Intel® VTune™ Profiler Release Notes and New Features

ID 764439
Updated 1/18/2024
Version 2024.0.1
Public

author-image

By

This page provides the current Release Notes for Intel® VTune™ Profiler (starting with Intel® VTune™ Amplifier XE 2017). The notes are categorized by major version, from newest to oldest, with individual releases listed within each version section.

NOTE: Intel® VTune™ Amplifier has been renamed to Intel® VTune™ Profiler starting with version 2020.

Click a release to expand it into a summary of new features and changes in that version since the last release.

The installation guides are posted separately:
View Installation Guide

For Intel® VTune™ Profiler usage, visit Intel® VTune™ Profiler User Guide

Where to Find the Release

If you have a qualifying license for this product (i.e. an unexpired free license or a paid license with active support at the time of the build date for this product release), you can download the installer by logging in to Intel® Software Development Products Registration Center and selecting the appropriate product.

You may need to create an account and/or register your product. For additional information please see:

Product Registration and Sign-Up FAQ

Where to Get Help

For technical support, including answers to questions not addressed in this product, visit the technical support forum, FAQs, and other support information at:

Register your product at Intel® Software Development Products Registration Center by providing your email address. Registration entitles you to free technical support, product updates and upgrades for the duration of the support term. It also helps Intel recognize you as a valued customer in the support forum.

System Requirements

Please see the Intel® VTune™ Profiler System Requirements.

New Features Support

2024.0.1

Overview

This is a patch release of Intel® VTune™ Profiler and contains new support for this Intel® Data Center Max GPU:           

Intel® Data Center Max 1100C GPU

  • A 300-watt double-wide AIC card with 56 Xe cores and 48 GB of HBM2E memory. Multiple cards can be connected via Intel Xe Link bridges.  This card is specifically designed to meet the requirements of US Government Export Control Regulations for export into China.

2024.0

Overview

This version of Intel® VTune™ Profiler contains improvements and additions in the following areas:           

Application Performance Snapshot

  • Xe Link observability
    Added two Xe Link related metrics to APS summary with average cross-GPU traffic bandwidth through Xe Links. It shows the average value per-node/per-gpu. It is also possible to get the value through CLI metrics report attributed to the device.

Hardware Support 

  • Support for 5th Gen Intel® Xeon® Server Processors (code named Emerald Rapids)
    Includes I/O analysis support
    Enabled Platform Diagram

OS Support

  • Fedora* 38 and 39
  • Ubuntu* 23.04
  • SUSE* Linux* Enterprise Server (SLES) 15 SP5
  • Windows* 11 23H2 

2023.2

Overview

This version of Intel® VTune™ Profiler contains improvements and additions in the following areas:

GPU Accelerators  

  • Better Hardware observability in the Platform Diagram.
    Identify Intel® Data Center GPU Max Series (formerly codenamed Ponte Vecchio) observability in the platform diagram.           

Application Performance Snapshot

  • Analyze the performance utilization of a particular GPU in the stack using Application Performance Snapshot.
    For each GPU, get insights into how much time is used for offload tasks and how the execution units are utilized.

Language Support 

  • Support for New Language Versions
    This release introduces support for Python 3.10.* in the Hotspots Analysis type for Windows and Linux systems.
    Explore performance analysis for .NET7 applications or native applications with .NET calls.

Intel® VTune™ Profiler - Platform Profiler 

  • Starting with the 2023.2 release, the Intel® VTune™ Profiler - Platform Profiler is available as a standalone application only.  
    The Platform Profiler analysis will no longer be available through the VTune Profiler GUI or command line. You can still collect platform behavior data by running the standalone application: https://www.intel.com/content/www/us/en/docs/vtune-profiler/user-guide/current/platform-analysis.html. The data collected with the Platform Profiler collector can be visualized with the Platform Profiler server.

Hardware Support 

  • Better Hardware observability in the Platform Diagram 
     Identify average HBM bandwidth on Intel® Xeon® Max Series CPUs (code named Sapphire Rapids HBM) in the platform diagram separately from DRAM bandwidth. 

