search

openworm / OpenWorm

Repository for the main Dockerfile with the OpenWorm software stack and project-wide issues
http://openworm.org
MIT License
2.69k stars 212 forks source link

Problem: Simulation with elastic body on GPU #59

Closed skhayrulin closed 11 years ago

skhayrulin commented 11 years ago

Simulation with elastic body doesn't run correctly on GPU (Tesla C1060) Problem With elastic body on GPU. It runs unstable meanwhile on CPU it runs correctly. I'll try to compare two simulations on CPU and on GPU

skhayrulin commented 11 years ago

The places with potentialy bugs. First of all, line https://github.com/openworm/Smoothed-Particle-Hydrodynamics/blob/master/src/sphFluid.cl#L928 potentialy we can write to acceleration[ id ] simultaneously many time. Secondly, lines https://github.com/openworm/Smoothed-Particle-Hydrodynamics/blob/master/src/sphFluid.cl#L948-L949 the same problem with sortedVelocity[id] and sortedVelocity[jd].

skhayrulin commented 11 years ago

First movement of worm)) (accidental movement), result of bug fixing. It's not work yet, but looks very funy) crawling_worm

gidili commented 11 years ago

Very funny indeed :)

slarson commented 11 years ago

Woah! Is there a movie of this somewhere?

On Thu, Feb 28, 2013 at 12:49 PM, Giovanni Idili notifications@github.comwrote:

Very funny indeed :)

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/59#issuecomment-14231678 .

skhayrulin commented 11 years ago

No @slarson unfortunately it's just a bug but it makes elastic worm similar to real worm). Sorry maybe my comment was't clear sorry

skhayrulin commented 11 years ago

Hi All,

I think we can close this issue. This bug is fixed. What was changed: 1) struct elastic connections: jd -> id of j elasic particle rij0-> initial distance between two elastic particles val1-> unused property in future I plan we can use it for storage charge of spring val2 ->unused property in future I plan we can use it for storage id of body to which ths
connection is belong. 2) Function compute_elasticForces was completely changed.

a-palyanov commented 11 years ago

Sergey, this is brilliant! Significant amount of work is done!

gidili commented 11 years ago

Nicely done @skhayrulin!

vellamike commented 11 years ago

Great work Sergey, I'm looking forward to diving into this code when I get a minute!

On 4 March 2013 12:35, Giovanni Idili notifications@github.com wrote:

Nicely done @skhayrulin https://github.com/skhayrulin!

— Reply to this email directly or view it on GitHubhttps://github.com/openworm/OpenWorm/issues/59#issuecomment-14378693 .

slarson commented 11 years ago

Bravo!