Questions and Answers
Carsten Haitzler (The Rasterman)
raster at rasterman.com
Sat Jan 3 01:11:21 CET 2009
On Sat, 03 Jan 2009 00:50:14 +0100 Lee Grime <lee.grime at gmail.com> babbled:
or... use something programmable like the omap3 series that comes with a dsp to
offload work to - arm has been working on doing cortex a9 which is multi-core.
forget glamo. it's legacy. if you are talking of "doing an SoC" you are in for
a hard time. i just did some benchmarks of tro650 vs gta02 vs beagleboard vs
desktop in software rendering:
as you can see - focusing on glamo and fixing it is not goign to net you
anything. just using a "modern SoC" netted an average gain of about 4-5x the
speed. and it's not even at its rated 600mhz and with slower ram. i'd expect a
good 15-30% up from that just for a production version (ie 600mhz and decent
speed ram). and thats not including any neon optimisations. i have that on my
todo list now i have silicon to play with.
you can go into details of the glamo if you want.. but it'll just rehash
things. the future for graphics is not discrete chips doing it - but cores on
your SoC. the 3530 has a fast cpu core - fast memory bust with even an l2 cache
- AND it even has a (closed) 3d core (SGX). the cpu alone is fast enough to do
nice gfx on at the resolutions you'd expect on a phone - throw in the dsp and
you're sailing along. producing your own SoC is likely going to net you
something that maybe in 5 years approximates the SoC's of 5 years ago in power
consumption and performance. if you want a solution and want it in a
reasonable time - simply a better SoC is the go. (what you want exists already
in production - just a matter of using it :))
> I come from a hardware background, chip design mainly, but analogue
> (note the spelling :-) ) and DSP(MSc) are still strong points. Done
> chip design for 15 years. Now I do not have a great deal of time at the
> moment, what with a 3 week old baby and stuff!, but if we can get a few
> other like minded people together, I am sure we can produce our own open
> source SoC. And if the credit crunch kicks in properly, even more time.
> I have heard all about the problems with the crappy Glamo chip. Why not
> have a small CPLD as a co-processor, into which we load a 'codec' for
> whatever we are doing at the time, say mp3 decoding, or some video
> codec. We can get cheap and low power enough CPLD's or FPGA's these
> days to perform this job.
> Lets make this thing really open. Could even do the GSM part open
> source. No more problems with NDA's etc. If you can get to 100K units
> our own ASIC should become viable.
> What do you think?
> On Sat, 2009-01-03 at 07:11 +0800, Sean Moss-Pultz wrote:
> > Dear Community
> > Here are my answers to your great questions:
> > > Q1. The end of the year is a time to look back on the year
> > > achievements. So where does Openmoko stands now from a business point
> > > of view ? Could you comment on sustainability, on sales numbers, on
> > > geographic markets and customer categories ?
> > Let me begin by addressing the market aspects of the question, because I
> > find this part more fundamental. Openmoko was built from the tools and
> > knowledge of the Internet. Our argument for the necessity of an open
> > phone stems from an observation that the Internet breaks down mass
> > markets by making it economically attractive for companies to address
> > niche market segments. We saw a real business opportunity caused by this
> > divergence. So we set out to build a product capable of allowing a
> > company to reach out to its customers and allowing its customers to talk
> > back and to talk to each other. This product was the Neo 1973.
> > We pioneered a new breed of Consumer Electronics companies. The products
> > we build expand our community. The expanded community makes for better
> > products and more sales. Ad infinitum. Perhaps one day people will look
> > back and call such corporate and community teamwork the first, of many,
> > Social Electronics companies.
> > FreeRunner, our major product milestone of 2008, started selling in
> > July. We were more conservative this time with our launch. There were no
> > pre-orders. We sold first through distributors. Our own order processing
> > was entirely automated. Last time it was (painfully) manual. I hope
> > everyone who reads this list will appreciate how much effort FreeRunner
> > took. Openmoko is a very small company. We succeeded in building a
> > smartphone only because we didn't have enough experienced people to tell
> > us that it couldn't be done.
> > While far better than the Neo 1973, FreeRunner had its share of
> > problems. The ancient TI Calypso had registration and SIM card
> > compatibility issues. Audio quality, while slowly improving, still is
> > not perfect. But I can accept these growing pains. We are improving. We
> > are much better than our last product. And community effects continue to
> > materialize in mind-blowing ways. (I will never forget the speed at
> > which we all fixed the GPS issue.)
> > "Back to the Basics" was our response to your public and private
> > comments. We continue to refocus internal efforts around these ideas.
> > Paroli represents a phone application with a feature set reduced to the
> > bare minimum that is still useful. FSO is our base that will let you
> > easily build what you want. The entire system is becoming more stable as
> > we increase our efforts to have our kernel downloadable from kernel.org.