Usability 

  • Intel VTune Profiler now allows for filtering kernels by name.  
    This is especially useful for applications that has many kernels, such as Gromacs, LAMPS etc.
  • GPU Adapter names will be replaced by short GPU Aliases.  
    • GPU Adapter names will be replaced by aliases in Summary, Grid and Timeline
    • Collection and Platform Info section will contain full names for GPU adapters, their BDFs and aliases.
    • The aliases will be stable: the same GPU adapters will have the same aliases for all results that are collected on the machine.
    • The alias should have the following format: "GPU 0", "GPU 1", "GPU N"

2023.1

Overview

This version of Intel® VTune™ Profiler contains improvements and additions in the following areas:

GPU Accelerators  

  • Support XeLink cross-card traffic profiling
    Get visibility into XeLink cross-card traffic for issues such as stack-to-stack traffic, throughput and bandwidth bottlenecks. Identify  imbalances of traffic between CPU and GPU through a GPU topology diagram.           

Debug Support 

  • Support for Large PDB Files 
    Starting with the 2023.1 version, on Windows* systems, you can resolve debug information from PDB files larger than 4GB. 

OS Support

  • Support Windows* 10 22H2 and Windows 11 22H2. 
  • Intel® VTune Profiler for macOS is now deprecated and will be discontinued in a future release. Learn other options to view results on macOS.

Hardware Support 

  • Support for New Architectures 
     VTune Profiler now supports the following generations of processors in client and server platforms: 
    • Intel® Xeon® Max Series CPUs (code named Sapphire Rapids HBM) 
      • Identify performance gained from HBM. Run Intel® VTune Profiler for each mode (HBM only, Flat, Cache) to identify which profile offers the best performance. 

2023.0.0

Overview

This version of Intel® VTune™ Profiler contains improvements and additions in the following areas:

GPU Accelerators

  • Stall Factor Information in GPU Profiling Results
    When you run the GPU Compute/Media Hotspots analysis to profile applications running on Intel® Data Center GPU Max Series (code named Ponte Vecchio) devices, you can now see the reasons for stalls in Xe Vector Engines (XVEs), formerly known as Execution Units (EUs). Use this information to better understand and resolve the stalls in your busiest computing tasks. For more information, see Analyze Xe Vector Engine (XVE) Stalls.
  • Metric Groups for Multiple GPUs
    When you run the GPU Compute/Media Hotspots analysis to profile an application executing on multiple Intel GPUs, you can now see metric information grouped by Intel microarchitecture family. See metrics for every GPU architecture family in a new consolidated view. To learn more, see Analysis Results for Multiple GPUs. 
  • Updated Metrics for Multiple GPUs
    GPU metric information in the Summary tab of the HPC Performance Characterization View have been enhanced to better represent data collected from multiple GPUs. 

Application Performance Snapshot

  • Updated Metrics for Multiple GPUs
    GPU metric information in the Application Performance Snapshot HTML reports have been enhanced to better represent data collected from multiple GPUs.

Language Support 

  • Support for New Language Versions
    This release introduces support for Python 3.9.0 in the Hotspots Analysis type for Windows and Linux systems. 

Platform Support 

  • Support for Legacy Processors
    VTune Profiler now supports the following generations of processors in client and server platforms: 
    • Server CPUs: Intel® Xeon® processor v3 and newer families.
    • Client CPUs: Intel® Core™ 4th generation processors and newer families. 

The 2023 version of VTune Profiler does not support processors older than the versions listed above. To analyze performance on older processors, use an older version of VTune Profiler. 

Hardware Support 

  • Support for New Architectures 
    VTune Profiler now supports the following generations of processors in client and server platforms: 
    • Fourth generation of Intel® Xeon® Scalable Processor (code named Sapphire Rapids) 
    • 13th generation of Intel® Core™ Processor (code named Raptor Lake) 
    • Intel® Data Center GPU Max Series (code named Ponte Vecchio) 
    • First generation of Intel® Arc™ High-performance Discrete GPUs (code named Alchemist). This support includes: 
      • Explicit support for SYCL, DirectX, Intel® Media SDK, OpenCL™, and OpenMP offload software technologies. 
      • Support for multi-GPU systems. You can now profile all Intel GPU devices, including integrated and discrete GPUs. 
      • Support for GPU Offload and GPU Hotspots analyses, including source level in-kernel profiling. 

