[chimerax-users] Bottleneck Large Surfaces

[Anderson, Jacob]

Thank you for the intuition - it makes the limit make more sense.

Thanks again,

~jacob
________________________________
From: Tom Goddard <goddard at sonic.net>
Sent: Tuesday, November 24, 2020 10:35 PM
To: Anderson, Jacob <jacob_r_anderson at hms.harvard.edu>
Cc: chimerax-users at cgl.ucsf.edu <chimerax-users at cgl.ucsf.edu>
Subject: Re: [chimerax-users] Bottleneck Large Surfaces

Hi Jacob,

ChimeraX only uses multiple threads for a few calculations, such as computing solvent excluded molecular surfaces or Coulombic electrostatic potential.  Calculations like computing the clip cap surface which is spending most of its time trying to triangulate the polygonal path where the surface intersects the plane would be tricky to divide into multiple threads.  Actually it could also be spending a fraction of a second computing the intersections of say a 50 million triangles with the plane -- it does not have any fancy oct-tree decomposition to accelerate that computation so it has to check every single triangle on every move.

Tom

On Nov 24, 2020, at 7:25 PM, Anderson, Jacob <jacob_r_anderson at hms.harvard.edu<mailto:jacob_r_anderson at hms.harvard.edu>> wrote:

Dear Tom,

I did not realize the clipping cap was running on one CPU at max. Turning that off eliminated any delay. Thanks for your quick reply. I will also switch to simple light, taking your advice.

~jacob
________________________________
From: Tom Goddard <goddard at sonic.net<mailto:goddard at sonic.net>>
Sent: Tuesday, November 24, 2020 10:19 PM
To: Anderson, Jacob <jacob_r_anderson at hms.harvard.edu<mailto:jacob_r_anderson at hms.harvard.edu>>
Cc: chimerax-users at cgl.ucsf.edu<mailto:chimerax-users at cgl.ucsf.edu> <chimerax-users at cgl.ucsf.edu<mailto:chimerax-users at cgl.ucsf.edu>>
Subject: Re: [chimerax-users] Bottleneck Large Surfaces

Hi Jacob,

  I would not expect any delay rotating the surface.  Maybe you are clipping  with a near clip plane and it is constantly recomputing the clipping plane cap on every motion.  In that case you could turn off clip plane caps ("surface cap false").  Even if you do that if you are using Full or Soft lighting with near clipping then the ambient shadowing of those lighting mode will update on every motion since the clipped away part constantly changes.  That casts shadows from 64 directions basically equivalent to rendering 64 frames and will slow down even the fastest graphics card.  To remedy that switch to simple lighting (Toolbar button, or command "light simple") which does not cast shadows.  If you are not using clipping then I bet something else is updating on the CPU and you will have to provide more details of what exact ChimeraX features you are using (what kind of coloring, what other models are shown, what tools are you using, ...).

Tom


On Nov 24, 2020, at 7:05 PM, Anderson, Jacob <jacob_r_anderson at hms.harvard.edu<mailto:jacob_r_anderson at hms.harvard.edu>> wrote:

With large colored surfaces of maps ~2.4GB, I experience a lag/delay of several seconds during rotation and translation. Is it obvious where the bottleneck might be that causes the lag?  I am running ChimeraX on a 3090 gpu w/ 128gb ram, & 12 core CPU.  Might there be ways to change other graphics settings to make the manipulations have a less observable delay?

~jacob
_______________________________________________
ChimeraX-users mailing list
ChimeraX-users at cgl.ucsf.edu<mailto:ChimeraX-users at cgl.ucsf.edu>
Manage subscription:
https://www.rbvi.ucsf.edu/mailman/listinfo/chimerax-users<https://urldefense.proofpoint.com/v2/url?u=https-3A__plato.cgl.ucsf.edu_mailman_listinfo_chimerax-2Dusers&d=DwMFAg&c=WO-RGvefibhHBZq3fL85hQ&r=WAfw3qrUG1ShU2YpwE9tWl1TX9X2iVUkDLLxcqiRjEo&m=4trpeUI99BIsNu4qjm3wtfP3QglOdtMSEn0cM4gLavk&s=6tAMU78zrOpkbJcOBY7YiFPDZJEIMw0EX7ozXtQFuYM&e=>

_______________________________________________
ChimeraX-users mailing list
ChimeraX-users at cgl.ucsf.edu<mailto:ChimeraX-users at cgl.ucsf.edu>
Manage subscription:
https://www.rbvi.ucsf.edu/mailman/listinfo/chimerax-users<https://urldefense.proofpoint.com/v2/url?u=https-3A__plato.cgl.ucsf.edu_mailman_listinfo_chimerax-2Dusers&d=DwMFAg&c=WO-RGvefibhHBZq3fL85hQ&r=WAfw3qrUG1ShU2YpwE9tWl1TX9X2iVUkDLLxcqiRjEo&m=a4zx6EGVx35lmzrqwFwDFy3uhBWn9smXsiuHlgU3tcQ&s=MDWuDurZF1tp10TBGMTZPbvSJfFRZQE1_tXHyW2N99g&e=>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://plato.cgl.ucsf.edu/pipermail/chimerax-users/attachments/20201125/ef4ea0d3/attachment.html>