Difference between revisions of "GPU621/Intel Parallel Studio Inspector."

From CDOT Wiki
Jump to: navigation, search
Line 78: Line 78:
 
Using Intel Inspector we can see where the race condition occurs in the code
 
Using Intel Inspector we can see where the race condition occurs in the code
  
[[File:RaceCondition_Inspector.jpg|950px]]
+
[[File:RaceCondition_Inspector.jpg|800px]]
  
 
''Indicating where the data race occurs''
 
''Indicating where the data race occurs''

Revision as of 22:36, 8 December 2021

Group Members

  1. Elliot Kyei
  2. Joel Raymond
  3. Muhammad Faiz

Project Overview

This project was created to provide detailed documentation on Intel Parallel Studio Inspector, how it works, and demonstrate key features that will help reduce write parallel

Intel Parallel Studio Inspector

Intel Parallel Studio Inspector is a dynamic tool that helps users detect memory and threading errors in their serial and multithreaded applications. Intel Inspector is available on Windows and Linux operating systems and works with C, C++, C#, and Fortran programming languages. For this project, we will be using Visual Studio 2019 with Inspector.

WorkFlow.png Intel Inspector Work Flow

Visit this site for more information on how to install Intel Inspector: https://www.intel.com/content/www/us/en/developer/tools/oneapi/inspector.html#gs.i68l9p

How to use Intel Inspector

Finding Memory Issues

Proper memory management is a common issue within programs, and it can be especially difficult to track. Intel Inspector can locate different memory issues such as memory leaks, memory corruption, allocation / de-allocation API mismatches, inconsistent memory API usage, illegal memory access, and uninitialized memory reads.

Once Intel Inspector is installed and you have a program you wish to analyze, navigate to:

Tools > Intel Inspector > New Analysis

New Analysis.jpg

Levels of Analyzation: Speed vs Thoroughness

An Intel Inspector window will pop up, with different features. At the top, Intel Inspector lists 3 different types of analysis levels that can be selected.

Analyszation Levels.jpg

Detect Leaks
The least thorough and intensive analysis. This setting reduces the stress on the system and cuts the resources and time to perform the analysis. As a result, this will produce a much faster analysis but will find a limited set of errors and provide fewer details

Detect Memory Problems
This setting indicates a medium-scope memory error analysis. It will increase time, resources, and load on the system when performing the analysis. This is a deeper level of analysis to find memory issues but is slightly slower.

Locate Memory Problems
This maximizes the scope of the memory analysis. Using this setting will also maximize the time, resources, and load on the system to perform the analysis. This will detect an extensive range of memory issues, display the context of the problem and the highest degree of information available.

Memory Analysis Options

Below are explanations of some important memory analysis options that a user may consider selecting:

Detect resource leaks – detect if a GDI object is not deleted, or a kernel object is not closed. Useful for Windows GUI Applications

Enable interactive memory growth – detect if a region of memory has been allocated, but not deallocated during a certain time of the program’s execution.

Enable on-demand memory leak detection – detect if a region of memory has been allocated, but not deallocated during a certain time of the program’s execution and is not reachable. (No pointer to the memory location still exists)

Remove Duplicates – When this setting is on, Intel Inspector will not display all incidents of detection in the Code Location

Stack frame depth – Sets the ...

Starting a Memory Analysis

To start a memory analysis, the left-field must be set as “Memory Error Analysis”. After the desired options are checked, the user can click the “start” button located on the right-hand side to begin the analysis.

Start Analysis.jpg

Finding Nondeterministic Threading Errors

Intel Parallel Studio can help users detect a variety of threading errors such as data race conditions, deadlocks, lock hierarchy violations, and cross-thread stack access errors. These threading errors are usually non-deterministic and can be hard to reproduce.

Race Condition

Intel Inspector can be used to detect race conditions in programs. The following code is an example of race conditions occurring. In it, 5 threads are created to increment the value of an object several times. A race condition occurs when the threads race for the same data, therefore the value will be inconsistent and output different results. Normally in a data race, it becomes hard to locate data manually but with the help of Intel Inspector, it is much faster.

RaceConditions.jpg

RaceCondition Output.jpg

Values below 20000 are caused by race conditions

Using Intel Inspector we can see where the race condition occurs in the code

RaceCondition Inspector.jpg

Indicating where the data race occurs

Deadlock

Deadlocks can potentially happen when dealing with multi-threads. When 2 threads or more are stuck waiting for each other and trying to access the recourse but are being locked by the previous threads. In case of a deadlock, the program may run fine on the first try but the lock will eventually come up and crash the program. The following program will demonstrate deadlock occurring. The code involves resources protected by mutex locks. Their orders are m1->m2 or m2->m1. In some cases, 2 threads may cause a deadlock when they are waiting for a mutex owned by the other.

Deadlock Code.jpg

Deadlock Output.jpg

Deadlock occurs and the program doesn’t end

Using Intel Inspector, we can also detect deadlocks in the program

Deadlock Inspector.jpg

Points to mutex.lock function

Deadlock Inspector.jpg

Indicating where it occurred in the code

The intel inspector allows us to quickly locate where the error occurs and we can move on to debug the program and find the optimal solution.