Commits

Breadcrumbs

History for

onlowerC99

User selector

All users

All time

Commit History

Commits on Aug 2, 2016

• Fix compile warnings.

  abergeron

  committed
  3446142

  Copy full SHA for 3446142

Commits on Aug 1, 2016

• Add GA_CUDA_EXIT_ON_ERROR at cuda_wait/record

  tsirif

  committed
  003a740

  Copy full SHA for 003a740

Commits on Jun 10, 2016

• Actually save the error code so it can be reported.

  abergeron

  committed
  c255969

  Copy full SHA for c255969

Commits on Jun 8, 2016

• Add a way to get a better description of blas errors.

  abergeron

  committed
  4eeac6e

  Copy full SHA for 4eeac6e

Commits on May 29, 2016

• cuda 8 API changes to cublasSgemmEx

  kashif

  committed
  9690136

  Copy full SHA for 9690136

Commits on May 24, 2016

• Hide the definiton of gpuarray_buffer_ops and provide a new less

  Show description for 8e3bed4
  abergeron

  committed
  8e3bed4

  Copy full SHA for 8e3bed4

• Fix cuda stuff for gpucontext.

  abergeron

  committed
  fcf3775

  Copy full SHA for fcf3775

Commits on Apr 19, 2016

• Fix compile errors that slipped in during rebase.

  abergeron

  committed
  07d9102

  Copy full SHA for 07d9102

• Add comment about the meaning of flags in the new Batch methods.

  abergeron

  committed
  a0e8046

  Copy full SHA for a0e8046

• Fix last two nits in the batch kernels.

  abergeron

  committed
  333fbce

  Copy full SHA for 333fbce

• Fix the batch kernel compilation.

  abergeron

  committed
  2d47ff6

  Copy full SHA for 2d47ff6

• Declare the batch kernels extern "C" for reachability.

  abergeron

  committed
  7c209c6

  Copy full SHA for 7c209c6

• Fix allocation size of handle.

  abergeron

  committed
  c1d5c95

  Copy full SHA for c1d5c95

• Fix errors in kernel code.

  abergeron

  committed
  36d65e3

  Copy full SHA for 36d65e3

• Make the blas ops slightly less fragile by having placeholder functions for the missing functionality.

  abergeron

  committed
  b36cc50

  Copy full SHA for b36cc50

• Fix build of the cuda part and prototypes.

  abergeron

  committed
  52685c0

  Copy full SHA for 52685c0

• Initial code for batch version of all the blas functions.

  abergeron

  committed
  f521b9c

  Copy full SHA for f521b9c

Commits on Apr 14, 2016

• use size_t for threshold computation

  cooijmanstim

  committed
  1ea8a45

  Copy full SHA for 1ea8a45

Commits on Apr 13, 2016

• implement multiple dispatch for dgemmBatch

  cooijmanstim

  committed
  e4ba190

  Copy full SHA for e4ba190

• switch sign of threshold

  cooijmanstim

  committed
  a776e5d

  Copy full SHA for a776e5d

• rgemmBatch -> rgemmBatch_3d, fix bugs

  cooijmanstim

  committed
  05bcc86

  Copy full SHA for 05bcc86

• cublas sgemmBatch: first stab at multiple dispatch implementation

  cooijmanstim

  committed
  ff13899

  Copy full SHA for ff13899

Commits on Sep 8, 2015

• Make sure we stay in the context when we manipulate the cublas handler.

  abergeron

  committed
  192484e

  Copy full SHA for 192484e

Commits on Sep 2, 2015

• Add ASSERT_BUF() where needed.

  abergeron

  committed
  977cd57

  Copy full SHA for 977cd57

Commits on Sep 1, 2015

• Add CMake black magic to check for cublasSgemmEx

  abergeron

  committed
  92e0acf

  Copy full SHA for 92e0acf

• Rename cuda_mark() to cuda_record()

  abergeron

  committed
  81d2b1d

  Copy full SHA for 81d2b1d

• Stop masking the error with cuda_enter().

  abergeron

  committed
  919f3fa

  Copy full SHA for 919f3fa

• Stop generating the blas files and stop using the gnarly macro-filled blas file.

  abergeron

  committed
  0ba987a

  Copy full SHA for 0ba987a

Commits on Apr 22, 2014

• Rename everything.

  abergeron

  committed
  430309f

  Copy full SHA for 430309f

Loading