Code

Half matrix and components#1708

Code

Merged

yhmtsai merged 16 commits intodevelopginkgo-project/ginkgo:developfrom half_matrixginkgo-project/ginkgo:half_matrixCopy head branch name to clipboardDec 3, 2024

+2,822-1,356

Commits

Commits on Dec 3, 2024

• instantiation/testing/next/prev/stub type definition

  yhmtsai

  committed
  502df54

  Copy full SHA for 502df54

• half option

  yhmtsai

  committed
  c53ab57

  Copy full SHA for c53ab57

• device type mapping

  yhmtsai

  committed
  6f14b13

  Copy full SHA for 6f14b13

• consider custom namespace for thrust::complex<__half> and benchmark

  yhmtsai

  committed
  476cf28

  Copy full SHA for 476cf28

• atomic and cooperative_groups

  yhmtsai

  committed
  7cc8c6f

  Copy full SHA for 7cc8c6f

• fix math and device_numeric_limit

  yhmtsai

  committed
  a85f462

  Copy full SHA for a85f462

• array operation in half

  yhmtsai

  committed
  7b4829c

  Copy full SHA for 7b4829c

• matrix with half

  yhmtsai

  committed
  561f173

  Copy full SHA for 561f173

• device_matrix_data and mtx_io

  yhmtsai

  committed
  036485a

  Copy full SHA for 036485a

• components such as array/iterator/segmented_array test with half

  yhmtsai

  committed
  b29a8f6

  Copy full SHA for b29a8f6

• matrix test with half

  yhmtsai

  committed
  2be0042

  Copy full SHA for 2be0042

• base such as composition/combination with half and corr. test

  yhmtsai

  committed
  8d3e4b5

  Copy full SHA for 8d3e4b5

• test_utils test

  yhmtsai

  committed
  b2fa55a

  Copy full SHA for b2fa55a

• constexpr restriction for nvc++

  yhmtsai

  committed
  8910f83

  Copy full SHA for 8910f83

• cuda with CC<70 and hip do not support 16 bit atomic. throw error or fallback to a working version if it is the case for matrix

  yhmtsai

  committed
  0421615

  Copy full SHA for 0421615

• implement half shuffle via 32 bit impl

  yhmtsai

  committed
  8190bf6

  Copy full SHA for 8190bf6