Code

Chi sqrd gradient#147

Code

Open

marziarivi wants to merge 34 commits intomasterratt-ru/montblanc:masterfrom chi_sqrd_gradientratt-ru/montblanc:chi_sqrd_gradientCopy head branch name to clipboard

+1,597-81

Commits

Commits on Jul 15, 2016

• GPU chi squared gradient with respect ro Sersic parameters

  marziarivi

  committed
  be1c515

  Copy full SHA for be1c515

• CPU chi squared gradient with respect ro Sersic parameters

  marziarivi

  committed
  9a5623a

  Copy full SHA for 9a5623a

• added chi squared gradient option

  marziarivi

  committed
  667aec1

  Copy full SHA for 667aec1

• added chi squared gradient option

  marziarivi

  committed
  ddc2f0a

  Copy full SHA for ddc2f0a

Commits on Jul 18, 2016

• Sersic gradient example

  marziarivi

  committed
  21dc3af

  Copy full SHA for 21dc3af

• Sersic gradient example

  marziarivi

  committed
  05433fd

  Copy full SHA for 05433fd

Commits on Jul 20, 2016

• uncommented multiplication by the G term

  marziarivi

  committed
  7541c01

  Copy full SHA for 7541c01

Commits on Jul 28, 2016

• added minor optimisations

  marziarivi

  committed
  b59b4e0

  Copy full SHA for b59b4e0

Commits on Aug 1, 2016

• move gradient elements product by 6 in the kernel

  

  Marzia Rivi

  committed
  6ad91b9

  Copy full SHA for 6ad91b9

• changed X2_grad classifier, now SIMULATOR_OUTPUT

  

  Marzia Rivi

  committed
  d083bd5

  Copy full SHA for d083bd5

• minor changes

  

  Marzia Rivi

  committed
  32734c2

  Copy full SHA for 32734c2

Commits on Aug 2, 2016

• minor changes

  

  Marzia Rivi

  committed
  30f1001

  Copy full SHA for 30f1001

• fixed bug #152 as in the master merge #156

  

  Marzia Rivi

  committed
  e0d0467

  Copy full SHA for e0d0467

Commits on Aug 3, 2016

• created multi-gpu gradient example

  

  Marzia Rivi

  committed
  0761ef5

  Copy full SHA for 0761ef5

• add X2 gradient computation within multi-gpu solver

  

  Marzia Rivi

  committed
  0e4e4a1

  Copy full SHA for 0e4e4a1

• add X2 gradient computation version 5

  

  Marzia Rivi

  committed
  6f7e2c0

  Copy full SHA for 6f7e2c0

• removed print instructions

  

  Marzia Rivi

  committed
  11b85ea

  Copy full SHA for 11b85ea

• fixed misalignment

  

  Marzia Rivi

  committed
  7a95988

  Copy full SHA for 7a95988

Commits on Aug 17, 2016

• bug fixed

  

  Marzia Rivi

  committed
  fc486a3

  Copy full SHA for fc486a3

• Composite solver merged with the master

  

  Marzia Rivi

  committed
  4738328

  Copy full SHA for 4738328

Commits on Aug 18, 2016

• minor change

  

  Marzia Rivi

  committed
  3d5d068

  Copy full SHA for 3d5d068

• bug fixed: X2_grad initialisation

  

  Marzia Rivi

  committed
  7bd82e9

  Copy full SHA for 7bd82e9

• minor change

  

  Marzia Rivi

  committed
  8e0866e

  Copy full SHA for 8e0866e

Commits on Aug 19, 2016

• change X2_grad classifier into GPU_SCRATCH to manage the case where gradient is not enabled

  

  Marzia Rivi

  committed
  5abd1e2

  Copy full SHA for 5abd1e2

• Fixed X2_grad array management in case of more src chunks

  

  Marzia Rivi

  committed
  a65b50f

  Copy full SHA for a65b50f

Commits on Aug 24, 2016

• Merge branch 'master' into chi-simon

  sjperkins

  committed
  4becebd

  Copy full SHA for 4becebd

Commits on Aug 25, 2016

• Pass results via dict, instead of tuple

  Show description for 7457d38
  sjperkins

  committed
  7457d38

  Copy full SHA for 7457d38

Commits on Sep 8, 2016

• Revert "Pass results via dict, instead of tuple"

  Show description for 15e99b8
  

  Marzia Rivi

  committed
  15e99b8

  Copy full SHA for 15e99b8

• replaced atomic addition in the shared memory with the block_reduce function

  

  Marzia Rivi

  committed
  4879f6b

  Copy full SHA for 4879f6b

Commits on Sep 19, 2016

• fixed block_reduce bug in case of block dimensions not factors of nbaselines and ntimes

  

  Marzia Rivi

  committed
  6e30f12

  Copy full SHA for 6e30f12

Commits on May 10, 2017

• Enabled single polarization for GPU version

  

  Marzia Rivi

  committed
  2795b5f

  Copy full SHA for 2795b5f

Commits on May 11, 2017

• allowed solver reading context from the slvr_cfg

  

  Marzia Rivi

  committed
  a3b51fd

  Copy full SHA for a3b51fd

Commits on Sep 19, 2017

• bugs fixed: E_beam and G_term matrix default, RimeBSqrt function for npol=1

  

  Marzia Rivi

  committed
  fc46cbb

  Copy full SHA for fc46cbb

Commits on Sep 22, 2017

• bug fixed for the case npol=1: final reduction didn't take into account the case for a single channel where num threadIdx.x = 1 < 3

  

  Marzia Rivi

  committed
  4d06dbd

  Copy full SHA for 4d06dbd