 0
PCI Adapter ID (vendor 0x9004, device 0x8078) PCI slot 1
PCI Adapter ID (vendor 0x9004, device 0x8078) PCI slot 2
PCI Adapter ID (vendor 0x1011, device 0x0001) PCI slot 3
PCI Adapter ID (vendor 0x9004, device 0x8278) PCI slot 4
PCI Adapter ID (vendor 0x9004, device 0x8278) PCI slot 5
Video: MVP unit 0 version 1.4
with no AV Card or Camera.
Vice: TRE


By default the Adaptec card only has one of its channels presented externally, so to make use of it one could
simply route a cable from the other channel to an external socket rigged to fit in the slot normally occupied by
the 1600SW adapter or dual-head board, or one could cut the casing in some other manner. Using a simple
cable extender, I found that for an R5K system I had to cut away one of the thin metal strips from the edge
of the PCI tray (this is easy to do, just wobble it bac & forth with a pair of pliers and it will fall off), while for
an R10K system I didn't need to make any changes, though routing the cable must be done carefully so as not
to impede the cooling of the CPU.

Either way, this is a simple method of having better disk I/O, ie. three external channels for storage.

A warning though: be careful about buying a 3940 card as some of the ones advertised are not what
they seem, usually misdescribed by sellers. The card should look like this .

Cheers! :)

Ian.
mapesdhs wrote: My thanks to Jirka Tolvårige for helping with this (he supplied the modified kernel file).Ian.

Nice, I'm very happy this works. I buyed last week the same card off ebay.de for 6 €.
Mac Version Roms.
May be Fu can put the 3940UW to The O2 Hardware Aggregator?
omelett
Nice work Ian!

O2 Hardware Aggregator is absolutely a good idea.

Ian, you might also consider asking Jirka Tolvårige if he'd allow you to post the specifics on how adp78.o was modified. It might prove helpful to someone trying a similar mod on yet another piece of hardware.
***********************************************************************
Welcome to ARMLand - 0/0x0d00
running...(sherwood-root 0607201829)
* InfiniteReality/Reality Software, IRIX 6.5 Release *
***********************************************************************
recondas wrote: Nice work Ian!

O2 Hardware Aggregator is absolutely a good idea.

Ian, you might also consider asking Jirka Tolvårige if he'd allow you to post the specifics on how adp78.o was modified. It might prove helpful to someone trying a similar mod on yet another piece of hardware.


I've installed the GNOME Hexadecimal Editor ghex 1.2.1
Then I made a copy of the adp78.o file from the /usr/cpu/sysgen/IP32boot directory.
All files in there are read-only, so I modified it with the chmod command -

Code: Select all

chmod 666 Copy_of_adp78.o

Then in ghex - Open file - I choosed the Copy_of_adp78.o
Then I searched the Cipher 8178 and replaced it with 8278, like recondas says in the thread (Link below). (Edit-Find, then 8178 -search for hex-data) I had to do it twice, this Number appears at two places.
Then I saved the file, but I forgot to chmod it back to 444 /:) and send it to Ian.

viewtopic.php?f=3&t=17554

omellet
Jirka Tolvårige?! That's not a name... (er det?!)
omelett wrote: (Edit-Find, then 8178 -search for hex-data) I had to do it twice, this Number appears at two places.


For reference, I diffed the files and here are the two instances:

Code: Select all

-0014420 3405 8178 0000 3825 3c06 0000 0c00 0000
+0014420 3405 8278 0000 3825 3c06 0000 0c00 0000

-0015260 3409 8178 1049 0065 0240 2025 0000 2825
+0015260 3409 8278 1049 0065 0240 2025 0000 2825
Twitter: @neko_no_ko
IRIX Release 4.0.5 IP12 Version 06151813 System V
Copyright 1987-1992 Silicon Graphics, Inc.
All Rights Reserved.
got two
so just time to wait come from canada and i'll try it out inside my o2's
thx all
Coffee for the mind, pizza for the body, sushi for the soul

:O2+: r7k350 :O2: r7k600 :O2: r12k400 sw1600 :Octane2: 2x400 v10
radrob wrote: got two
so just time to wait come from canada and i'll try it out inside my o2's
thx all


I hope they're the right model!

Hey, with respect to your 600MHz O2, any chance you could run the Blender test on your
system please? I'd really like to know how it compares to an R12K/400 and the R7K/350. See:

http://www.sgidepot.co.uk/blender.html

Test files:

http://www.sgidepot.co.uk/depot/blender ... ips.tar.gz
http://www.eofw.org/bench/test.blend

Thanks!! 8)

Ian.
thanks to dave for the heads up, this is now on the The O2 Hardware Aggregator
A follow-up to this with some solid results...

I tested an R5K/300 O2 fitted with the dig vid board and dual-channel Adaptec, with an array containing eight old/slow 73GB drives,
running Illusion 6.1v8; diskperf gave:

Code: Select all

# req_size  fwd_wt  fwd_rd  bwd_wt  bwd_rd  rnd_wt  rnd_rd
#  (bytes)  (MB/s)  (MB/s)  (MB/s)  (MB/s)  (MB/s)  (MB/s)
#---------------------------------------------------------
1245184   50.89   57.31   49.65   49.15   49.64   50.25
2490368   52.83   60.90   52.19   55.90   52.25   56.42


which is more than enough for D1 PAL. I used an Octane with DIGVID as a D1 source and then tested Illusion on the O2 for real-time
capture and playback, which did indeed work ok with the material stored as RGB frames.

