v0.5.0

What's Changed

• docs: update README for v0.4.0 changes by @theHamsta in #54
• ci: update actions/checkout to v6 by @theHamsta in #55
• fix: add_quantize/add_dequantize for enterprise by @theHamsta in #57
• feat: add missing methods on Tensor and give Tensor its own file by @theHamsta in #56
• fix: don't publicly export check_network by @theHamsta in #58
• docs: reference documentation for IExecutionContext::setTensorAddress by @theHamsta in #60
• feat: add ICastLayer::{set,get}ToType by @theHamsta in #62
• trtexec-rs by @theHamsta in #49
• Determine TensorRT version from header by @theHamsta in #61
• feat: add nvinfer1::IProfiler by @theHamsta in #63
• refactor!: rename ExecutionContext getters and give ExecutionContext its own file by @theHamsta in #66
• refactor!: clean up getter names and doc references in OptimizationProfile by @theHamsta in #68
• refactor!: clean up docs and method names in CudaEngine by @theHamsta in #67
• refactor: deprecate tensor_dtype and more doc links for CudaEngine by @theHamsta in #69
• feat: add ExecutionContext::set_communicator by @theHamsta in #73
• chore: remove duplicate v_1_4 tags around docs and code by @theHamsta in #74
• feat: more methods on ExecutionContext by @theHamsta in #75
• feat: add CudaEngine::name by @theHamsta in #77
• Profile.json by @theHamsta in #70
• fix: fix library loading (again) by @theHamsta in #81
• fix(onnxparser)!: Make sure the parser can't be dropped before build by @theHamsta in #72
• fix!: add missing builder lifetime to BuilderConfig by @theHamsta in #80
• refactor!: remove get_ prefix from getters by @theHamsta in #76
• fix: undeprecate Tensor::get_type (.r#type is awkward) by @theHamsta in #83
• publish: bump version to v0.5.0 by @theHamsta in #84

Full Changelog: v0.4.0...v0.5.0

v0.4.0

What's Changed

• Expose all methods of IBuilderConfig by @theHamsta in #15
• Expose IHostMemory by @theHamsta in #24
• feat!: refactor for stronger use of autocxx by @theHamsta in #28
• refactor: let NetworkDefinition::add_input accept a i64 shape by @theHamsta in #31
• ci: add a configuration with onnxparser disabled by @theHamsta in #34
• Allow taking ownership of convolution weights by @theHamsta in #32
• build: Try to avoid invalidation of trtx-sys(build.rs) by @theHamsta in #35
• feat: add handler for logging with log crate by @theHamsta in #11
• Add some more methods on layers by @theHamsta in #37
• Add cross-references to C++ doxygen online documentation by @theHamsta in #30
• feat: add support for v_1_4 layers (IMoELayer, IDistCollectiveLayer) by @theHamsta in #39
• fix: ensure lock is hold when loading library by @theHamsta in #38
• feat: implement methods on PoolingLayer by @theHamsta in #44
• Quantize/Dequantize/DynamicQuantize methods by @theHamsta in #45
• ci: reorder checks to put most relevant first by @theHamsta in #40
• Add support for TensorRT non-RTX (mainly for debugging/validation of RTX) by @theHamsta in #36
• Add missing functions of the ResizeLayer by @mtavenrath in #48
• feat: add GridSampleLayer by @theHamsta in #47
• feat: add more logging to trtx by @theHamsta in #42
• feat: add methods on AttentionLayer by @theHamsta in #46
• feat: make Tensor implement Clone, Copy by @theHamsta in #43
• Bump version v0.4 by @theHamsta in #50
• ci: remove mock feature from publish workflow by @theHamsta in #52
• ci: really fix publish workflow by @theHamsta in #53

Full Changelog: v0.3.1...v0.4.0

v0.3.1

What's Changed

• docs: update library docs with changes from README in #8 by @theHamsta in #10
• fix: only load the TRTLIB once by @theHamsta in #12
• deps: update depends to newest versions by @theHamsta in #21
• feat: dynamically load TRT libraries on demand by @theHamsta in #13
• build: transform TensorRT-Headers if necessary by @theHamsta in #14
• Add support for missing layers required by RustNN and bump version number to 0.3.1 by @mtavenrath in #22

Full Changelog: v0.3.0...v0.3.1

v0.3.0

What's Changed

• Fix segfaults and CUDA / TRT-RTX dependency issues by @mtavenrath in #1
• Migrate FFI bindings to autocxx and add support for the TRT-RTX INetworkDefinition interface by @mtavenrath in #2
• Fix compilation on Linux by casting IHostMemory::size usage by @theHamsta in #5
• Allow dynamic loading of TensorRT lib by @theHamsta in #3
• Vendor TensorRT headers by @theHamsta in #6
• chore: remove CUDA wrappers by @theHamsta in #9
• docs: README update for #3 and #6 by @theHamsta in #8
• Bump version to 0.3.0 by @mtavenrath in #7

New Contributors

• @mtavenrath made their first contribution in #1
• @theHamsta made their first contribution in #5

Full Changelog: v0.2.0...v0.3.0