Commits

Branch selector

 main

User selector

jananisriram

Datepicker

All time

Commit History

Commits on Aug 1, 2024

• Add Inductor torch.compile benchmark for jagged_layer_norm operator

  Show description for 043b07d
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  043b07d

  Copy full SHA for 043b07d

• Add Inductor torch.compile benchmark for jagged_softmax operator

  Show description for 55f03bb
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  55f03bb

  Copy full SHA for 55f03bb

• Add Inductor torch.compile benchmark for jagged_mean operator

  Show description for c031f41
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  c031f41

  Copy full SHA for c031f41

• Add Inductor torch.compile benchmark for jagged_sum operator

  Show description for d7f6346
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  d7f6346

  Copy full SHA for d7f6346

• Refactor jagged operator benchmarks in TritonBench with helper functions for readability

  Show description for 31ebbaa
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  31ebbaa

  Copy full SHA for 31ebbaa

Commits on Jul 18, 2024

• Add variable-length loop Triton kernel benchmark for jagged_softmax

  Show description for 11ebabb
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  11ebabb

  Copy full SHA for 11ebabb

• Add simple fused Triton kernel benchmark for jagged_softmax

  Show description for c75d1a5
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  c75d1a5

  Copy full SHA for c75d1a5

• Add torch.sum benchmark for jagged_softmax operator

  Show description for f50c632
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  f50c632

  Copy full SHA for f50c632

Commits on Jul 15, 2024

• Add torch.sum benchmark for jagged_layernorm operator (#2376)

  Show description for e26cd75
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  e26cd75

  Copy full SHA for e26cd75

Commits on Jul 3, 2024

• Add variable-length loop Triton kernel benchmark for jagged_mean operator (#2356)

  Show description for cf24fa1
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  cf24fa1

  Copy full SHA for cf24fa1

• Add simple fused Triton kernel benchmark for jagged_mean operator (#2355)

  Show description for 1e79c04
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  1e79c04

  Copy full SHA for 1e79c04

• Add padded, torch.sum benchmark for jagged_mean operator (#2354)

  Show description for 2155130
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  2155130

  Copy full SHA for 2155130

• Add padded, torch.nanmean benchmark for jagged_mean operator (#2353)

  Show description for 5238ff1
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  5238ff1

  Copy full SHA for 5238ff1

• Add torch.mean benchmark for jagged_mean operator (#2352)

  Show description for 8146267
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  8146267

  Copy full SHA for 8146267

Commits on Jun 27, 2024

• Add plots variable along 1 input parameter to jagged_sum operator in TritonBench

  Show description for 847d38e
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  847d38e

  Copy full SHA for 847d38e

• Add Triton kernels with fast variable-length loop to jagged_sum benchmark (#2341)

  Show description for ec0909d
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  ec0909d

  Copy full SHA for ec0909d

Commits on Jun 26, 2024

• Extend support to varying block sizes on middle dimension of 3D tensors for sum operator

  Show description for 2f6ea58
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  2f6ea58

  Copy full SHA for 2f6ea58

• Add plots variable along 1 dimension to sum operator in TritonBench (#2330)

  Show description for 2643d15
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  2643d15

  Copy full SHA for 2643d15

Commits on Jun 21, 2024

• Add variable seqlen and sparsity parameters to jagged_sum benchmark (#2324)

  Show description for 1425f68
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  1425f68

  Copy full SHA for 1425f68

Commits on Jun 20, 2024

• Add unit tests on CPU for TritonBench features (#2323)

  Show description for a529b5a
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  a529b5a

  Copy full SHA for a529b5a

• Add simple fused Triton kernel for jagged_sum operator (#2322)

  Show description for a3af772
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  a3af772

  Copy full SHA for a3af772

Commits on Jun 18, 2024

• Add input_shape metric to jagged_sum operator (#2313)

  Show description for 5f6633d
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  5f6633d

  Copy full SHA for 5f6633d

• Add jagged_sum operator for padded nested tensors to TritonBench (#2305)

  Show description for 40b376d
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  40b376d

  Copy full SHA for 40b376d

Commits on Jun 17, 2024

• Add asymmetric shapes to test sum Triton kernels (#2310)

  Show description for bbeba0a
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  bbeba0a

  Copy full SHA for bbeba0a

Commits on Jun 14, 2024

• Refactor code for sum Triton kernels (#2303)

  Show description for 339ccfd
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  339ccfd

  Copy full SHA for 339ccfd

• Extend support to varying block sizes on both dimensions for 2D matrices (#2302)

  Show description for f4cbf78
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  f4cbf78

  Copy full SHA for f4cbf78

Commits on Jun 13, 2024

• Extend support for x_only = True extra metrics through canonical name retrieval (#2300)

  Show description for 86904ca
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  86904ca

  Copy full SHA for 86904ca

• Add jagged_sum operator for unpadded nested tensors to TritonBench (#2299)

  Show description for 576b2b2
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  576b2b2

  Copy full SHA for 576b2b2

Commits on Jun 12, 2024

• Add support for reducing across the middle dimension for 3D matrices using the sum Triton kernel (#2297)

  Show description for 55c975e
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  55c975e

  Copy full SHA for 55c975e

• Add support for x_only = True extra metrics during table generation (#2294)

  Show description for 217d8f8
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  217d8f8

  Copy full SHA for 217d8f8

• Add support for reducing across individual dimensions for 2D matrices using the sum Triton kernel (#2295)

  Show description for 3ecaae9
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  3ecaae9

  Copy full SHA for 3ecaae9

Commits on Jun 7, 2024

• Add command line argument parsing for reduction dimensions in Triton sum kernel (#2284)

  Show description for 9f2ab74
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  9f2ab74

  Copy full SHA for 9f2ab74

Commits on Jun 6, 2024

• Add sum reduction operator to TritonBench (#2282)

  Show description for 2d8999b
  

  jananisriram

  authored and

  facebook-github-bot

  committed
  2d8999b

  Copy full SHA for 2d8999b