> > Our technology plan is stabilized at this point. Development priorities
> > for the next six months are clear.
> > Openmoko's goal can be written in one sentence: We want to build
> > consumer products that package the best parts of the FOSS world into
> > products that are relevant and inspiring to ordinary people.
> > "Inspiring" is the key part of this goal in my mind. Here's a list, off
> > the top of my head, of the things we've inspired to date:
> > * A small project inside of FIC to become an independent company
> > * 24 distributors around the world to join us in getting our products
> > to more people
> > * The development of entire distributions (as opposed to just
> > applications)
> > * Industrial designers to remix our CAD files
> > * A very interesting Framework initiative
> > * A documentation list with the most amazing Community Update emails
> > Using inspiration as a metric, without a doubt, Openmoko was a massive
> > success in 2008. If you want to compare us to Apple and their iPhone I
> > would be flattered, but I also think it's a strange comparison. We are
> > very a small company. Sales are just enough (around 10,000 to date) to
> > survive. Apple has been around for 30 years - Openmoko just under two.
> > I'm extremely excited about our position going into 2009. At the same
> > time, I'm very realistic about the road we are traveling. Success will
> > take everything we've got.
> > > Q2. The god of January has two faces, one looking back and one looking
> > > forward, so... Juergen, Eric in Japan and practically everybody else
> > > want to know when will the GTA03 phone be released, if it will have
> > > 3G/3.5G, a camera and the kitchen sink. Can you tell us anything yet?
> > > Where do you want Openmoko to be in December 2009?
> > We sold the FreeRunner based on "Openness" alone in 2008. We had no real
> > competition in the marketplace. We were very lucky. But times are
> > changing. Go ask your non-developer friends what's the most open phone
> > in the world. They'll say the iPhone or G1. At least that's what I'm
> > hearing.
> > Nobody will doubt the value of openness for the mobile industry anymore.
> > This seems like good news at first glance. But what openness are they
> > talking about? Look around and you'll find it's pretty different than
> > what we've been talking about. Yes, the very definition of openness is
> > changing. This troubles me because we cannot influence markets with our
> > words — only our products. And the quality of our products is not world
> > class yet. The bar has been seriously raised. Time is running out. We
> > need to find a way to lead again. I don't believe playing catch up will
> > work. Something fundamental needs to change.
> > I've been thinking a lot about art this past year and how it relates to
> > what we do. Specifically, I've been extremely interested in categories
> > and perception. The difference between new art and copies is in the
> > ability to show contrast. The difference between new art and a mistake
> > is in perception. A change in perception can only happen when people no
> > longer think in terms of the same categories.
> > One of my favorite examples comes from the 20th century artist Vladimir
> > Malevich. Why would somebody, leading a comfortable career following the
> > latest trends, come up with something like Suprematism out of the blue?
> > What changed in mind of Malevich to go from painting vibrant fluid
> > landscapes to a white square on a white canvas? I think I understand now.
> > As an entrepreneur, I know that our best work is done when our resources
> > are limited because there is no other choice. Necessity is, after all,
> > the mother of all invention. Openmoko needs our best work to thrive in
> > 2009. Our goal will be to generate more contrast, enough to change
> > people's perception, enough to stay visible.
> > This is all I want to say about GTA03 at this point.
> > > Q3. The mandatory dogfood question was submitted by Yorik Moko:
> > > "Do you use the FreeRunner as your daily phone, which distro
> > > and what is your favorite application ?"
> > I use two phones daily. My FreeRunner boots (now) Om 2008.12. The
> > last 3 months before that I've been using Om 2008 TESTING. My second
> > phone is a Blackberry Curve, used not out of love but out of a desperate
> > form of addiction to the digital world. Sometimes I wake up in the
> > middle of the night just to check email.
> > This past year, I've had too much email and too much technology. More
> > and more I find myself looking for natural, handmade things. At the same
> > time, though, I'm definitely not willing to give up my technology. I
> > want both a phone that is digital, but in a strange way also analog. I
> > hope to build one next year. Let's see if we can get there!
> > > Q4. Todd would like to know exactly what are your feelings about
> > > Android and its impact on the Openmoko OS.
> > I would like to clarify one thing first: We don't see Android as the
> > competition. This might surprise you. But it shouldn't. Would a painter
> > (I would ask) view new paint, a blank canvas, or paintbrushes as the
> > competition? Surely they would see them as tools, as raw materials,
> > waiting for their imagination to transform them into something real.
