onnxruntime/onnxruntime
Chi Lo aa5e36456a
[TRT EP] Fix multithreading bug of getting the corrupted trt engine instance (#17507)
Revert to the old TRT EP behavior of securing the whole compute_function
by lock_guard.

Current TRT EP which only puts lock_guard around a critical section
(obvious wrong) inside compute_function.
The issue can happen where one thread is updating the engine in
compute_function whereas another thread still accesses the
stale/corrupted engine instance in compute_function, for example, the
code outside the critical section, `int total_bindings =
trt_engine->getNbBindings()`.

So, make the whole compute_function the critical section should be okay.
2023-09-12 07:37:45 -07:00
..
contrib_ops Flash Attention v2 MHA (#17227) 2023-08-31 13:52:21 -07:00
core [TRT EP] Fix multithreading bug of getting the corrupted trt engine instance (#17507) 2023-09-12 07:37:45 -07:00
python Enable optimize_by_onnxruntime call for float32 unet model (#17483) 2023-09-10 16:50:50 -07:00
test [QNN EP] Update QNN SDK to version 2.14.1 (#17467) 2023-09-11 21:07:50 -07:00
tool/etw Run clang-format in CI (#15524) 2023-04-18 09:26:58 -07:00
wasm Add training WASM generation to Web CI pipeline (#17319) 2023-09-08 15:49:47 -07:00
__init__.py ExecutionProvider API refactor - move allocator from EP level to SessionState level and indexed by OrtDevice (#15833) 2023-06-19 17:44:45 -07:00
ReformatSource.ps1 Run clang-format in CI (#15524) 2023-04-18 09:26:58 -07:00
ReformatSourcePython.bat
VSCodeCoverage.runsettings