Known Issues

Intel® VTune™ Profiler 2023.0.0 may not include all the latest functional and security updates. A new version of Intel® VTune™ Profiler is targeted to be released by March 2023 and will include additional functional and security updates. Customers should update to the latest version as it becomes available.

VTUNE PROFILER 2024.0 IS INCOMPATIBLE WITH glibc 2.37-13 ON FEDORA OS

VTUNE PROFILER 2024.0 PREVIOUSLY COLLECTED PLATFORM PROFILER RESULTS CANNOT BE OPENED WITH VTUNE GUI DUE TO SEPARATION FROM VTUNE PACKAGE

Results previously collected with VTune Platform Profiler (VPP) can no longer be opened in the VTune GUI as the platform profiler has been separated from the primary VTune product.

VTUNE PROFILER 2024.0 MODULEFILE INCORRECTLY DERIVES COMPONENT ROOT PATH AS "/"

The modulefile included with VTune Profiler 2024.0 incorrectly derives the component root path as "/". This file is located at:

<install-dir>/vtune/2024.0/etc/modulefiles/vtune/2024.0

Workarounds for this issue include:

1. Click here to download the fixed tcl file and replace the existing 2024.0 file. 

2. Instead of using "module load" to set up the environment variables, run:

$ source <install-dir>/vtune/latest/vtune-vars.sh

VTUNE PROFILER 2024.0 HBM AND TMA METRIC LIMITATION (DOC-11651)

The HBM Bound metric (in Microarchitecture Exploration, Memory Access, and HPC Performance Characterization) is only functional in HBM cache mode. The metric may not show a proper value in the HBM only and HBM flat modes.

VTUNE PROFILER 2023.2 KERNEL PANIC WHEN RUNNING HARDWARE-BASED SAMPLING COLLECTIONS ON FEDORA DISTRIBUTIONS RUNNING LINUX KERNEL VERSION 5.17 OR LATER (VASP-29437)

For some hardware-based analysis types, particularly Memory Access and Microarchitecture Exploration, VTune may cause a kernel panic on systems with Fedora distributions running Linux kernel version 5.17 or later. This has been fixed for the upcoming release. To work around this problem, unload the VTune sep driver by running:

$ sudo <VTune install dir>\latest\sepdk\src\rmmod-sep

This will prevent VTune from using the problematic driver and use Linux Perf instead.

 

VTUNE PROFILER 2023.2 GPU ANALYSIS MAY FAIL WITH INTEL DATA CENTER GPU MAX SERIES (CODE NAME PONTE VECCHIO)

With the recent GPU software stack (starting with agama-ci-devel-679 and newer) some VTune 2023.2 GPU analyses may fail with Intel® Data Center GPU Max Series (PVC). Affected analysis types are: GPU Offload analysis and GPU Compute\Media Hotspots analysis in "Compute Basic" mode.

Corresponding command lines for the analyses:
vtune -collect gpu-offload (default mode)
vtune -collect gpu-hotspots -knob characterization-mode=global-local-accesses (non-default mode)

The displayed error message would be the following:
vtune: Warning: Cannot start collection of GPU events
vtune: Collection failed
vtune: Internal Error

To avoid the failure, a user should remove or comment out lines #38 and #39 in VTune configuration script: <VTUNE_INSTALL_DIR>/config/analysis_type/include/gpu_metric_sets.xsl:

<item>HOST_TO_GPUMEM_BYTE_READ</item>
<item>HOST_TO_GPUMEM_BYTE_WRITE</item>

It is recommended to save the original script in the same directory and restart the VTune GUI if it's running on a target machine.

VTUNE PROFILER 2023.2 PLATFORM PROFILER DOESN'T SUPPORT 12th GEN INTEL CORE PROCESSORS (CODE NAMES ALDER LAKE AND RAPTOR LAKE) (VASP-29214)

VTUNE PROFILER 2023.2 IMPORTING PLATFORM PROFILER DATA FROM 8th GEN INTEL CORE PROCESSORS (CODE NAME COFFEE LAKE) FAILS WITH THE PLATFORM PROFILER SERVER (VASP-29198)

This is one known issue on 2023.2, may be fixed in future release.

