Code

Nathan's pull request#18

Code

Open

n8xm wants to merge 69 commits intoCIS565-Fall-2013:masterCIS565-Fall-2013/Project3-Simulation:masterfrom n8xm:masterCopy head branch name to clipboard

+78,049-521

Commits

Commits on Oct 19, 2013

• Now we have simple force calculation working.

  

  nmarshak1337

  committed
  ba1310f

• Got the basic height field working.

  

  nmarshak1337

  committed
  8e8b438

• Well, now we no longer hit any debug asserts.

  

  nmarshak1337

  committed
  b38c305

• Now we have the naive force calculation working.

  

  nmarshak1337

  committed
  aac4c75

• Merge remote-tracking branch 'upstream/master'

  

  nmarshak1337

  committed
  1563889

• Now we can draw the planets as squares. Yay!

  

  nmarshak1337

  committed
  0cb976d

• Now we have our nice spacetime shader. Time to move onto using shared

  Show description for 3ab774e
  

  nmarshak1337

  committed
  3ab774e

• Performance comparison done on naiveAcc. Time to implement sharedAcc.

  

  nmarshak1337

  committed
  0657af0

• Code is down for shared memory, and it compiles. Next step is to see if it

  Show description for 753c509
  

  nmarshak1337

  committed
  753c509

• Working with shared memory!

  

  nmarshak1337

  committed
  f41cbcb

Commits on Oct 20, 2013

• Added profile. Bumped up the number of planets. Now I also have a tile

  Show description for 12b7d7a
  

  nmarshak1337

  committed
  12b7d7a

• Code is down for tiling, but with our small initial example it doesn't

  Show description for 9236bd4
  

  nmarshak1337

  committed
  9236bd4

• Shared works for 32 bodies (one tile)

  

  nmarshak1337

  committed
  2837e89

• Now using shared memory works for 64 bodies

  

  nmarshak1337

  committed
  da3f054

• Now working on 1024 bodies. Shared memory actually makes it worse with

  Show description for 452bcaf
  

  nmarshak1337

  committed
  452bcaf

• Added implicit Euler

  

  nmarshak1337

  committed
  8227fca

• Using a cutoff radius gets us better results.

  

  nmarshak1337

  committed
  9879c9d

• Last commit before we try prefetching

  

  nmarshak1337

  committed
  9ea2448

• Loop unrolling appears to work!

  

  nmarshak1337

  committed
  1bc38a5

• Now we don't have to manually adjust the tile size. Still appears to be

  Show description for 7406d23
  

  nmarshak1337

  committed
  7406d23

• Changed blockSize so that now we pass it in

  

  nmarshak1337

  committed
  807ec3d

• Went ahead and proved that the tile size CAN be larger than the block

  Show description for 0217fef
  

  nmarshak1337

  committed
  0217fef

• Now we can print out the elapsed time for the first 100 kernel executions.

  

  nmarshak1337

  committed
  32584c9

Commits on Oct 21, 2013

• Last commit before we attempt to add collisions!

  

  nmarshak1337

  committed
  7a315e3

• Code down for collisions

  

  nmarshak1337

  committed
  1da2776

• Almost ready to add in collisions.

  

  nmarshak1337

  committed
  1530f48

• Added collisions. Not quite working... is the radius too small?

  

  nmarshak1337

  committed
  5a1feee

• Collisions seem to work. Radius is small

  

  nmarshak1337

  committed
  7c7e571

• It works! Penalty force seems a little funny though

  

  nmarshak1337

  committed
  099fc9b

• Elastic collisions work!

  

  nmarshak1337

  committed
  7ef81c9

• Collision detection works well, with nice clumping.

  

  nmarshak1337

  committed
  f50fd26

• Going back to no collisions for nao.

  

  nmarshak1337

  committed
  6f91518

• Last commit before adding RK4

  

  nmarshak1337

  committed
  a5c67e8

• Updated README

  

  nmarshak1337

  committed
  681b3e8

• Revert "Updated README"

  Show description for b91bb5a
  

  nmarshak1337

  committed
  b91bb5a

• Revert "Last commit before adding RK4"

  Show description for 74a57a3
  

  nmarshak1337

  committed
  74a57a3

• Collision detection is now back in!

  

  nmarshak1337

  committed
  c25a416

• Now we are back to the "standard" demo. Time to add flocking!

  

  nmarshak1337

  committed
  105963f

• Now collisions conserve energy (or at least are better at conserving

  Show description for c93f600
  

  nmarshak1337

  committed
  c93f600

• Duplicated the nbody project in order to start with flocking.

  

  nmarshak1337

  committed
  d43a784

Commits on Oct 22, 2013

• Well, the star mass is zero. I'll figure out how to get rid of it later.

  

  nmarshak1337

  committed
  c6fb6a0

• Well, now we have a pointer to the device position vector

  

  nmarshak1337

  committed
  5c1440c

• Got the guy to work. It's a little buggy though.

  

  nmarshak1337

  committed
  524ebee

• Added control law. Dunno if it works though.

  

  nmarshak1337

  committed
  a4acc1e

Commits on Oct 23, 2013

• Well, now we get the ball moving independetly of the other stuff

  

  nmarshak1337

  committed
  2121915

• Merge branch 'master' of https://github.com/nmarshak1337/CUDA-NBody-Sim

  Show description for df5be3b
  

  nmarshak1337

  committed
  df5be3b

• Now it actually works.

  

  nmarshak1337

  committed
  a454db5

• Got seek working. Disabled collisions, because it causes horrible,

  Show description for 0954019
  

  nmarshak1337

  committed
  0954019

• Got arrival working, I think.

  

  nmarshak1337

  committed
  794b798

• Separation works!

  

  nmarshak1337

  committed
  f0a7147

• Added enum for behavior type

  

  nmarshak1337

  committed
  311df47

• Switching between states works. Separation spazzes a bit when agents are

  Show description for 99021bf
  

  nmarshak1337

  committed
  99021bf

• Prevented separation from spazzing.

  

  nmarshak1337

  committed
  9a8a4e5

• Leader following works

  

  nmarshak1337

  committed
  6e67966

• Turned the planets into triangles. I just need the triangles to rotate nao

  

  nmarshak1337

  committed
  26ba57f

• Got triangles to rotate! Yippee!

  

  nmarshak1337

  committed
  287d8ad

• We've proved that we change rotation dynamically. Now it's time to

  Show description for 28f76da
  

  nmarshak1337

  committed
  28f76da

• Got the rotation array working.

  

  nmarshak1337

  committed
  ac97e3f

• Angles are expressed as FLOATS now.

  

  nmarshak1337

  committed
  792eb30

• Added angular velocity array

  

  nmarshak1337

  committed
  b7252f8

• Code is down for calculation of control torque.

  

  nmarshak1337

  committed
  56269b1

• Base framework is now there for angular accelerations.

  

  nmarshak1337

  committed
  b0a3074

• Well, now we have torques. Phew!

  

  nmarshak1337

  committed
  4744d27

• Ready to drop in numerical integration of angle/angular velocity

  

  nmarshak1337

  committed
  a633f8c

• Numerical integration of angular acceleration is there. Now to give it a

  Show description for bba641a
  

  nmarshak1337

  committed
  bba641a

• Angle control works... sort of!

  

  nmarshak1337

  committed
  89ffc51

• Got angle control to work!

  

  nmarshak1337

  committed
  3e93738

• Updated README

  

  nmarshak1337

  committed
  790f6d8

• We are back to using shared, tile size and block size are both 128

  

  nmarshak1337

  committed
  0d4df48