Interview with Joseph Palmer
Posted: 2-Dec-2005
A couple of months back,
Cale Lewis and I (Andrew Lampert) put our heads together and came up
with some burning questions about the development of the BeBox and various
BeBox prototypes that were built along the way. Given the nature of our
questions, we headed straight to the source - Mr BeBox himself, Joseph Palmer -
to seek some answers.
Huge thanks to Cale Lewis for actually conducting the interviews and collating
the responses, and to Joseph Palmer and Guillaume Desmarets for taking time
out of their busy schedules to respond thoughtfully to our questions.
Note that I've edited some questions and answers for clarity and brevity.
Without further ado, here is the interview.
Cale Lewis: Could you give us a time frame where work on the Quad604
board took place at Be Inc. and if possible some idea of what it cost to
develop?
Joseph Palmer: That project started in mid '96, I don't have any
notes from that era (it was before blogging) so I can't really be more
specific.
Cost? I don't know -- probably 8-10 man months plus $20,000 or so for
the fabs and parts. There were three boards, the Motherboard, the CPU &
Memory daughter card and a logic analyzer probe card).
CL: Was there more than one board built?
JP: We fabed a couple of each of the boards, and populated one base and
one CPU card with two 604 processors.
I also had to rig up a regulator card for the CPU, which needed 3.3v and
2.5V back in the days when power supplies
didn't offer it.
CL:
I've read about the multi-processing
challenges Be faced with the PowerPC 603 processor.
Could you give us an idea of the major challenges you faced in designing and
building the quad-processor BeBox and was BeOS fully working on it or only
partially?
JP:
We were solving for several variables:
- It had to fit into the BeBox form factor. We didn't want to have to
re-do the case for a larger motherboard.
- We had to cool 4 processors in a very small area. We were planning to
use two 60MM fans to force air through a plenum that would surround the heat
sinks, and back panel was going be modified to use two of the IO openings as a
hot air exhaust. The idea was to get the hot air out of the box.
- Power was a real issue, I ended up placing a power connector on the CPU
daughter card because there were not enough extra pins on the main board.
- We wanted it to be low cost -- that led to the concept of using an 8 layer
board for the CPU, and a 4 layer design for the main board. All the scary high
speed stuff stayed on the CPU card, and only the relatively slower 32 bit wide
33 MHz PCI bus went down through the connector. I used the extra pins of the
64bit PCI connector for the interrupts and other odd signals.
The project was canceled in the bring-up stage, low level boot code was
running, the new interrupt
system was being debugged, but I'm sorry to report that the BeOS never
booted on the system.
CL:
Having heard of dual processor 200MHz BeBoxen existing, could you give us an
idea of how many were built and where any might be now if they still exist?
Were they considered 'revision 9' or just an upgraded revision 8 133 MHz BeBox?
JP:
Yes, a few were built, Guillaume Desmarets was the lead on that project, while
I was working on the Quad Box. It was a new board, there were extensive changes
to the power supplies to support the 200 MHz processors.
(Editor's Note: See below for more on the 200MHz BeBox
from Guillaume Desmarets)
CL:
Were there any other BeBox prototypes built or worked on such as a 6 or 8 way
BeBox board?
JP:
No, the quad box was as far as it went.
CL:
Are there any other projects planned or ideas considered at Be Inc. that our
BeBox owners here or BeOS fans at large might find interesting that you
remember from your time working there?
JP:
In the early days there was a big interest in telephony, there was a project to
build an ISA card that would act as a modem, a mini PBX and an answering
machine. That kind of fell by the wayside.
There was a lot of talk about BBS systems in the early days, the was a plan to
have a server room with a modem bank so that people could log in from their
BeBoxes to get software updates and do email, etc. It's kind of funny to think
back on that now, since at the time I think every Be employee had some basic
Internet access at home at the time. We'd talk about BBS at work, then go home
and get email, read news, use Archie and ftp... This was before the web and
HTML pages really took off.
We also had a notion of having an X-86 card that would plug in to the PCI bus,
that would make the BeBox either a PPC system running BeOS or an Intel system
for Windows. (That was back in the Windows 3.1 days.) We thought that it would
have sent the wrong message about the BeOS, and Intel closed the issue when
they obsoleted the chip set required to make it work.
CL:
We hear about the troubles the Amiga-related and Pegasos hardware developers
have had, do you think there is or in the foreseeable future will be any market
for a custom-built hardware platform for some of the niche OSes? I guess you
might say, is a 'BeBox' for Zeta or Haiku an idea that you think could work?
JP:
I think any "third platform" (AMD/Intel and Apple/PPC being the first
two) is pretty much going to be relegated to hobby status. These days there are
just so many great motherboard-CPU options out there that building your own is
akin to building your own hard drives. There's not much point.
There's always room for custom hardware: laptops, PVRs, AV systems and
such, but without a defined market and enough volume, I just don't see much
need for "yet another platform".
CL:
How about a system done with a combination of components such as a currently
produced dual CPU mainboard and other complimentary hardware integrated with
the OS into a tightly knit package such as Apple does but without the variety
of models they support. One or two finely tuned models that also might support
Linux or *BSD if need be, or do you feel it's too late/early for something like
this to have much of a chance?
JP:
I see three "platforms" that are already in that space.
The SFF (Small Form Factor) world has a lot of really nice, compact hardware. I
just put together a Shuttle system this week, there's definitely a feel of the
BeBox spirit there.
