NTT Indy Mysteries

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
A while back when visiting an ex-SGI employee, I obtained three NTT Indys. These were not the usual "branded" configuration of the Indy, but instead are black with the NTT logo, and are in a set-top box configuration with an IR port on front and a custom video board, with composite outputs and an ATM network connection.

The story about these NTT Indys that I was told, was that SGI collaborated with NTT (the Japanese telecom) to develop them as an ATM video conferencing solution. NTT would provide the ATM network, and SGI would provide the hardware. The ex-SGI employee I talked to was a member of the project (as well as the WebForce project, etc.). Apparently the video conferencing initiative never panned out, but he still had three of the prototype units.

I will put up pictures eventually (and Ciao has some, as well as one of the units I gave to him), however it's worth noting that it's not just the hardware in these units that is custom, but also the boot PROM as well.

Here is what is presented over the serial console at boot time:
Code:
PROM Export Version 2.7
Init took 629 msec

PROM menu

1) Boot settop
2) Enter PROM monitor
r) Restart PROM

Option?
The machines want to boot over the network and do not contain a hard drive or any local IRIX installation:
Code:
Option? 1
Starting settop download.
Loading params bootp()nt:bootparams.08:00:69:09:46:f8
Cannot open 'bootk(nt:bootparams.08:00:69:09:46:f8)nt:tvkern' - server not responding
bootk(nt:bootparams.08:00:69:09:46:f8)nt:tvkern: boot error - server not responding
  
PROM menu
[...]
PROM monitor options and environment variables are somewhat different...
Code:
Option? 2
PROM monitor, type 'exit' to return to menu.
>> help
Commands, args within [] are optional
  setenv [-p] var val           - Set environment variable
  unsetenv var ...              - Remove environment variable(s)
  printenv [var ...]            - Print environment variable(s)
  resenv [-p]                   - Reset environment variables and NVRAM
  version                       - print verbose version of PROM
  date                          - Print date of RTC chip
  boot                          - boot system ($bootdev/$bootfile)
  off                           - Power-down
  hinv                          - Hardware inventory
  exit                          - Return to PROM menu
  res                           - Restart PROM
  resa                          - Reset ATMizer
  loada [-d] bootp()[host:]file.xx [[bootp()[host:]]file.cnf]
                                - Load ATMizer code
  help                          - Print command overview
  ?                             - Print command overview
Devices, args within [] are optional
  bootp()[host:]file
  bootk([[host:]paramsfile])[host:]kernel
  boota([bootvpi],[bootvci])[host:][kernel]
>> printenv
dbaud=9600
rbaud=14400
netaddr=208.99.99.2
eaddr=08:00:69:09:46:f8
autoload=n
console=d
bootdev=bootk(nt:bootparams.08:00:69:09:46:f8)
bootfile=nt:tvkern
vcxo=0x7ed
volume=87
nbootries=9999
passwd
interactive=y
tvmode=ntsc
usedebugger=1
ng1=ntsc
ide=g
logtype=3
logsize=512
verbose=BT
noir=1
ipfsn=0
>>
They also output a bit of video out from the composite graphics port, which I will capture and upload here along with pictures of the unit, at a later date.
 

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
It's also worth mentioning that Ciao's unit has some problems with the serial console that mine does not. I suspect timing issues, as the text is often corrupted / garbled, with occasionally readable sections. Perhaps a slipping UART clock?

In any case, as can be seen by the scope capture below, the async serial timing on mine is pretty good! 104.3us is 9588 baud; of a 9600 baud target, well in tolerance.

2
 

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
The composite output comes from a separate board that isn't a regular graphics adapter and thus would not show the PROM boot menu or any IRIX boot messages / login screen. There is likely special software and drivers on whatever the netbooted image would be that determine what to display from the composite output and how to deal with the inputs.

Ciao took a picture of the composite output, below:
125
 

TriOx

New member
Aug 1, 2019
9
9
3
I had one of these set top boxes, back in the day, but it didn't have a black case. The case was normal Indy blue but zero badges or logos. At the time I was looking for R4400 and R5k cpu modules and remember being quite dis-appointed with both the R4600-133 and the video board not being usable in Irix.
 

