Nathan's pull request#18

Open

n8xm wants to merge 69 commits into

CIS565-Fall-2013:masterfrom

n8xm:master

Commits on Oct 19, 2013

Now we have simple force calculation working.
nmarshak1337
committed
Got the basic height field working.
nmarshak1337
committed
Well, now we no longer hit any debug asserts.
nmarshak1337
committed
Now we have the naive force calculation working.
nmarshak1337
committed
Merge remote-tracking branch 'upstream/master'
nmarshak1337
committed
Now we can draw the planets as squares. Yay!
nmarshak1337
committed
Now we have our nice spacetime shader. Time to move onto using shared
nmarshak1337
committed
Performance comparison done on naiveAcc. Time to implement sharedAcc.
nmarshak1337
committed
Code is down for shared memory, and it compiles. Next step is to see if it
nmarshak1337
committed
Working with shared memory!
nmarshak1337
committed

Commits on Oct 20, 2013

Added profile. Bumped up the number of planets. Now I also have a tile
nmarshak1337
committed
Code is down for tiling, but with our small initial example it doesn't
nmarshak1337
committed
Shared works for 32 bodies (one tile)
nmarshak1337
committed
Now using shared memory works for 64 bodies
nmarshak1337
committed
Now working on 1024 bodies. Shared memory actually makes it worse with
nmarshak1337
committed
Added implicit Euler
nmarshak1337
committed
Using a cutoff radius gets us better results.
nmarshak1337
committed
Last commit before we try prefetching
nmarshak1337
committed
Loop unrolling appears to work!
nmarshak1337
committed
Now we don't have to manually adjust the tile size. Still appears to be
nmarshak1337
committed
Changed blockSize so that now we pass it in
nmarshak1337
committed
Went ahead and proved that the tile size CAN be larger than the block
nmarshak1337
committed
Now we can print out the elapsed time for the first 100 kernel executions.
nmarshak1337
committed

Commits on Oct 21, 2013

Last commit before we attempt to add collisions!
nmarshak1337
committed
Code down for collisions
nmarshak1337
committed
Almost ready to add in collisions.
nmarshak1337
committed
Added collisions. Not quite working... is the radius too small?
nmarshak1337
committed
Collisions seem to work. Radius is small
nmarshak1337
committed
It works! Penalty force seems a little funny though
nmarshak1337
committed
Elastic collisions work!
nmarshak1337
committed
Collision detection works well, with nice clumping.
nmarshak1337
committed
Going back to no collisions for nao.
nmarshak1337
committed
Last commit before adding RK4
nmarshak1337
committed
Updated README
nmarshak1337
committed
Revert "Updated README"
nmarshak1337
committed
Revert "Last commit before adding RK4"
nmarshak1337
committed
Collision detection is now back in!
nmarshak1337
committed
Now we are back to the "standard" demo. Time to add flocking!
nmarshak1337
committed
Now collisions conserve energy (or at least are better at conserving
nmarshak1337
committed
Duplicated the nbody project in order to start with flocking.
nmarshak1337
committed

Commits on Oct 22, 2013

Well, the star mass is zero. I'll figure out how to get rid of it later.
nmarshak1337
committed
Well, now we have a pointer to the device position vector
nmarshak1337
committed
Got the guy to work. It's a little buggy though.
nmarshak1337
committed
Added control law. Dunno if it works though.
nmarshak1337
committed

Commits on Oct 23, 2013

Well, now we get the ball moving independetly of the other stuff
nmarshak1337
committed
Merge branch 'master' of https://github.com/nmarshak1337/CUDA-NBody-Sim
nmarshak1337
committed
Now it actually works.
nmarshak1337
committed
Got seek working. Disabled collisions, because it causes horrible,
nmarshak1337
committed
Got arrival working, I think.
nmarshak1337
committed
Separation works!
nmarshak1337
committed
Added enum for behavior type
nmarshak1337
committed
Switching between states works. Separation spazzes a bit when agents are
nmarshak1337
committed
Prevented separation from spazzing.
nmarshak1337
committed
Leader following works
nmarshak1337
committed
Turned the planets into triangles. I just need the triangles to rotate nao
nmarshak1337
committed
Got triangles to rotate! Yippee!
nmarshak1337
committed
We've proved that we change rotation dynamically. Now it's time to
nmarshak1337
committed
Got the rotation array working.
nmarshak1337
committed
Angles are expressed as FLOATS now.
nmarshak1337
committed
Added angular velocity array
nmarshak1337
committed
Code is down for calculation of control torque.
nmarshak1337
committed
Base framework is now there for angular accelerations.
nmarshak1337
committed
Well, now we have torques. Phew!
nmarshak1337
committed
Ready to drop in numerical integration of angle/angular velocity
nmarshak1337
committed
Numerical integration of angular acceleration is there. Now to give it a
nmarshak1337
committed
Angle control works... sort of!
nmarshak1337
committed
Got angle control to work!
nmarshak1337
committed
Updated README
nmarshak1337
committed
We are back to using shared, tile size and block size are both 128
nmarshak1337
committed