Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on Sep 29, 2022

• Renamed to tinycudann32 to use FLOAT32 all the time.

  YOUSIKI

  committed
  b8be606

  Copy full SHA for b8be606

Commits on Sep 12, 2022

• PyTorch extension: permit installing without PyTorch CUDA being present

  Show description for 5f0d27a
  Tom94

  committed
  5f0d27a

  Copy full SHA for 5f0d27a

Commits on Aug 26, 2022

• Make deletion of GPUMemoryArena safe via reference counting

  Tom94

  committed
  ee210be

  Copy full SHA for ee210be

Commits on Aug 23, 2022

• PyTorch bindings: fix compilation on some compilers

  Show description for a11ccc9
  Tom94

  committed
  a11ccc9

  Copy full SHA for a11ccc9

Commits on Jul 13, 2022

• Multi GPU support

  Tom94

  committed
  5a734c1

  Copy full SHA for 5a734c1

Commits on Jul 12, 2022

• Ensure FullyFusedMLP isn't used with compute capability 70

  Tom94

  committed
  7f2ab26

  Copy full SHA for 7f2ab26

Commits on Jul 5, 2022

• Add fmt::format for terser string manipulation

  Show description for 4e8f235
  Tom94

  committed
  4e8f235

  Copy full SHA for 4e8f235

Commits on Mar 25, 2022

• Misc formatting

  Tom94

  committed
  ee585fa

  Copy full SHA for ee585fa

• Remove superfluous space in license header

  Tom94

  committed
  dd2ecfc

  Copy full SHA for dd2ecfc

• Fix indentation

  Tom94

  committed
  087d07c

  Copy full SHA for 087d07c

• Minor updates of comment

  ventusff

  committed
  b35e196

  Copy full SHA for b35e196

• Add d(dydx)_dx support & overload pos_fract

  ventusff

  committed
  83ac516

  Copy full SHA for 83ac516

Commits on Mar 18, 2022

• Unified module_function & throw Error directly at base class

  ventusff

  committed
  97ff02e

  Copy full SHA for 97ff02e

• Minor fix comments

  ventusff

  committed
  243db3c

  Copy full SHA for 243db3c

• Add partial support for second-order derivatives (only for grid.h's input)

  ventusff

  committed
  cb30a9f

  Copy full SHA for cb30a9f

Commits on Mar 17, 2022

• Fix PyTorch bindings compilation

  Tom94

  committed
  a3acb35

  Copy full SHA for a3acb35

Commits on Mar 15, 2022

• Update CUTLASS to 2.8 (and track as submodule)

  Show description for 1018206
  Tom94

  committed
  1018206

  Copy full SHA for 1018206

Commits on Mar 14, 2022

• Make GPUMatrix slice-able and unify encoding/network API as byproduct

  Tom94

  committed
  16a5d59

  Copy full SHA for 16a5d59

Commits on Feb 28, 2022

• PyTorch bindings: add function to free temporary memory allocations

  Tom94

  committed
  5c00809

  Copy full SHA for 5c00809

Commits on Feb 24, 2022

• PyTorch bindings: online compile fully_fused_mlp.cu with arch>=70

  Tom94

  committed
  fb8f845

  Copy full SHA for fb8f845

Commits on Feb 22, 2022

• PyTorch bindings: handle CPU tensors, but emit a warning

  Tom94

  committed
  ce54e70

  Copy full SHA for ce54e70

• PyTorch bindings: ensure target GPU arch is passed to nvcc

  Tom94

  committed
  953a8d0

  Copy full SHA for 953a8d0

Commits on Feb 21, 2022

• Homogenize batch size granularity across the codebase

  Tom94

  committed
  eb76667

  Copy full SHA for eb76667

Commits on Feb 17, 2022

• PyTorch: expose more info to Autograd & to print()

  Tom94

  committed
  8170062

  Copy full SHA for 8170062

Commits on Feb 16, 2022

• PyTorch bindings: automatically pad batch size

  Show description for a3acd81
  Tom94

  committed
  a3acd81

  Copy full SHA for a3acd81

• Allow building the PyTorch bindings without NNs (just the encoding; much faster build)

  Tom94

  committed
  70356c5

  Copy full SHA for 70356c5

• Only compute param gradients of torch requires them

  Tom94

  committed
  143a7f5

  Copy full SHA for 143a7f5

• Always expose fp32 encodings in addition to fp16

  Tom94

  committed
  4f01c04

  Copy full SHA for 4f01c04

Commits on Feb 15, 2022

• With ForwardCtx in place, don't require the same stream for the forward vs. backward pass

  Tom94

  committed
  1383bd6

  Copy full SHA for 1383bd6

• Decouple forward() context from Object

  Tom94

  committed
  7f126fe

  Copy full SHA for 7f126fe

Commits on Feb 14, 2022

• Add PyTorch extension

  Tom94

  committed
  2a3de3e

  Copy full SHA for 2a3de3e

Loading