CiaoTime

Public Enemy Number One
Jan 15, 2020
45
27
18
Vancouver, BC, Canada
Good evening! It's been a few months since I've played around with the NTT Indy system, but I'll share some of my experiences with it here. Let's start off with a picture of the case.

NTTfront.png

Internally, the NTT Indy is a stock Indy for the most part: the motherboard and CPU are normal, and predictably rather low end. (R4600, 133mhz, 16mb RAM.) There are no physical provisions to install a floptical drive nor a hard drive internally: the normal mounting cage for a hard drive isn't present inside.

Beyond the custom boot ROM, the main component that makes the NTT Indy special is the "Bambi" daughterboard, which is a large single-height GIO32 board that takes up much of the space inside of the system. It provides composite video input/output/passthrough, an integrated Fore ATM network interface, an infrared receiver (presumably for some sort of remote control), and a few other functions. Here's an internal picture of the system with the Bambi board installed.

NTTbambi.png

Here's a close-up view of the ports at the back of the system.

NTTbackpanel.png

The ports on the top row are all connected to the Bambi board. Ports A1 and A2 are intended to be connected with an audio patch cable, and ports B1 and B2 are also wired together via a separate cable. Elf thankfully has one instance of the B1 to B2 cable, which I was able to pin out while I was visiting. It's also worth mentioning that the usual keyboard and mouse ports are present on the motherboard itself -- but they are blocked by the sheet metal of the case!

System startup is relatively uneventful - the composite output displays the Zoetrope splash screen once it's initialized, while the 13W3 output displays a garbled version of the same thing. This is from the 13W3 output:



Beyond that, the system waits for a serial command once it's initialized. As Elf had mentioned, the only options are either to enter the PROM or to attempt to netboot off of an as-yet unknown device.

Unfortunately, my particular Indy has an issue when connecting to it over serial, which leads to some rather... interesting behaviors, in the PROM.

Code:
NTT#1 - bilerhahn, built ed ep 20 12:5:01 DT 1995
                                                 CP board
                                                           CP is 13 MHz 4600
  Primary d-cach: 16 kbytes  Seconary cche: 0 kbytes
                                                     Memory sze: 16 Mbytes
NG1 board
 Bt5 rev 0x, VC2 r0x0
                       No extenaLL CS1562, $ics is inor!
 Using eault28ca06 forBt445PL
  Bt445 ECL(from IC1562) clock np
Bambi boa
  PC15   XND loadd
                    MVU rev02
 rookTre Encder initializd
  AMizer Frmware Rev▒▒▒▒
                        ▒r▒R>> hnv
PROM xportVeron 27
                  NTT #1 - builder hahn builtWed Sep20 2:58:01 PD 995
CPU board
           CPUis 133 MH R00
                            rimaryi-cache: 1kytes
  Scndar cche: 0 kbyteses
  Mmor sie: 16 Mbyts
  Bt445 re 0xd, V2rev0x
  No external PLL(CS1562), $ics is igned !
                                            Env variable $pll is no set
 Usingdault 28ca06 for B445 PLL
                                 Bt445 ECL (from ICS1562) clock inpu
ambibard
 HPC.5
 XNDloaded
            MVre 0x2
 BrookTree Encode initilize
  ATMizer Firmware isi 4.
So, it's a bit of a mess, and makes it impossible for me to easily dump my particular system's ROM at this time. It's only an issue with serial transmission and loopback: the system itself does seem to receive serial inputs perfectly well. Strange stuff.

I haven't yet tried to install the Bambi board in a system running IRIX yet, though I have doubts as to how much of it will function without drivers and/or the custom boot ROM.

To end off, here's a link to all of the pictures I've taken of the machine, as well as some other prototypes from Elf's personal collection. Happy hunting!
 
  • Like
Reactions: Elf

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
I remember that glitchy serial port well :)

If you are ever in the area again, bring it by and we can troubleshoot it some!
 

About us

  • Silicon Graphics User Group (SGUG) is a community for users, developers, and admirers of Silicon Graphics (SGI) products. We aim to be a friendly hobbyist community for discussing all aspects of SGIs, including use, software development, the IRIX Operating System, and troubleshooting, as well as facilitating hardware exchange.

User Menu