Balmer, please note that there are no final decisions yet about whether the next phase of work with Dr. Aspuru-Guzik's team will be GPU enabled.

While we would love to have a new GPU project on World Community Grid, that is not the primary factor for determining which research to run. Instead, we first ask the scientists to pick the research application most valuable to their work. Then, we assess whether it is GPU enabled and whether those GPU capabilities are beneficial to the researchers.

While we are considering a GPU-enabled research application in our next phase of work with Harvard, those discussions are still ongoing. Therefore, no final decision has yet been made as to whether the next phase of that work will be GPU-enabled.

Don't mean to dampen your spirits, just didn't want to raise your hopes up just yet :)

adamradocz, the researchers ultimately select which research application they want to run on World Community Grid. We take on the technical work of incorporating that app into the World Community Grid platform (e.g. adding checkpointing capabilities, addressing security loopholes, incorporating it into BOINC's data input and output mechanisms, etc).

Sometimes, those apps already run on GPU, but in many cases, they don't. If the app was developed by the research team itself, they weigh the effort of developing a GPU version against the expected acceleration of running on GPU's (the GPU capabilities may not be applicable to the entire app, and therefore may only be speeding up a small portion of an entire calculation). i.e. it is up to the app developers to modify it to run on GPU, and up to us to then integrate that version into the Grid.

In other cases, the scientists may select a research app that another project has already run on World Community Grid but where the GPU version is newer than the version that's already on the Grid. In those cases, we evaluate with the researchers whether the expected acceleration would justify the time it would take to onboard the newer, GPU-enabled version.