Comment Easier said than done! (Score 1) 391
I've been watching these guys' web site for a while, and it is great that they now have something decent to show for their work!
It's important to realise that going from a prototype to a real marketable product may take a *lot* of effort, particularly in this field. If you don't provide the highest quality and features, you'd have to compete mainly on speed. If you provide all the features and quality needed for rendering movies, you have a very complex product. It's not clear where this project is headed, but it seems to be mainly a research vehicle at the moment.
They say that Programmable shading, Parallel rendering and Global illumination are on their research agenda, so they're probably quite a way from a full-featured rendering system.
At Advanced Rendering Technology, we jumped in at the deep end! We took in over £14m of investors' cash to (try to) create a full-featured hardware ray tracing appliance. The products are still being marketed at http://www.art-render.com/, but the original company ran out of cash and has been liquidated. A brief explanation of what happened is below.
The business was created in March 1995 from the results of my Ph.D work at Cambridge University (UK). The chip, the AR250 was loosely based around the system described in my thesis "Real-time ray tracing on an advanced HDTV framestore", published in 1993 (not available online, but see patent).
The business was built on a shoestring at the start, getting early investment installments of just £50,000 ($75,000 then). With this funding, we set out detailed create proof-of-concept and a full business plan. The plan agreed was to develop and market a rendering appliance, based on:
A) A multi-million transistor graphics processor, comprising ray/object intersection pipeline, programmable shading coprocessor, and microprocessor-based control circuit.
B) Parallel processing accelerator boards with up to 64 processors
C) An embedded Linux system (we used Dec Alpha initially)
D) An efficient and robust RenderMan Shading Language Compiler
E) Rendering/driver software capable of accurate area lighting, motion blur, depth of field, volumetric effects, even for multi-million object scenes and resolution of 6k x 4k or higher.
F) A suite of application plug-in network RenderMan drivers, with support for 3D-Studio Max, Alias|Waverfront, Soft Image, as well as RenderMan direct
G) Throughput of "up to 1000x" existing software solutions on contemporary PC hardware
and once we had done all that lot, we'd shrink it to a single card and in due course move it to a real-time high-end visualisation solution.
Some "experts" said it was actually *impossible* to parallelize ray tracing in the way we had claimed! Others just said it was too ambitious.
Investors were *very* enthusiastic about the project, but demanded unrealistically short timescales with no slack. Process quality, management, testing and documentation went out of the window. They also demanded veto rights over any significant changes to the plan
To stick within the budget, we could only afford to employ one experienced chip designer (increased from the initial plan of nil(!)), and we couldn't have access to a full set of software tools either.
To stick within the schedule, we had to assume the chip would be "right first time", and would be saleable within months of receipt from the fab. The software systems would have to be built concurrently with chip design. There was no time or budget for fixing design issues with the chips. The first prototypes had to go to paying customers.
It became clear in 1996 that we needed access to an additional chip development tool not budgeted for. This led to a six month(!) impasse with one of the investors, which used its veto to conduct a technical review of the development process.
Once we had this tool, we switched fabs to one which supported embedded ARM and 0.35um process with VHDL signoff. Only it turned out they didn't really support ARM after all. In fact *none* of the fabs with ARM at 0.35um would deal with UK start-ups(!) So we designed our own 32-bit embedded microprocessor with ARM instruction set, and switched fabs a second time.
It took six months from getting the chip design compiler tool to handing over a complete netlist to the fab. This includes developing the new ARM-compatible core, and verification.
It took another six months for the foundry to make it (we went to the bottom of the pile!). The design worked first time, leaving a few months to shoehorn a huge pile of software into a box and get someone to buy it.
Of course the software wasn't ready to go (we had no Shading Language Compiler yet, nor most of the plug-ins we needed). After a lot of (controversial) re-working of the software, we got code that built, ran, and made pretty pictures. In the summer of '98, we persuaded someone to buy a unit. It only ran on (low-end) 3D-Studio Max, and still had no compiler. The investors were overjoyed! All we had to do was find a bunch of people like the first customer, and sell it to them too. We'd all be rich!
Some were ecstatic over the blinding speed - up to 120x speedup for some customers' ray tracing needs. Or about 12% of what we had promised.
