Folding Forum

Right now Core 17 with GPU folding is not providing any data for Viewer (or trajectory for 3rd party API). Is there a concrete plan to implement or not really as the scientific value is less compared to the entertainment value ?

You're right, the API just returns an empty PyON array. My viewer just resorts to its demo protein for this. This is an important thing to fix, since viewers are a fairly prominent part of FAH client software and do help get newcomers into F@h.

Which client version? AFAIK it's supposed to be fixed in 7.4.0 (now in open beta), but I don't have a 8900 running on 7.4.0 yet. Maybe I'll upgrade later tonight...

David

You mean this point from change log : "Read Core 0x17 JSON visualization files." ?

But sounds like only work local; not on remote machines ... run my folding rigs mainly headless and control remote via ssh and iPad.

See viewtopic.php?f=94&t=25600&p=256618#p256618

David

Oh that's right. If FAHClient can read core 17 files, it should be able to return trajectory data via its API.

On a slightly different subject, the protein split glitch (where the protein seems broken/exploded) is another issue with the official and third-party viewers. The problem is upstream: FAHClient sometimes has issue reading protein data from the cores, and sometimes shows atoms all over the place or pieces of the protein that are far away. (The cores rely on Fourier transforms for their work, so a conversation back to a single 3D protein is necessary.) FAHClient also filters out bonds that are chemically impossibly long, which unfortunately also has the consequence that it's afterwards impossible to know which pieces hook up with each other. The bond filter needs to move from FAHClient to FAHViewer, which wouldn't have any visual effect to the end user, but would allow third-parties to hook into the API and get raw data. I've got some code that should be able to identify the glitched proteins, and with the bond information it should be fixable. But that's a different issue.

Looks like the next Starwars movie: little x-wings attacking the DeathStar



On a serious note: often it looks like H2O molecules .... Fewer times also more complex once ...

I have an 8018 WU running now on 7.4.0, & the 3-D viewer works, I see now. I know 8900 runs on Zeta, so not exactly the same thing, but I thought it worth mentioning, for the score-keepers.