Commits

Branch selector

 master

User selector

All users

All time

Commit History

Commits on Jul 13, 2018

• Merge branch 'dev'

  weiHelloWorld

  committed
  1fe5abe

  Copy full SHA for 1fe5abe

• updated assertion, to better locate error

  weiHelloWorld

  committed
  2f8aece

  Copy full SHA for 2f8aece

Commits on May 21, 2018

• added Softmax activation function, updated corresponding test

  weiHelloWorld

  committed
  45be010

  Copy full SHA for 45be010

Commits on May 4, 2018

• add timer, check CUDA compilation time

  weiHelloWorld

  committed
  963abae

  Copy full SHA for 963abae

Commits on Apr 16, 2018

• Python wrapper

  weiHelloWorld

  committed
  22312c0

  Copy full SHA for 22312c0

• updated

  weiHelloWorld

  committed
  be789bb

  Copy full SHA for be789bb

• Merge branch 'dev'

  weiHelloWorld

  committed
  13f41b1

  Copy full SHA for 13f41b1

• fixed computation

  weiHelloWorld

  committed
  534e456

  Copy full SHA for 534e456

• added CUDA implementation for pairwise distances as input (part 3 of unknown)

  weiHelloWorld

  committed
  01e40ef

  Copy full SHA for 01e40ef

• added CUDA implementation for pairwise distances as input (part 2 of unknown)

  weiHelloWorld

  committed
  2b4fc37

  Copy full SHA for 2b4fc37

Commits on Apr 15, 2018

• added CUDA implementation for pairwise distances as input (part 1 of unknown)

  weiHelloWorld

  committed
  9ed75b7

  Copy full SHA for 9ed75b7

Commits on Apr 13, 2018

• 1. fixed computation, 2. added test

  weiHelloWorld

  committed
  c6c7323

  Copy full SHA for c6c7323

• fixed issue of 'Conditional jump or move depends on uninitialised value(s)' related to class ANN_Force, this issue may be related to inproper assignment and initialization of 2D vector. There is s…

  Show description for 0f74458
  weiHelloWorld

  committed
  0f74458

  Copy full SHA for 0f74458

• updated

  weiHelloWorld

  committed
  e2c6817

  Copy full SHA for e2c6817

Commits on Apr 2, 2018

• added pairwise distances as inputs, temp

  weiHelloWorld

  committed
  65cf1d6

  Copy full SHA for 65cf1d6

Commits on Jun 21, 2017

• updated tests, support var number of atoms, one of purposes is to see compilation time vs. num of atoms

  weiHelloWorld

  committed
  8f1e60b

  Copy full SHA for 8f1e60b

• updated makefile and tests

  weiHelloWorld

  committed
  db458fc

  Copy full SHA for db458fc

• updated readme and Cmakelist

  weiHelloWorld

  committed
  64dc79e

  Copy full SHA for 64dc79e

Commits on May 26, 2017

• updated readme and test

  weiHelloWorld

  committed
  e246775

  Copy full SHA for e246775

Commits on May 4, 2017

• Create Licence.md

  weiHelloWorld

  authored
  562d4a0

  Copy full SHA for 562d4a0

Commits on Apr 28, 2017

• updated

  weiHelloWorld

  committed
  ead0ed7

  Copy full SHA for ead0ed7

Commits on Apr 7, 2017

• actually coor_center_of_mass is one component, each thread of thread 0/1/2 only needs to handle one component

  weiHelloWorld

  committed
  3ccc57e

  Copy full SHA for 3ccc57e

Commits on Apr 3, 2017

• added larger testing case, this is used to test valid number of threads (if the number of threads is too large, it will give error due to synchronization issue)

  weiHelloWorld

  committed
  8103215

  Copy full SHA for 8103215

Commits on Apr 2, 2017

• SAME FIX for cuda platform

  weiHelloWorld

  committed
  e2f7a9a

  Copy full SHA for e2f7a9a

• IMPORTANT FIX for reference platform, for the case with translation degrees of freedom removed

  weiHelloWorld

  committed
  4b697e1

  Copy full SHA for 4b697e1

Commits on Mar 14, 2017

• added comment

  weiHelloWorld

  committed
  3c7f45d

  Copy full SHA for 3c7f45d

Commits on Mar 13, 2017

• 1. updated tests, 2. added remove_translational_degrees_of_freedom to input

  weiHelloWorld

  committed
  a145113

  Copy full SHA for a145113

Commits on Mar 12, 2017

• temp update

  weiHelloWorld

  committed
  8b3db0b

  Copy full SHA for 8b3db0b

• updated test and ignore

  weiHelloWorld

  committed
  f76a7cd

  Copy full SHA for f76a7cd

• 2-step code generation, purposes: 1. - create code with minimal computation, where C++ works for pre-preprocessing optimization, 2. reduce memory access by replacing variable in shared memory with …

  Show description for 34c1d9f
  weiHelloWorld

  committed
  34c1d9f

  Copy full SHA for 34c1d9f

• reduced memory access, speed: 640 ns/day

  weiHelloWorld

  committed
  e4ff171

  Copy full SHA for e4ff171

• added helper function

  weiHelloWorld

  committed
  73fdbbe

  Copy full SHA for 73fdbbe

• make constant value localized, ~10% faster

  weiHelloWorld

  committed
  3930384

  Copy full SHA for 3930384

• fixed repeated computation for force by different threads

  weiHelloWorld

  committed
  f98630e

  Copy full SHA for f98630e

• first version of parallelization, benchmark: 470-500 ns/day, = 300 CPU cores

  weiHelloWorld

  committed
  fec2e84

  Copy full SHA for fec2e84

Pagination

Previous

Next