Before presenting the roadmap, some info on the current status:

The project status is updated as of January 2008...

• Base GPU Network: up and running since 2002 with 5-15 nodes online in average
• Search Engine: up and running since 2005, not reachable since July 2007
• Terragen Movie Generation: up and running, anonymous ftp to store computed images available thanks to LWM-Media. A part of Titan, most movies were computed in 2005. A feasibility study for Terragen 2 was done.
• Seismograph: connected only once to GPU in 2006, results were seen in Whiteboard
• File Distributor: up and running since 2007, it is close to production stage.
• Mutella Linux client: recompiled with changes, but still not useable for GPU.
• Orbit Reconstruction: computation of orbits is fast and accurate, but initial position with orbital elements is close to JPL solution for T=Tp, but not for other periods.
• A completely new paradigm, Persistent Database on WAN, is released with 0.942.
• Eternity II plugin released with modifiable milestones.

Does a Roadmap exist?

As the interested reader might imagine after reading the introduction and some futuristic thoughts, it is difficult to setup a "roadmap" for this project. It is better to think as the project goal in term of reaching the pinnacle of a mountain which was never violated before. So there is no road, but only a tiny path defined by the rocks. There is no map because the map still has to be written. And there is the blue sky over there... which suddenly can transform into a blizzard, where everyone is lost.

Keep in mind that even if the project is up since 2002, it only left the base camp for the first journey...

One of the three rope columns which compose the project

Instead of having a roadmap, the project is divided into three rope columns with a direction and current open targets, where the column should put nails and support camps. Though the three lines follow different paths and act independently, they have to keep in touch and try to follow the schedule.

First rope: Resistance

The direction for the first column is to resist with the current equipment. Sherpas operate GPU nodes, ftp servers, websites, file distributors, standalone crawlers and other devices created by the project. They upgrade their router firmware because of increased traffic and fight against bad documentation full of mistakes. Doing so, the first line will spread the word and compute interesting movies.

• Operate normal GPU nodes
• Operate entry nodes

Second rope: Freedom

The second column tries to free source code from proprietary extensions, and to compile it on a free platform like Lazarus, so that GPU can run on any operating system.

The second line tries to improve the source, to free GPU from scalability limits. The second line understands the Distributed Search Engine as a vehicle to spread information which is not offensive, but of a certain quality.

Sherpas of the second line work also on an Open Source license, which prohibits military use.

• Compile Mutella clients compatible with GPU nodes which can run on Linux
• Improve the proposed Pacifistic Public License
• Provide a new GPU core without GUI
• Work on Open tasks on Sourceforge

Third rope: Discovery

Sherpas of the third column live shortly but intensely. They die in impossible missions. From their failures, everyone can learn where pitfalls lie. The third line also keeps a list of ideas, which should be implemented soon or later.

• Build new interesting computations into the framework (e.g. do not check for the obvious climate warming, but propose and implement calculations to reduce it)
• GPU as an agent based framework (PDF, chapter 6.2)
• GPU as sensor network (earthquake sensors, weather stations,...)
• GPU on the Persistent WAN db layer
• Rewrite entire code!

The One Laptop Per Child Project became reality. Will it be the same for GPU?

Feedback on this document

This document needs revision. Please contact the mailing list for feedback on this document. Thanks a lot!