Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on May 14, 2019

• t

  

  Roman Dzhabarov

  committed
  d2e7eb8

• [onnxifi] More debug info in case of failures.

  

  Roman Dzhabarov

  committed
  35ed7c8

• Add copyOutputs trace event to HabanaDeviceManager (pytorch#2906)

  Show description for ff44209
  

  jackm321

  authored and

  facebook-github-bot

  committed
  ff44209

• Eliminate duplicate allocation for weights in model loaders (pytorch#2865)

  Show description for 96db0f1
  

  jackm321

  authored and

  facebook-github-bot

  committed
  96db0f1

• Some cosmetic changes in the HabanaBackendTest. (pytorch#2904)

  Show description for bd7f51c
  

  Roman Dzhabarov

  authored and

  facebook-github-bot

  committed
  bd7f51c

• Unify the scripts for downloading database and caffe2/onnx models. (pytorch#2875)

  Show description for 3ec5904
  

  Chi Zhang

  authored and

  facebook-github-bot

  committed
  3ec5904

• Make debug printing friendly to gcc 4.8.5

  Show description for 0456859
  

  bertmaher

  authored and

  facebook-github-bot

  committed
  0456859

• Make build_llvm take RELEASE from environment if provided (pytorch#2901)

  Show description for a6f73a4
  

  bertmaher

  authored and

  facebook-github-bot

  committed
  a6f73a4

• always generate TraceEvents when doing debug dumping (pytorch#2902)

  Show description for 1141b07
  

  nickgg

  authored and

  facebook-github-bot

  committed
  1141b07

Commits on May 13, 2019

• Adding RecommendationSystemTest for SparseLengthSum based models (pytorch#2894)

  Show description for 93787fe
  

  nickgg

  authored and

  facebook-github-bot

  committed
  93787fe

• add TensorPool and speed up adjust batch (pytorch#2857)

  Show description for f2e9258
  

  nickgg

  authored and

  facebook-github-bot

  committed
  f2e9258

• Removed deviceConfig specialization. Added string map for parameters. (pytorch#2874)

  Show description for b1e2f24
  

  gcatron

  authored and

  facebook-github-bot

  committed
  b1e2f24

• Use onnxifi output tensors directly (pytorch#2884)

  Show description for 47a137e
  

  jackm321

  authored and

  facebook-github-bot

  committed
  47a137e

• Make 1 library per lib/ subdirectory (pytorch#2891)

  Show description for e02b873
  

  bertmaher

  authored and

  facebook-github-bot

  committed
  e02b873

• Tweak GlowErr message (pytorch#2893)

  Show description for f5149dc
  

  jackm321

  authored and

  facebook-github-bot

  committed
  f5149dc

Commits on May 11, 2019

• Refactor Backends libraries (pytorch#2877)

  Show description for 8fa1cf5
  

  bertmaher

  authored and

  facebook-github-bot

  committed
  8fa1cf5

Commits on May 10, 2019

• Remove uses of explicit result indices where possible (pytorch#2886)

  Show description for a2f64d0
  

  jfix71

  authored and

  facebook-github-bot

  committed
  a2f64d0

• Create test recipes in temp directory (pytorch#2887)

  Show description for 8895b35
  

  bertmaher

  authored and

  facebook-github-bot

  committed
  8895b35

• Hoist QuantizationProfile instructions right after the last node that updates their inputs to shorten lifetimes of buffers (pytorch#2698)

  Show description for 3abdf8d
  

  opti-mix

  authored and

  facebook-github-bot

  committed
  3abdf8d

• Allow Quantized output, save Function's name.

  Show description for 36e709c
  

  artemrakhov-glow

  authored and

  facebook-github-bot

  committed
  36e709c

• Fix selectRepFunc() not to return null. (pytorch#2882)

  Show description for 6195409
  

  artemrakhov-glow

  authored and

  facebook-github-bot

  committed
  6195409

• add commandline option to dump glow traces for debugging

  Show description for 79f698a
  

  nickgg

  authored and

  facebook-github-bot

  committed
  79f698a

• Log topology's IO tensors for debugging

  Show description for 35242be
  

  bertmaher

  authored and

  facebook-github-bot

  committed
  35242be

• fix Onnxifi TraceEvent propagation and implement 'I' and 'X' events in C2-Artillery interface

  Show description for cdbe0db
  

  nickgg

  authored and

  facebook-github-bot

  committed
  cdbe0db

• Add BatchMatMul node to allow backends to prevent lowering (pytorch#2868)

  Show description for d65e305
  

  jfix71

  authored and

  facebook-github-bot

  committed
  d65e305

• FullyConnectedNodeKind is not supported by Interpreter backend (pytorch#2879)

  Show description for 43026fd
  

  Man Wang

  authored and

  facebook-github-bot

  committed
  43026fd

• Let backends decide if they support kernel stacking for specific instructions

  Show description for 1e067bf
  

  opti-mix

  authored and

  facebook-github-bot

  committed
  1e067bf

• Updated ONNX to latest. (pytorch#2881)

  Show description for 24d443a
  

  Alexander Yermolovich

  authored and

  facebook-github-bot

  committed
  24d443a

Commits on May 9, 2019

• SpaceToDepth (pytorch#2860)

  Show description for 4a3eb12
  

  Alexander Yermolovich

  authored and

  facebook-github-bot

  committed
  4a3eb12

• Reduce multi device startup latency by minimizing lock time (pytorch#2856)

  Show description for e2862b4
  

  nickgg

  authored and

  facebook-github-bot

  committed
  e2862b4

Commits on May 8, 2019

• This fixes an issue with OpenCL in the runtime. The module was being cleared but OpenCL needs tensor type information. This adds a new release method to the tensor that frees the data and converts …

  Show description for 3201a3b
  gcatron

  committed
  3201a3b

• Added a command line flag to toggle activations memory allocation reuse

  gcatron

  committed
  0586bda

• Set up ShipIt (pytorch#2869)

  Show description for 1219b41
  arunm-git

  authored
  1219b41

• [tracing] fix ScopedTraceBlock reordering bug (pytorch#2853)

  nickgg

  authored
  a1846cd

Commits on May 7, 2019

• [Habana] Enable gather and move test from fbcode. (pytorch#2867)

  rdzhabarov

  authored
  e2f1c54

Pagination

PreviousNext