Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Aug 28, 2025

• workaround CUDA 13 API changes

  jeffhammond

  committed
  04248b6

  Copy full SHA for 04248b6

• add nvidia target for convenience

  jeffhammond

  committed
  7540a73

  Copy full SHA for 7540a73

• reorg nvshmem build stuff

  jeffhammond

  committed
  d664379

  Copy full SHA for d664379

Commits on Aug 27, 2025

• cursor found and fixed this bug

  jeffhammond

  committed
  6fc6257

  Copy full SHA for 6fc6257

Commits on Aug 19, 2025

• GPU MODE slides

  Show description for 5f56ed0
  jeffhammond

  authored
  5f56ed0

  Copy full SHA for 5f56ed0

Commits on Aug 16, 2025

• ignore

  jeffhammond

  committed
  23cd9b3

  Copy full SHA for 23cd9b3

• fix compiler errors

  jeffhammond

  committed
  fd425d8

  Copy full SHA for fd425d8

• user buffer in nccl

  jeffhammond

  committed
  88495de

  Copy full SHA for 88495de

• untested sketch of put version w nvshmem

  jeffhammond

  committed
  a8c2724

  Copy full SHA for a8c2724

• move the barrier - it's a 0.1% difference

  jeffhammond

  committed
  f66cd14

  Copy full SHA for f66cd14

• fix duplication nvshmem get

  jeffhammond

  committed
  7b81341

  Copy full SHA for 7b81341

• simple explanation of transpose.py

  jeffhammond

  committed
  fc7c697

  Copy full SHA for fc7c697

• plot transpose idea

  jeffhammond

  committed
  0ce8870

  Copy full SHA for 0ce8870

• update profiling

  jeffhammond

  committed
  fdaeb6c

  Copy full SHA for fdaeb6c

• fix profiling output

  jeffhammond

  committed
  7cd2f95

  Copy full SHA for 7cd2f95

• nvshmem profiling

  jeffhammond

  committed
  2251c87

  Copy full SHA for 2251c87

• CUDA events

  jeffhammond

  committed
  8ea5509

  Copy full SHA for 8ea5509

• add cuda events

  jeffhammond

  committed
  5e5a44c

  Copy full SHA for 5e5a44c

• implement perftest (no accumulate) in transpose cuda and cublas

  jeffhammond

  committed
  a421b9e

  Copy full SHA for a421b9e

• change BW measurement to use 4x bytes not 2x

  jeffhammond

  committed
  6d09934

  Copy full SHA for 6d09934

• disable ranges since dependencies are annoying

  jeffhammond

  committed
  182caf6

  Copy full SHA for 182caf6

• perf test measurement

  Show description for 8fd91cb
  jeffhammond

  committed
  8fd91cb

  Copy full SHA for 8fd91cb

• remove bulk variant since not applicable

  jeffhammond

  committed
  d241948

  Copy full SHA for d241948

• homogenize source

  jeffhammond

  committed
  2b831ef

  Copy full SHA for 2b831ef

• transpose nvshmem get on device non-naive variants

  jeffhammond

  committed
  7371559

  Copy full SHA for 7371559

• on device get version for naive

  jeffhammond

  committed
  de90d97

  Copy full SHA for de90d97

Commits on May 29, 2025

• grid stride params good now

  jeffhammond

  committed
  320c170

  Copy full SHA for 320c170

• fix block strided issues

  jeffhammond

  committed
  a47aec5

  Copy full SHA for a47aec5

• header

  jeffhammond

  committed
  f390def

  Copy full SHA for f390def

• block strided transpose

  jeffhammond

  committed
  26d8418

  Copy full SHA for 26d8418

Commits on Apr 17, 2025

• build updates

  jeffhammond

  committed
  933eab0

  Copy full SHA for 933eab0

• cleanup

  jeffhammond

  committed
  f662bb5

  Copy full SHA for f662bb5

• move barrier as permitted by nvshmem

  jeffhammond

  committed
  9cd0890

  Copy full SHA for 9cd0890

• disable launch bounds query code

  jeffhammond

  committed
  8e1b05f

  Copy full SHA for 8e1b05f

• in-kernel barrier will not work without redesigning the algorithm

  jeffhammond

  committed
  89eee87

  Copy full SHA for 89eee87

Pagination

Previous

Next