> > Android is a software stack that is both complex and high quality. But
> > software is just one small piece of what a person interacts with when
> > making a phone call. My fascination with the mobile phone, at least to a
> > large extent, stems from the fact that it runs complex software that
> > most people don't know or care about. Please make no mistake about the
> > importance of this last statement: ordinary people don't care what
> > software their phone uses. You might think this trivializes what Google
> > did. That is not my intention. I have nothing but the greatest respect
> > for what Andy Rubin's team pulled off - that was an engineering
> > masterpiece. But I see Openmoko as a product company. First and foremost
> > we must concern ourselves with impacting the lives of ordinary people.
> > Praising (for too long) a software stack would be placing technology in
> > a far more powerful position than (I believe) it belongs.
> > > Q5. Kosa and Marco wondered if you could say us something about the
> > > management that doesn't seem to be loved by great hackers like Harald
> > > and Carsten. For example, what is your analysis of the controversies
> > > that led Om2008 to ship with Qtopia's predictive keyboard.
> > I am puzzled each time I hear people complain about "management" in
> > Openmoko. Management (myself obviously included) goes out of our way to
> > provide the best environment we can. People, like Harald and Raster, are
> > given immense freedom to work on what they feel is most important.
> > Management is responsible for the results of their efforts. So if we
> > make changes that are unpopular, it's only because the realities of
> > shipping a working product under a limited amount of resources (time,
> > money, etc...) became too great. It's not because we don't believe in
> > them or their abilities.
> > Let me give everyone a bit more background into the keypad issue. We
> > first saw the Qtopia predictive keypad back in February of 2008, and
> > became extremely exited. This keypad, we believed, had the potential to
> > become better than anything on the market.
> > We asked Raster to integrate this keypad into Om 2008 and extend it to
> > make it more hacker friendly (i.e., usable from places like the
> > terminal). After two months of more or less silence he showed us his own
> > version, written from scratch. The design was a work in progress. And
> > the dictionary was far inferior to what Qtopia had already. An internal
> > battle started that lasted until one month before Om 2008 was set to be
> > released when our product manager, Will Lai, couldn't take it anymore.
> > He asked another engineer to just get the Qtopia keypad working.
> > At that point Raster's keypad was getting stable. It had many new
> > features. But basic text entry was still not as good as Qtopia's. Major
> > parts of Om 2008, in the meantime, were still not finished (like the
> > Glamo or network manager).
> > Openmoko (the company) needs to focus on simplifying. We need to limit
> > ourselves to building what doesn't already exist. We cannot constantly
> > try to build better components from scratch. Our resources are just too
> > limited for that. Openmoko is trying to repackage the essentials (just
> > enough) to make people feel inspired. What's not there is often times
> > more inspiring than what is there.
> > I emailed Raster, the other day, asking if my current perspective
> > corresponds with his. The main motivation for writing a new keypad from
> > scratch, he said, had to do with his ability to (easily) extend Qtopia's
> > code. C++ and qt were not familiar to him. And he wanted something with
> > more configuration options. To get there with Qtopia, he thought, would
> > take more time then writing a new one from scratch.
> > All I ask is that we please don't continue this debate. We have both
> > now. FSO is using Illume's keypad, so future Om releases will most
> > likely do the same.
> > Raster's keyboard, without a doubt, inspired many people in our
> > community. We've arrived at a point we can look at the past and find out
> > what we have experienced and what meaning it has. I'll leave the
> > conclusions up to you.
> > > Q6. There are many Openmoko communities in languages other than
> > > English now. Swap38 asks how can they best contribute back to the
> > > project, and if there are plans to organize the diversity around a
> > > common model, as OpenOffice.org did for example. What is the status of
> > > the internationalization and localization effort ?
> > To the best of my knowledge, there have only been a few coordinated
> > localization efforts on our lists. I would love to see more.
> > Localization is very interesting to me since I believe FOSS provides a
> > unique opportunity to think locally, but act globally.
> > Probably the best way forward is for people to volunteer on the
> > community list. Key documentation, language strings, howtos, all could
> > use translations. If there is anything Openmoko can do to better
> > organize these efforts, please let us know. We really do need a lot of help.
> > > Q7. Let's be reflexive, kind of "we have to talk about our
> > > relationship, dear". Tilman, supported by Michelle, wonders to which
> > > extend the community has met your expectations. More precisely, could
> > > you name an area where it did not, perhaps like an application that
> > > you would have expected be written already ?
> > The first part of this question is very hard for me to answer. I don't
> > think it's possible to talk about a community meeting the expectations
> > of an individual. I only wanted to inspire people, not demand results
> > from them.
> > From day one, I believed that Openmoko needed to deconstruct the
> > smartphone to its bare essentials, and then build up something new that
> > properly embraced diversity (the more human face of complexity). Along
> > this journey we were sidetracked into reinventing the smartphone many
> > times. I'm not sure why remaking what already exists emits such a strong
> > gravitational pull - but it does.