  • Importing valid result data into the Platform Profiler server (both on Linux and on Windows 11) fails with no obvious error message to the user.
  • Server log contains the following message: Load failed on dataset. Dataset has no emon_results_metrics file and no formula file.

 

VTUNE PROFILER 2023.2 EMON DATA IS NOT PROPERLY LOADED ON HYBRID PLATFORMS (DOC-11343)

This is one known issue on 2023.2, will be fixed in future release. A workaround is available. 

  • VTune does not load EMON data properly from hybrid system where events are not applicable to both core types. 
  • The issue is currently observed on Windows OS only and not reproduced on Linux.
  • The Performance Snapshot and Microarchitecture Exploration analysis types in summary mode are impacted.

As a workaround, run the collection in the command line with the following option: --run-pass-through=--aggregated-counting

For example, to run the Performance Snapshot analysis on the matrix sample, run:

> vtune -collect performance-snapshot --run-pass-through=-aggregated-counting -- matrix.exe

 

VTUNE PROFILER 2023.2 MAX DRAM BANDWIDTH ESTIMATION MAY BE CALCULATED INCORRECTLY ON PLATFORMS WITH HBM (VASP-29158)

  • The estimation of max DRAM bandwidth value is incorrect on platforms with HBM in flat mode.

 

VTUNE PROFILER 2023.1 USER MODE SAMPLING FAILS WITH APPLICATION LINKED WITH GLIBC 2.35 (VASP-28157)

This is one known issue on 2023.1,  will be fixed in future release.

  • Hotspots and threading user mode sampling and memory access / consumption, and I/O analysis types with sep sampling will fail when application is linked with glibc 2.35 on Ubuntu 22.04. 
  • All cases using Perf and VTSS succeed with this application.
  • Other than the I/O, Memory Access, and Memory Consumption, all cases using SEP also succeed.

 

VTUNE PROFILER 2023.1 HARDWARE EVENT-BASED SAMPLING FAILS *PATCH AVAILABLE* (VASP-28956)

This is a known issue in the SEP driver packaged with update 2023.1, and will be fixed in a future release. The problem was reported on SLES15 SP4 but could happen on other Linux distros. A standalone SEP patch with the fix is available here: https://www.intel.com/content/www/us/en/developer/articles/code-sample/vtune-profiler-sampling-driver-downloads.html 

On certain workloads, the issue is seen after the 3rd or 4th runs, and on some systems it is observed in the 1st or 2nd run.

One of the following behaviors is observed randomly with each failure:

  • making the system go unresponsive and rebooting
  • system hangs in between
  • aborts the workload in the middle of collection and exits VTune

Common Limitations

RUNNING TIME IS ATTRIBUTED TO THE NEXT INSTRUCTION (VASP-3854)

To collect data about time-intensive running regions of the target, Intel® VTune™ Profiler interrupts the execution of target threads and attributes the time to the context IP address.

Due to the collection mechanism, the captured IP address points to the instruction after the most time-intensive one. This causes the running time to get attributed to the next instruction (or sometimes to subsequent instructions) in the Assembly view. In rare cases, this can also lead to incorrect attribution of running time in the source. The time can be erroneously attributed to the source line after the actual hot line.

If the inline mode is ON and the program has small functions inlined at the hotspots, this can cause the running time to be attributed to the wrong function since the next instruction can belong to a different function in tightly inlined code.

VTUNE CRASHES ON FREEBSD 11/12 AFTER PRESSING CTRL-C DURING COLLECTION (VASP-17676)
Upon termination of collection on FreeBSD with a Ctrl-C operation, the next VTune hardware event-based collection fails with this error- PMU resource(s) currently being used by another profiling tool or process: another PMU collection session may be running.

As a workaround, use -command stop to stop the collection.

COLLECTION MISSING DATA FOR CORE_0 ON SKYLAKE-X SYSTEMS (VASP-11480)
On some systems based on Intel microarchitecture code name Skylake-X with enabled SMT mode, VTune™ Profiler hardware sampling analysis can collect data against less logical CPUs than available on the system. For example, the problem was reproduced on Intel® Core™ i9-7900X 10C20T CPU installed on an Asus* PRIME X299-DELUXE motherboard.

