Commits

Breadcrumbs

History for

onv3.8

User selector

All users

All time

Commit History

Commits on Oct 5, 2020

• Fix input ndims check for regions function

  

  pradeep

  authored and

  umar456

  committed
  096221d

  Copy full SHA for 096221d

• Add support for cuda 11.1 and compute 8.6

  umar456

  committed
  db3c333

  Copy full SHA for db3c333

• Fix index copy constructor by retaining array based index

  Show description for c903e57
  umar456

  committed
  c903e57

  Copy full SHA for c903e57

• Fix features copy constructor for multithreaded usage

  Show description for 465013c
  

  pradeep

  authored and

  umar456

  committed
  465013c

  Copy full SHA for 465013c

Commits on Aug 21, 2020

• Update select_compute_arch to support CUDA 11 and Ampere

  Show description for 1405a94
  

  umar456

  authored and

  pradeep

  committed
  1405a94

  Copy full SHA for 1405a94

Commits on Aug 20, 2020

• Add CUDA 11 to Toolkit2MaxCompute. Add Ampere sm to compute2cores

  

  umar456

  authored and

  pradeep

  committed
  2fb662a

  Copy full SHA for 2fb662a

• Add functions to set and get the cache directory for runtime kernels

  Show description for fe97af7
  

  umar456

  authored and

  pradeep

  committed
  fe97af7

  Copy full SHA for fe97af7

Commits on Aug 17, 2020

• Create a more portable macro that disables deprecated warnings (#2997)

  Show description for 87b94ad
  umar456

  authored
  87b94ad

  Copy full SHA for 87b94ad

Commits on Aug 16, 2020

• Fix bias factor of variance in var_all and cov functions

  

  pradeep

  committed
  13e1904

  Copy full SHA for 13e1904

• Add new std dev APIs with bias parameter

  

  pradeep

  committed
  690936f

  Copy full SHA for 690936f

• Add new covariance APIs with bias enum instead of bool

  

  pradeep

  committed
  d5789cb

  Copy full SHA for d5789cb

• Add new variance APIs with bias enum argument instead of bool

  Show description for f79d438
  

  pradeep

  committed
  f79d438

  Copy full SHA for f79d438

Commits on Aug 15, 2020

• Fix errors and warnings in RNG for CUDA 9.0

  umar456

  committed
  e62aab0

  Copy full SHA for e62aab0

• Formatting and rewording of warning messages

  Show description for 3e01de4
  umar456

  committed
  3e01de4

  Copy full SHA for 3e01de4

• Add utility header included from cuda_fp16.hpp for CUDA 9

  Show description for 0e8d5fd
  umar456

  committed
  0e8d5fd

  Copy full SHA for 0e8d5fd

• Fix checkAndSetDevMaxCompute when the device cc is greater than max

  Show description for 6b314a9
  umar456

  committed
  6b314a9

  Copy full SHA for 6b314a9

Commits on Aug 14, 2020

• Remove assert that check that signal/filter types have to be the same

  

  umar456

  authored and

  pradeep

  committed
  d849785

  Copy full SHA for d849785

Commits on Aug 13, 2020

• Correct interp helper calls in other kernels that use it

  

  pradeep

  committed
  8df8e6f

  Copy full SHA for 8df8e6f

• Address perf regression in approx after dim based interop was introduced

  

  pradeep

  committed
  5360328

  Copy full SHA for 5360328

Commits on Aug 12, 2020

• Fix problems in RNG for older compute architectures with fp16

  

  umar456

  authored and

  pradeep

  committed
  fe0c8d5

  Copy full SHA for fe0c8d5

• Further improve the box muller function by fixing rounding issues

  Show description for dffa8c5
  

  umar456

  authored and

  pradeep

  committed
  dffa8c5

  Copy full SHA for dffa8c5

• Fix warning in compile module about the length of the error message

  

  umar456

  authored and

  pradeep

  committed
  4d1c47b

  Copy full SHA for 4d1c47b

• Fix randn by passing in correct values to boxMuller

  Show description for c022b37
  

  umar456

  authored and

  pradeep

  committed
  c022b37

  Copy full SHA for c022b37

• Build Unified backend without OpenCL or CUDA installed if necessary

  Show description for 706924f
  

  umar456

  authored and

  pradeep

  committed
  706924f

  Copy full SHA for 706924f

• Move initializer_list constructor implementation to the header

  Show description for 8fd7eec
  

  umar456

  authored and

  pradeep

  committed
  8fd7eec

  Copy full SHA for 8fd7eec

Commits on Aug 8, 2020

• Add f16 support for histogram

  umar456

  committed
  ead53b9

  Copy full SHA for ead53b9

Commits on Aug 4, 2020

• Add min driver versions for CUDA 11

  

  pradeep

  committed
  95919f3

  Copy full SHA for 95919f3

Commits on Jul 31, 2020

• Add missing opencl-arrayfire interop fns in unified backend

  

  pradeep

  committed
  edb189c

  Copy full SHA for edb189c

Commits on Jul 17, 2020

• Add missing non-const Array::getNode method in CPU/OpenCL backends

  Show description for a67c346
  

  pradeep

  committed
  a67c346

  Copy full SHA for a67c346

Commits on Jul 15, 2020

• Use .bin extension for kernel binaries that are saved to disk

  

  pradeep

  committed
  75e3c6c

  Copy full SHA for 75e3c6c

• Check exceptions in CUDA compileModule and Log binary write/load failures

  

  pradeep

  committed
  58dc98e

  Copy full SHA for 58dc98e

• Enable disk caching support for OpenCL kernel binaries

  

  pradeep

  committed
  9cd94d3

  Copy full SHA for 9cd94d3

Commits on Jul 10, 2020

• Fix barrierOR fn in confidence connected opencl kernel (#2969)

  Show description for 64855cb
  

  pradeep

  authored
  64855cb

  Copy full SHA for 64855cb

Commits on Jul 9, 2020

• Add move constructor and operator= for dim4, Array<T> SparseArray<T>

  

  umar456

  authored and

  pradeep

  committed
  0308a5c

  Copy full SHA for 0308a5c

Commits on Jul 7, 2020

• Fix the svd ndims check to limit inputs to 2D arrays

  

  pradeep

  authored and

  umar456

  committed
  1682bcc

  Copy full SHA for 1682bcc