IPU_STABLE_SDK_2.3.1_v8

ODLA-POPART

• Fixed a bug in multi-thread initialization.
• Fixed a bug in QManager initialization.

CMB Demo v0.2.4

1. Update vODLA Adapter to avoid redundant file transformation.

v0.8.0

This release contains the following major changes since v0.7.2:

• Enhance dynamic shape and dynamic batch support
• Support TF switch/merge control flow
• Support serialization of IR to protobuf format
• Improve fp16 support
• More ops support for Caffe, ONNX & Tensorflow
• More optimization on fusion and instruction simplification
• Bug fixes

ODLA:

• Add Cambrian MagicMind support
• Enhance ODLA/Popart support
• Upgrade ODLA/TRT to support cuda 11.4.2 / TensorRT 8.2.0

IPU_STABLE_SDK_2.3.1_v7_2

This release contains the following major changes since IPU_STABLE_SDK_2.3.1_v7

• Add tid print in vODLA Adapter

• Modified vODLA Adapter error message in odla_ExecuteComputation.

• Reset the QManager status when create the new queue to fit potential error handling scenario.

IPU_STABLE_SDK_2.3.1_v7

• Check the execution mode when do the async computation

• Make all resources used in thread intialized before start the thread

• debug core dump problem with vodla device reuse

• Add e.what() in print when catch exception

IPU_STABLE_SDK_2.3.1_v4_2

• add tid print in vODLA Adapter odla_ExecuteComputation

IPU_STABLE_SDK_2.3.1_v4_1

• add dynamic shape

IPU_STABLE_SDK_2.3.1_v6

• callback function definition changes, and handle the error raised by poplar SDK

• Move computation init to odla_BindToArgument to avoid early call odla_CreteContext before compuation graph constructed in ASYN call

• Fixed include sequence error when using makefile.

• Modified typedef vodla_context to keep consistence with vodh_sdk, and fixed several bug for odla_context.

• callback function definition changes, and handle the error raised by poplar SDK

• Move computation init to odla_BindToArgument to avoid early call odla_CreteContext before compuation graph constructed in ASYN call

• Add the mode string for pipeline_async

• Add more debug information for async context

• return the error by callback

• Use pipeline_async to create context when execution mode is UNKNOWN

• fixed a bug about adapter in debug mode

• fix for wrong simplify_preprocess option

• fix vODLA Adapter data reuse bug.

IPU_STABLE_SDK_2.3.1_v5

• Implemented vODLA adapter in parallel setting.

• Updated callback function and corresponding error handler.

• callback function definition changes, and handle the error raised by poplar SDK

• Move computation init to odla_BindToArgument to avoid early call odla_CreteContext before compuation graph constructed in ASYN call

IPU_STABLE_SDK_2.3.1_v4

• Add information to check the cache sdk version and the working sdk version

• Set the log level to info for test

• Add environment variable for error injection test

• format the code for lint check

• Add information to show the ENGION OPTIONS set

• Make it recompile when cache version mismatch

• Remote the environment for error simulator to make it set by environment variable

• Add error injector for temporarily test

• Fix the problem when there's not the error injection config file

• Move the error injector to the odla_CreateComputation to make sure it will be called for each inference request

• Add try catch to computation create and init

• make the error handling code at one place

• Use config file setting for batch per step when it's inconsistent with ODLA set

• set the log level when error injector used

• pass the error code to caller

• Make sdk version mismatch cache recompile when error injector set

• Make the popart log level take POPART_LOG_LEVEL firstly

• change format to pass lint

Co-authored-by: tianboh 54729592+tianboh@users.noreply.github.com