> > Around the end of 2007, I had an "out of body" moment when I realized
> > where the path we were going was leading us. In short, we realized that
> > if we ship with a reduced set of smartphone applications then we show
> > the past, and promise the future. Everyone will compare what is missing.
> > And we can only say, we are open; all that standard stuff you expect in
> > a phone will come. In short, we've turned the future into the past.
> > If we launch with just a single future phone application then we deliver
> > the future, and forget the past. When people ask where a past feature
> > is, "Hey where are contacts?" we'd say, "Contacts are in the past". The
> > phone of the future will manage this for you, even tell you before you
> > call if your contact is online, offline, around the corner, or only
> > taking voicemail. We must move beyond the phone. Support for legacy
> > stuff will come.
> > We tried to refocus the company around these ideas. This led to an
> > application called Diversity. The basic idea is the following:
> > Neos talk to other Neos using a self-creating, self-healing, global
> > free (WiFi) network. The software system, code named Diversity,
> > consists of many clients (Neos) talking to servers and, at a later
> > time, self-connecting, using mesh-like interactions.
> > When powered on, Neos continuously expand the construction of the
> > free network by automatically searching out free WiFi spots and
> > adding their locations to a database for other Neos to use. A simple
> > reward system based on visual / auditory feedback will let people
> > know that a new spot was discovered.
> > Communication is done on the free (WiFi) network with only one
> > exception: When a Neo cannot find a free network to connect to, it
> > will dial into the non-free (GSM) network for only the time required
> > to download the nearest free WiFi spot locations. Using GPS, the Neo
> > is then able to navigate to that position and communicate for free.
> > Communication is simply a layer on top of the free network. It can be
> > auditory, visual, or textual.
> > "Free your Phone" was our marketing focus up to that point. But we
> > realized this was not enough. Even when the phone is completely opened,
> > it's still not free. There's another force keeping it's potential locked
> > up. This is the carrier's network. They limit what can be plugged in and
> > what the plugged in devices can do. So we thought we needed to "Free the
> > Network".
> > To make a long story short, execution got in the way of ideas (something
> > a bit too familiar in Openmoko). Parts of the project went into
> > Locations. But the full idea of constructing a parallel, free
> > communication network never materialized. Maybe people had similar ideas
> > (I was reading about kismet on our list a while back) but I think we
> > were in the best position to make it happen from a product standpoint.
> > One day we'll come back to these goals, but not until we ship a robust,
> > well-designed phone. Until then, we'll stay focused on getting "Back to
> > the basics".
> > If a serious community effort would be established around these goals,
> > there's a very good chance Openmoko would design a phone specifically
> > for this purpose. I still believe this device could become the icon of
> > openness.
> > > Q8. On the other hand, Tim and others would like to know the most
> > > interesting or fun thing you have seen the Freerunner used for ?
> > Guillermo Sureda-Burgos' work (and essay) on open industrial design
> > ranks high on my list. So do Numpty Physics and Duke Nukem. But I think
> > the most surprising thing (as in we didn't see it coming) was the
> > explosion of new distributions: FDOM, SHR, even Debian. Being able to
> > swap out entire software stacks with the change of an SD card gives a
> > whole new meaning to the word customization.
> > > Q9. And where do you think could any person do the most for the
> > > project right now? Think of if that person would be ideal and have all
> > > the skills needed. And then, which skills are in most need right now
> > > and in the future?
> > There are so many ways people can contribute that it's hard for me to
> > single out a specific task or accomplishment. Development, testing, bug
> > fixing, organizational, and documentation tasks are all important.
> > Perhaps the best way to figure something out in this regard is for
> > people to post their qualifications and interests to the community list
> > and we all can provide direction based on that.
> > > Q10. To conclude, can we have a video of you in Openmoko T-shirt and
> > > geek-cap, having a phone chat with star hackers, using Neo FreeRunners
> > > at both ends, telling each other your favorite FreerRunner joke?
> > > Please?
> > I'm in San Francisco now, without a video camera. So this one will have
> > to wait until I return to Taiwan. When we're done filming, I'll
> > post to the community list.
> > Hopefully you all got this far. I know this email is long. It was quite
> > an experience for me to write so much. (Minh you did an amazing job
> > compiling these questions!) We've come a long way together. I'm
> > really looking forward to 2009. We have great opportunities all over the
> > place. So what matters is that we continue to be hopeful for the future.
> > And, together, we take the bold steps needed to get there.
> > Happy New Year!
> > Sincerely,
> > Sean Moss-Pultz
> > _______________________________________________
> > Openmoko community mailing list
> > community at lists.openmoko.org
> > http://lists.openmoko.org/mailman/listinfo/community
> Openmoko community mailing list
> community at lists.openmoko.org
------------- Codito, ergo sum - "I code, therefore I am" --------------
The Rasterman (Carsten Haitzler) raster at rasterman.com
More information about the community