Another less developed area is in "Silent Computing". There's a lot
of work going into fans and heat pipes, and a few folks are pushing the
envelope into fully passive cooling. This really interests me because I'd like
to be able to do audio recording without all the fan noise.
The third platform is the hardware to support the serious gamers and moders.
These are the garage hot-rodders of this generation, and they've got an entire
sub-industry of people building exotic cases and cooling systems to support
their hobby. Almost every motherboard maker has an entry here - dual CPU
systems, extensive voltage and clocking control. Then there's Fans with
lights, UV reactive colored connectors, cases with windows, custom cases in
every shape, color and style.
Years ago I wrote a piece called "It's Dark In The
Box" - not any more. (I myself have not fallen for this, I've not placed any lights
in my box, but my second-hand graphics card came with blue LEDs installed at
the factory.)
As for the OS market, Linux came along about the same time as the BeOS, and I
think many of the programmers who we had hoped would develop for Be instead
found a home in the Linux camp. I'm very impressed with Linux these days, It's
actually just as easy, if not easier to install Red Hat or SUSE today as it is
to install Windows. The only thing that keeps me using Windows is the CAD
applications that I need.
CL:
Are you still working at Danger Inc. and if so, can you tell us anything about
the type of things you're working on?
JP:
I left Danger on July 8th, 2005. If you read my blog, you'll note that I never
talk about work related things until they have become public. (Sorry.) Since
then I've been taking a self-funded sabbatical to shake out the cobwebs, but
lately I'm feeling like it's time to find the next rock that needs pushing.
CL:
Since you've so graciously consented to spend time answering our questions on
the BeBox, I guess we should give you a moment or two in return. Your site has a wide range of things and
thoughts on it. From the BeBox and high performance paper airplanes to
theremins and your photography, mixed in with some very timely,
thought-provoking material on the national debt and energy policy, usage, and
where we're headed with it both here in the US and worldwide. Is there any
there anything you'd like to add here that you think we should all know or
spend more time/effort thinking about?
(Editor's Note: Although this interview with Joe was conducted quite some
weeks ago, publication was unfortunately delayed due to personal
circumstances)
JP:
First, sorry It took so long to get back to you with this. Like many Americans
I've been dazed by the news coming out of the gulf coast, and particularly New
Orleans, and it's taken these two weeks to get to the point where I felt up to
talking about the old times.
It's a little funny to dig through my old Be memories, it's been like ten years
now, and much of the fine detail has become a bit fuzzy with time. I'm really
surprised that there is still any interest in the BeBox itself when technology
has left it in the dust. Besides, the real star of the show was always the
BeOS, and I always made that point whenever I spoke about the system. For
myself, I'm very proud of the work I did on the BeBox, and it was terrifically
rewarding in the sense that I got to go out in public to meet the people
interested in the the system and do demos, both the weekly demos in the Be HQ,
and at Macworld . (I was once accused of being so enthusiastic that my demos
would scare small children. My favorite trick was to take a deep breath, cycle
power on the system, then let it out when the desktop was ready for the Demo.
I swear there were looks on the crowd like--Oh my God, that man is going to
DIE!). Perhaps it's because it was so fun, and that I got to work with
amazingly talented an interesting people, that I see the years I spent working
on the BeBox in a different, more final light.
As for one thing to think about... I struggled with this for some time. At
first my thoughts went to Hurricane Katrina, where we witnessed a failure at
every level of government to respond to an event that had been predicted for
years -- an event that had been extensively planned for -- an event that gave
days warning -- an event that was so visible, and so huge that it could have
been witnessed from the surface of the moon with the unaided eye.
But we're facing challenges greater than Hurricane Katrina, challenges that
affect the entire world. We face the imminent end of cheap oil, a crisis that
our government is not planning for, at least not in public. We also face global
climate change, a crisis that the US administration not only refuses to plan
for, but refuses to even admit exists.
You don't need to live on the path if a hurricane, or on fault line, or in the
shadow of a terrorist target to be be affected. So sometime today, stand up,
look around, and think to yourself; "It will happen -- right here".
Then think; "What can I do about it?"
Thanks for your time and your thoughts Joe!
Guillaume Desmarets on the 200MHz BeBox
Joe was kind enough to put us in contact with Guillaume Desmarets. In a series of brief emails, here's what Guillaume had to say about the 200MHz BeBox.
Guillaume Desmarets:
My 200Mz BeBox is here with me, in my office in France :-). I think there were
only 2 of them that got completed before the project was canned, and I'm afraid
I don't know what happened to the other one.
After reading Joe's comments, I don't know that there is much more to be said.
The 200MHz BeBox was a pretty natural evolution for the previous models. The
only problem was that the CPU core required an extra voltage (I don't even
recall how much, I suppose 2.5V!). We didn't have much resources to do
extensive layout modifications, so we decided to leave the CPU section
untouched and simply added a switching supply in one of the last corner of the
PC that was not used :-). That supply was pretty dense at the time, and was
pushing the limits of what was doable. But that's the extent of the
modifications.
I have not powered that BeBox for something like 2 years, as I have been
moving a lot during that timeframe. Now that I have a decent office to
put around it, maybe it's time to bring it back to life :-)
Thanks again to Guillaume and Joe for their time, and I hope you've enjoyed
this trip through the memories and nostalgia of building the BeBox.
[ Return to the News Page ]
| 