As a workaround, upgrade the BIOS and the firmware of the board to 0802 or a newer version.

AN APPLICATION THAT ALLOCATES MASSIVE CHUNKS OF MEMORY MAY FAIL TO WORK UNDER VTUNE™ PROFILER
If a 32-bit application allocates massive chunks of memory (close to 2 GB) in the heap, it can fail to launch with VTune™ Profiler but execute independently without error. This happens because VTune™ Profiler requires additional memory when profiling an application. As a workaround, use a larger address space. For example, convert the project to 64-bit.

HARDWARE EVENT-BASED ANALYSIS MAY CRASH CERTAIN INTEL® CORE™ I7 PROCESSOR-BASED SYSTEMS WHEN DEEP SLEEP STATES ARE ENABLED
On some Intel® Core™ i7 processor-based (based on Intel microarchitecture code named Nehalem) systems with C-states enabled, sampling may cause system hanging due to a known hardware issue. To avoid this, disable the “Cn (ACPI Cn) report to OS” BIOS option before sampling with the VTune™ Profiler on Intel Core i7 processor-based systems.

INSTRUCTION SET REFERENCE DOCUMENT IS NOT POSITIONED ON DESCRIPTION OF PROPER INSTRUCTION
The reference information for assembly instructions can be opened in any PDF viewer, but only Adobe* Acrobat Reader* supports the positioning of the instruction reference document on the required page. To ensure correct functionality of this feature, install the latest version of Adobe Acrobat Reader.

VTUNE™ PROFILER DOES NOT SUPPORT PROFILING AN APPLICATION LAUNCHED UNDER DEBUGGER
The Hotspots, Threading, Memory Access with object tracking, and Input and Output analysis types provide incorrect results when analyzing an application launched under a debugger. VTune™ Profiler does not detect whether a debugger is attached to a profiled application. Make sure that no debugging tools are attached to the application that is profiled with the VTune™ Profiler.

ONLY ONE STACK FRAME IS SHOWN IN JAVA* CODE IF IBM* J9* JVM IS USED (VASP-3535)
Currently, IBM* J9* JVM does not support Java* stack unwinding.

RESULTS NOT COLLECTED FOR PROCESSES TERMINATED WITH TERMINATEPROCESS()
No results are displayed for a profiled process that is terminated by a different process calling TerminateProcess(). Use an alternate method to terminate the process.

DO NOT USE -IPO (LINUX*)//QIPO (WINDOWS*) OPTION WITH INTEL COMPILER VERSION EARLIER THAN 13.0. (VASP-3534)
If you are using a version of Intel® compiler that is 13.0 or older, use the additional option -inline-debug-info (Linux*)//debug:inline-debug-info (Windows*) to get performance data on inline functions. However, avoid using the /Qipo option. Currently this option disables the generation of the inline debug information in the compiler. Note that the Intel compiler integrated into the Microsoft Visual Studio* IDE uses the -ipo//Qipo by default in the Release configuration.

JAVA* SOURCE LINE MAY BE SHOWN INCORRECTLY FOR ORACLE JDK* 6U25 X64
Drilling down to Java* source code from VTune™ Profiler results may show incorrect source line. The issue occurs with Java* applications built with Oracle JDK* 6u25 x64. This is a JDK* problem. Refer to issues JDK-7051767 and JDK-7047624 in the Oracle bug database.

RESULT FINALIZATION MAY BE SLOW IF RESULTS ARE LOCATED IN A REMOTE NETWORK DIRECTORY (VASP-995)
The problem is caused by slow network access. To improve performance, use local directory for result storage.

CALL STACK CANNOT BE UNWOUND VIA NO RETURN FUNCTIONS (VASP-3350)
If the analyzed application contains functions without ret instruction (for example, calling exit()), unwinding the call stack to its caller and higher may fail (no stack frames shown).

ATTACHING THE VTUNE™ PROFILER TO A PROCESS MAY TAKE SOME TIME (VASP-2289)
Sometimes, your target application may complete execution before data collection begins. In this case, VTune™ Profiler may report an error. Increase the duration of your test application if necessary.

Command line option "vtune -command status" is currently supported only for user-mode sampling and tracing based analysis types, and not for EBS analysis types.

