Software Instrument Smart Placement

Table of Contents

• Introduction
• Short Coming of This Method
• Steps
  • 1. Recompilation of Libraries
  • 2. Recompilation of Application
  • 3. Test Run / Profiling Data Generation
  • 4. Profiling Data Extraction
  • 5. Extracted Data Processing
• 6. Linker update related to Smart Placement
• 7. Rebuilding the Application
• Benchmark Application

Introduction

This page goes over how to use software instrument smart placement.

Software instrumentation implementation make use of ti-arm-clang compiler's instrumentation feature.

Note

Make sure that ti-arm-clang compiler version is more than 3.1.0 STS.

Short Coming of This Method

1. Will not work for C++ code.
2. All static functions are never considered.
3. Internal memory constraint applications may not be able to use this.

Steps

1. Recompilation of Libraries

It needs to made sure that all the source code are built using -fprofile-instr-generate -fcoverage-mapping flags. Therefore, it is required to recompile entire SDK with these flags. To do this, please execute the following command on SDK top level.

gmake -j libs-scrub DEVICE=am263px

This will remove all compiled libraries. This will make sure to have a clean start. Then,

gmake -j libs DEVICE=am263px INSTRUMENTATION_MODE=yes

The above command will compile all the SDK libraries in instrumentation mode and in release profile. If it required to build libraries in debug mode then PROFILE=debug can be added at the end of the command.

Also, in Linux, gmake is to be replaced by make. The above commands assumes that current working directory is defined in MCU_PLUS_SDK_PATH environment variable.

2. Recompilation of Application

Application is also required to be compiled with those flags.

Note

The binary that is generated is an instrumented binary and its size is going to be more than the size of original application size. Therefore, it might be the case that linker.cmd file needs to be changed or if memory configurator is used then memory size needs to be changed.

For the changes that are required to be done for in memory configurator/linker script for instrumentation, please go through the following steps documented at How to add compiler generated sections for instrumentation binaries?.

Note

Although these changes are already at-least in empty SDK examples. So, changes can be copied and pasted directly from it and any more changes can be done on top of that.

Once those changes are done, execute following commands:

gmake -C path/to/folder/with/application/makefile scrub
gmake -C path/to/folder/with/application/makefile INSTRUMENTATION_MODE=yes

The above two commands assume that current working directory is defined in MCU_PLUS_SDK_PATH environment variable. Again, profile flag can be added as well to above commands.

3. Test Run / Profiling Data Generation

The instrumented binary needs to be run on the target. Make sure to run the application under correct scenario that would lead to good and complete profiling data generation. This is a very important step and quality and accuracy of the generated profiling data will determine quality of rest of the process.

This can be done connecting via CCS and can be read at CCS Launch, Load and Run

4. Profiling Data Extraction

1. Connect to target via CCS. Steps to do this can be found at CCS Launch, Load and Run
2. Open CCS scripting console CCS Tool Bar > View > Scripting Console and do below,

    js:> loadJSFile "C:/ti/mcu_plus_sdk/tools/smart_placement/coverage_dump.js"
   

   • NOTE replace C:/ti/mcu_plus_sdk with the absolute path to where the SDK is installed.

What this will do is, make a new .cnt file in same place as that of location of application’s .out file.

5. Extracted Data Processing

All the profiling data that is extracted, is stored in .cnt file. This .cnt file has to be processed and needs to be converted into a format that compiler will understand. To do this, type the following command:

gmake -C path/to/folder/with/application/makefile coverage

Again, if application is being built in debug mode, then make sure to provide PROFILE flag to the above command.

The output of this step would an ASM file with .S extension.

6. Linker update related to Smart Placement

Update linker as described at Changes in linker/Memory Configurator for Smart Placement.

Note

Although these changes are already at-least in Memory Benchmark with Smart Placement SDK examples. So, changes can be copied and pasted directly from it and any more changes can be done on top of that.

7. Rebuilding the Application

Recompile the application with new generated ASM file. To do this, add the file in makefile's ASMFILES_common variable. Note that, now INSTRUMENTATION_MODE flag is not required. So rebuild libs and application without these flags.

Benchmark Application

This demo provides a means of measuring the performance of a realistic application where the text of the application is sitting in various memory locations and the data is sitting in On-Chip-Memory RAM (referred to as OCM, OCMC or OCMRAM).

Note

All above steps are already done and generated ASM file has been renamed to annotations.S.

In case if above steps are required to run, then all linker related changes can be skipped as those are already taken into account.

Memory Benchmark with Smart Placement