NVIDIA/TensorRT-Incubator

Issues

• Revise `tp.Compiler` class

  #220 opened 2 months ago
  1

• Bug: 'arith.constant' op failed to verify that all of {value, result} have same type

  #218 opened 2 months ago
  0

• Refactor `convert_inputs_to_tensors` decorator to move automatic shape conversion logic to a separate helper

  #214 opened 2 months ago
  1

• Segfault in mlir-tensorrt

  #208 opened a month ago
  2

• Concatenating a shape tensor using the `*` op does not work with fill

  #207 opened 3 months ago
  4

• Slicing a shape with a negative index sometimes does not work

  #206 opened a month ago
  4

• Add __mod__ for tp.Tensor

  #201 opened 3 months ago
  0

• Add NaN/Inf checking mode

  #200 opened 3 months ago
  0

• Add tp.Pooling

  #195 opened a month ago
  0

• Add tp.pad

  #194 opened a month ago
  0

• Add tp.resize

  #193 opened a month ago
  0

• Dynamic shape handling

  #192 opened 2 days ago
  1

• Slice with None at dim should add a new dimension

  #188 opened 3 months ago
  1

• Optimization = 3 accuracy issue for nanoGPT

  #186 opened a month ago
  0

• Add an assertion layer

  #183 opened 3 months ago
  0

• Add unit test to measure execution overhead of Tripy

  #179 opened 2 months ago
  1

• Add performance regression testing infrastructure for nanoGPT sample

  #178 opened 20 days ago
  1

• Avoid extra memory allocation and device-device copy

  #177 opened 3 months ago
  0

• Pipe nvtx markers from Tripy to TRT engine

  #176 opened 3 months ago
  1

• Add support for setting stream priorities in mlir-tensorrt

  #172 opened a month ago
  1

• Update `convert_inputs_to_tensors` to require Python numbers

  #168 opened 2 months ago
  1

• Add verification for compiler

  #166 opened 16 days ago
  1

• Add tp.cumsum

  #164 opened 3 months ago
  2

• Add tp.repeat_interleave

  #160 opened 3 months ago
  1

• Performance improvements in Tripy

  #159 opened 3 months ago
  3

• Support memref's with negative strides

  #158 opened 3 months ago
  0

• Add multi-function support to Flat IR

  #155 opened a month ago
  0

• floats with integer tensors not working with `__r<op>__`s

  #154 opened 20 days ago
  1

• Allclose no longer works with integers

  #153 opened 3 months ago
  3

• Add Bazel build system for MLIR-TensorRT

  #141 opened 4 months ago
  0

• Fix Tensor.tolist()

  #134 opened 2 months ago
  5

• Create versioned doc

  #133 opened 4 months ago
  0

• `tp.Tensor` constructed over `np.arange()` with reverse indexing gives corrupted view.

  #129 opened 3 months ago
  2

• 'tensorrt.shuffle' op operand #0 must be a tensor of allowed TensorRT element types, but got 'tensor<2x3x2xi1>

  #127 opened 20 days ago
  1

• `infer_devices` failing for `tp.copy`

  #126 opened 4 months ago
  3

• [compile_fixture] `test_iota_from_shape_tensor[compile]` fails

  #120 opened 20 days ago
  2

• Investigate slow tests in L0

  #118 opened 2 months ago
  5

• Improve int64 support in TensorRT dialect

  #116 opened 4 months ago
  1

• Int64 with gather results in MLIR-TRT error

  #115 opened 4 months ago
  2

• [compile_fixture] Scale must be a constant tensor in quantize / dequantize op.

  #102 opened 20 days ago
  3

• Validate device while initializing Tensor with an DLPack object

  #78 opened 4 months ago
  1

• Evaluate implementing constant folding using numpy in concatenateOp

  #70 opened 20 days ago
  1

• Gather doesn't work with arange and rank 2 index

  #68 opened 4 months ago
  1

• Accept shape tensors in Compiler and propagate bounds

  #65 opened 4 months ago
  1

• Improve Tripy documentation

  #54 opened 4 months ago
  0

• Add integration tests for gather op

  #52 opened 4 months ago
  1

• `get_arg_bound` for static shape arg returns empty list

  #44 opened 4 months ago
  1