Code

[TOPI][OP] cuda for argwhere#6868

Code

Merged

kevinthesun merged 13 commits intoapache:mainapache/tvm:mainfrom zhiics:argwherezhiics/tvm:argwhereCopy head branch name to clipboardDec 4, 2020

+795-36

Commits

Commits on Dec 4, 2020

• argwhere

  zhiics

  committed
  68fb2f6

  Copy full SHA for 68fb2f6

• cuda schedule

  zhiics

  committed
  d91598b

  Copy full SHA for d91598b

• sort argwhere result

  zhiics

  committed
  b74d59f

  Copy full SHA for b74d59f

• Use single block and thrust to fix flaky behavior

  

  kevinthesun

  authored and

  zhiics

  committed
  6d83d85

  Copy full SHA for 6d83d85

• format

  zhiics

  committed
  1716846

  Copy full SHA for 1716846

• used dynamic strided_slice

  zhiics

  committed
  18e10bf

  Copy full SHA for 18e10bf

• Fix dynamic strided_slice

  

  kevinthesun

  authored and

  zhiics

  committed
  6824edf

  Copy full SHA for 6824edf

• try new strided_slice

  zhiics

  committed
  689732a

  Copy full SHA for 689732a

• Improve dynamic strided slice to bind data depedent shape var.

  

  kevinthesun

  authored and

  zhiics

  committed
  2524663

  Copy full SHA for 2524663

• all tests pass

  zhiics

  committed
  8c85766

  Copy full SHA for 8c85766

• remove print

  zhiics

  committed
  736dfe7

  Copy full SHA for 736dfe7

• use new strided_slice

  zhiics

  committed
  ea8ba0c

  Copy full SHA for ea8ba0c

• clean

  zhiics

  committed
  0a77250

  Copy full SHA for 0a77250