Biggest difference though is the Octane (a single-R12K/400) was - for example - way faster than the O2 for exporting a clip
out as Targa files, and the GUI interface responsiveness was far superior on the Octane.

But for practical purposes, the O2 does indeed work fine for real-time capture using one of these Adaptec dual-channel cards
and the kernel mod. Here's the O2's hinv, notice the extra SCSI channels:

Code: Select all

CPU: MIPS R5000 Processor Chip Revision: 10.0
FPU: MIPS R5000 Floating Point Coprocessor Revision: 10.0
1 300 MHZ IP32 Processor
Main memory size: 1024 Mbytes
Secondary unified instruction/data cache size: 1 Mbyte on Processor 0
Instruction cache size: 32 Kbytes
Data cache size: 32 Kbytes
FLASH PROM version 4.18
Integral SCSI controller 0: Version ADAPTEC 7880
Disk drive: unit 1 on SCSI controller 0 (unit 1)
Disk drive: unit 2 on SCSI controller 0 (unit 2)
CDROM: unit 3 on SCSI controller 0
Integral SCSI controller 1: Version ADAPTEC 7880
PCI SCSI controller 3: Version ADAPTEC 7880
Disk drive: unit 1 on SCSI controller 3 (unit 1)
Disk drive: unit 2 on SCSI controller 3 (unit 2)
Disk drive: unit 3 on SCSI controller 3 (unit 3)
Disk drive: unit 4 on SCSI controller 3 (unit 4)
PCI SCSI controller 4: Version ADAPTEC 7880
Disk drive: unit 1 on SCSI controller 4 (unit 1)
Disk drive: unit 2 on SCSI controller 4 (unit 2)
Disk drive: unit 3 on SCSI controller 4 (unit 3)
Disk drive: unit 4 on SCSI controller 4 (unit 4)
On-board serial ports: tty1
On-board serial ports: tty2
On-board EPP/ECP parallel port
CRM graphics installed
Integral Ethernet: ec0, version 1
Iris Audio Processor: version A3 revision 0
PCI Adapter ID (vendor 0x9004, device 0x8078) PCI slot 1
PCI Adapter ID (vendor 0x9004, device 0x8078) PCI slot 2
PCI Adapter ID (vendor 0x1011, device 0x0001) PCI slot 3
PCI Adapter ID (vendor 0x9004, device 0x8278) PCI slot 4
PCI Adapter ID (vendor 0x9004, device 0x8278) PCI slot 5
Video: MVP unit 0 version 1.4
AV: AV2 Card version 0, Camera not connected.
Vice: TRE

Graphics board 0 is "CRM" graphics.
Managed (":0.0") 1280x1024
32 + 32 bitplanes
board revision 2, CRM revision C, GBE revision B
Monitor 0 type: Unknown
Channel 0:
Origin = (0,0)
Video Output: 1280 pixels, 1024 lines, 50.00Hz (1280x1024_50)



Cheers! :)

Ian.
Wait a second -- you said only one of the two channels was external. Did you hack a flatcable into the case or find different Adaptec?
Now this is a deep dark secret, so everybody keep it quiet :)
It turns out that when reset, the WD33C93 defaults to a SCSI ID of 0, and it was simpler to leave it that way... -- Dave Olson, in comp.sys.sgi

Currently in commercial service: Image :Onyx2: (2x) :O3x02L:
In the museum : almost every MIPS/IRIX system.
Wanted : GM1 board for Professional Series GT graphics (030-0076-003, 030-0076-004)
jan-jaap writes:
> Wait a second -- you said only one of the two channels was external. ...

That was 2 months ago. :D


> ... Did you hack a flatcable into the case ...

Of course! 8)

Ian.
Hi Ian, I'v run blender render test on my O2 wit these results:
10:58.76 when resolution was set to 1280x1024_75
10:56.39 when res was set to 640x480_60

None CPU intensive background processes was running, whole system was only rendering.
:O2: R7000/600 576MB Ram CDRW 18+9Gb HDD
http://www.tomosgi.co.cc
Thanks!! What version of Blender did you use? (2.44 I hope)

Is there a hinv of your system somewhere I can link to on my results page?


Follow-up for this thread's main topic...

I tried Illusion on an R12K/400 O2 with the same dual-channel Adaptec card and the performance was
much better for using the application, but about the same for diskperf results. I'm wondering if accessing
both channels through one 32bit/33MHz PCI card is causing an issue somehow. Will try next with the
built-in UW channel plus just one of the extra card's channels...

The R12K did make a huge difference for using Illusion. Much more responsive menus, and the app loaded
up almost as fast as it does on an Octane/400.

Ian.
Yes blender was 2.44 downloaded from your site, no changes made to scene or to settings. I'v just pushed F12 key and wait...

and here is link to my actual hinv http://forums.nekochan.net/viewtopic.php?f=14&t=16718908
:O2: R7000/600 576MB Ram CDRW 18+9Gb HDD
http://www.tomosgi.co.cc