We immediately doubled the number of chip designers (to two), started work on a dual-core version to boost the dazzling 40MHz clock speed of the first product. The company switched focus to trying to address potential customers' feature requests.
All they wanted was more compatibility, more features, and a 10x boost in speed.
The investors piled in more money, and the developers slogged their guts out trying to deliver whatever sales guys said would clinch each deal. But the investors were uncomfortable with missed sales targets (typically 0.1x). Obviously better management was needed.
We had made a fantastic prototype, but not a marketable product. We had done so on an amazingly tight budget in a remarkably short timescale. The investors pumped in money progressively faster (although only acting when insolvency was imminent). But the herd of customers never arrived.
At the start of 1999, it was obvious to people that
1) potential customers loved the images, loved the product concept
2) it still had to be *much* faster to compete with scanline/Z-buffer
3) investors would be mad to fund a third generation chip before seeing sales success
4) the whole technology was too complex for a tiny start-up
5) "clever-yet-brittle" constructs had to be replaced or tamed through architectural enhancements
And it was also obvious to some that a sales-driven focus would not solve the key problems, and would lead to failure.
The architectural enhancements were well mapped out by mid '99, and I felt initial prototypes were very promising. But solving this technical problem wouldn't be sufficient, of itself, for success. We had to solve the company's structural and funding issues too.
The executive management went into overdrive to evaluate alternative approaches to resolving the key problems as a whole.
By this time, I had an outline of the third generation chip architecture. This was to be based on multiple (perhaps 8) cores, each with 4-way multithreading. The architecture was to be stripped down, with the minimum of baggage. If we did it today, people would say we had ripped off Sun's Niagra!
The executive management concluded that the *only* sensible way forward was to develop the third generation chip in a separate company, marketed for a much broader range of applications, particularly as a network processing chip (which was in its infancy in '99). Advanced Rendering Technology would enhance its focus on the rendering architecture and appliance, and would have a much reduced costs and risks from dispensing with the chip development. The broad opportunity of the multi-core SMT chip was too great to discard, but too risky to fund within ART.
The spinout plan was (to say the least) controversial with the board. In fact the board wasn't even empowered to adopt the plan, and it came down to whether investors would authorise it. The non-executive directors felt that the efforts to boost sales should be redoubled, and anything else wasted precious time and money. Investors certainly wouldn't back anything radical without (near) unanimous board support. The result was a five month impasse.
Ultimately, the situation had boiled down to two possibilities
1) carry on with existing plan (supported by key investors) or
2) spinning out the chip, cutting costs and focusing on graphics software (supported by key management)
At the end of the impasse, the company had no money left, and new investment became contingent on executives backing the existing plan. Variations on the spinout plan discussed led to investors to get worried they were to be ripped off(!)
I felt it would breach my duty to the company to back new investment based on a plan which was near certain to fail. So the investors made my removal another condition of investment. I was pushed out of the company in November 1999, and new investment arrived shortly after.
The company immediately dropped the multi-core SMT technology(!), and structural enhancements of the software were abandoned for fixing up existing issues. I was excommunicated (sent to Coventry as we say in Britain).
A new third generation ray tracing chip was embarked upon, based on a sophisticated configurable streaming architecture. The chip design team was boosted to three.
Another £5m was poured into the company over about 2.5 years, but investor stamina ran out in early 2002. The company went into administration, and has since been liquidated. Nothing has been returned to Investors.
The rendering appliance business, however, was bought out of administration by a team led by Dr. Daniel Hall, the other ART founder. The new company, ARTVPS is still selling the second generation chip (AR350), which was designed back around 1998/1999. Their web site proclaims great quality and performance, but it is unclear what exciting new products are in the pipeline.
The conclusions?
* The goals were achievable, but the irrational appetite for risk made success rather unlikely
* Only take money from investors who want to back the people!
* Manage expectations!
* Investor vetoes on board decisions damage the heart of governance
* Don't compromise on critical issues!
Adrian
--
Sincerest thanks to all who helped this project, staff, customers, investors, suppliers
drop me an email if you want more info!
amtw at linuxchip dot demon dot co dot uk
PS: I'm game for any interesting new project in this field, especially if it involves SMT, concurrency or graphics