Page 1 of 1
Why seams the folding of the protein be so slow nowadays?
Posted: Sat May 02, 2020 4:54 am
by Simplex0
Why seams the folding of the protein be so slow nowadays?
I remember a few years back when you bring up the Viewer and could se how the protein folded really fast but when I take a look at the protein folding in the Viewer now the protein is almost static.
When I look at the protein in the Viewer I might se a minor folding every 10 seconds or so, very slow.
Also, when I stop the rotation the folding also seams to stop. That cant be right?
Often when I try to use the Viewer while working on a really big protein like more than 200000 atoms I can not view the protein at all. Why is that?
Re: Why seams the folding of the protein be so slow nowadays
Posted: Sat May 02, 2020 4:33 pm
by MeeLee
I don't know if anyone is maintaining the Viewer.
But I remember distinctly, that 20 years ago, folding on CPU was just a few atoms.
Now it's hundreds, if not thousands of atoms in one WU.
Re: Why seams the folding of the protein be so slow nowadays
Posted: Sat May 02, 2020 6:53 pm
by PantherX
The FAHViewer in V7.6.9 has had a bit of TLC so it does work as expected in most cases. What version of the client are you using?
Re: Why seams the folding of the protein be so slow nowadays
Posted: Sat May 02, 2020 7:14 pm
by JimboPalmer
In the last 5 weeks, the F@H servers have been under intense load. One way of reducing that load would be to increase the length of time each WU was folding for. And I have noticed that my GPUs are getting long Work Unit run times. I think there is a limit to how big they can make a WU, I see some errors that the return file is too large in some cases. So my PCs are still doing the same PPD, but with fewer WUs.
This way they attract less server attention.
This is Win-Win unless you are annoyed your stats are 'lumpier'.
Re: Why seams the folding of the protein be so slow nowadays
Posted: Sat May 02, 2020 9:00 pm
by Simplex0
Thank you all.
I'm using version 7.6.9.
"But I remember distinctly, that 20 years ago, folding on CPU was just a few atoms.
Now it's hundreds, if not thousands of atoms in one WU."
Yes MeeLee that might bee the reason that the folding seams so slow nowadays, some of the proteins I run nowadays have more than 400000 atoms.
But when I press space bar to halt the rotation it should not halt the folding of the protein but that seams to be the case.
Re: Why seams the folding of the protein be so slow nowadays
Posted: Sat May 02, 2020 9:05 pm
by PantherX
Simplex0 wrote:...when I press space bar to halt the rotation it should not halt the folding of the protein but that seams to be the case.
Weird... I am running V7.6.9 and on the GPU Slot, I was able to pause the FAHViewer and my GPU Slot kept on folding without issues. Does it say "pausing" or "unpausing" in the log file?
Re: Why seams the folding of the protein be so slow nowadays
Posted: Sat May 02, 2020 9:13 pm
by bruce
It's not quite that simple. First of all, the complexity goes up faster than linearly. If you double the number of atoms, there are 4 times as many inter-atomic forces to be calculated. Second, many of the COVID projects involve a mostly fixed feature of the virus to be tested against a small segment of a potential antibody. The suggested vaccine component looks around for a suitable binding site while the virus segment remains in it's folded state.
Re: Why seams the folding of the protein be so slow nowadays
Posted: Sun May 03, 2020 5:51 am
by TPL
Simplex0 wrote:But when I press space bar to halt the rotation it should not halt the folding of the protein but that seams to be the case.
A tip: You can adjust rotating speed by arrow keys on both axes, X an Y. Slow both down to zero and you have stopped rotation.
Re: Why seams the folding of the protein be so slow nowadays
Posted: Sun May 03, 2020 9:26 am
by Jorgeminator
I'm sure he's not talking about folding, but about rendering. Simplex0, the viewer does not show folding in real-time, it renders the checkpoints (viewer frames) and interpolates between them to make it look smoother. The proteins have gotten larger but the viewer is still as efficient/inefficient as it was years ago, so the rendering capability is limited. If you throw hundreds of thousands of atoms on it then it will inevitably become a slideshow.
Simplex0 wrote:But when I press space bar to halt the rotation it should not halt the folding of the protein but that seams to be the case.
This confirms what I said above. Pressing space halts the *rendering* of the protein. It is still happily being folded by the client.