HARDWARE EVENT-BASED ANALYSIS MAY CAUSE UNPREDICTABLE SYSTEM BEHAVIOR ON PROCESSORS BASED ON INTEL MICROARCHITECTURE CODE NAMED SANDY BRIDGE
On processors based on Intel microarchitecture code name Sandy Bridge, hardware event-based sampling may cause unpredictable system behavior due to a known hardware issue (see erratum BK105 in http://www.intel.com/content/dam/www/public/us/en/documents/specification-updates/2nd-gen-core-family-mobile-specification-update.pdf ). To avoid this, avoid running Microarchitecture Exploration (formerly, General Exploration) or Custom Hardware event-based analysis with precise events on these systems unless a BIOS workaround for the erratum is provided.

VTUNE™ PROFILER MAY DETECT WHICH TIMER SOURCE TO USE INCORRECTLY ON INTEL® XEON® PROCESSOR E5-XXXX PROCESSORS
In User-mode Sampling and Tracing based analysis (Hotspots, Threading), the following message may be displayed in command line or Debug Window:

Warning: Cannot load data file `<path>\data.0\18385-18389.0.trace' (SampleCallback: timestamps aren't ascended!).
As a workaround, collect results from the command line using the -run-pass-thru option:

vtune -collect hotspots -run-pass-thru=-timestamp=sys -- <application>

HARDWARE EVENT-BASED ANALYSIS DOES NOT WORK IF MORE THAN 128 EVENTS ARE GATHERED SIMULTANEOUSLY (VASP-1992)
Decrease number of PMU events in analysis settings to resolve it.

ITT API TASK OR FRAME IS NOT SHOWN IN RESULTS IF ITS END OCCURS WHEN COLLECTOR IS INACTIVE (VASP-3354)
When ITT task or frame end notification occurs after collection was paused, or detach or stop command issued, the frame or task is not displayed in the result.

AN APPLICATION MAY EXPERIENCE THE STACK OVERFLOW EXCEPTION WHEN RUNNING UNDER THE VTUNE™ PROFILER (DOC-200)
An application allocating massive chunks of memory on a thread stack may experience the stack overflow exception and fail when running under the VTune™ Profiler, while running flawlessly on its own. This happens because VTune™ Profiler requires additional space on an application thread stack for profiling needs. To work around this problem, consider using larger thread stack space.

VTUNE™ PROFILER CANNOT LOAD DRIVERS INTO KERNEL
Most Android* kernels only allow signed drivers. If you try and install a driver (sep3_10.ko, pax.ko, vtsspp.ko, SOCWATCH1_3.ko or apwr3_1.ko), which was built after the kernel was built, you will get an error message stating that no signing key is found. To work around this problem, you must build drivers with the exact signing key created when the Android* OS (boot.bin) was built. Use drivers which have been integrated into the kernel.

STARTING PLATFORM PROFILER SERVER CAUSES SERVER CREDENTIALS ERROR: "DATABASE SERVER CREDENTIALS ARE INVALID OR CORRUPTED" (DXET-1990)
Platform Profiler server is unable to connect to the backend of the database due to an unexpected change in database credentials.

To work around this problem:

  • Verify that you have read/write access to the directory where Platform Profiler data is stored.
  • Verify that the Platform Profiler server is not running.
  • Open a Terminal window (Linux) or a Command Prompt window (Windows*) and type the following command:

           $ <vtune install dir>/vpp/server/vpp-server reset-db-credentials

  • Start the Platform Profiler server

DATA COLLECTED WITH PREVIOUS VERSIONS OF PLATFORM PROFILER IS NOT AVAILABLE AFTER UPGRADE TO VTUNE™ PROFILER 2020 (DXET-2014)
Platform Profiler 2020 allows users to choose the location for data storage. Previous versions of Platform Profiler stored the data in these locations:

  • Linux: /opt/intel/vtune-Profiler-platform-profiler/vpp-data
  • Windows: C:\ProgramData\Intel\VTune Profiler Platform Profiler\data

To view the data collected with previous versions of Platform Profiler:

  • Linux: Start the Platform Profiler server with the --data-dir option:

           <vtune install dir>/vpp/server/vpp-server start --data-dir /opt/intel/vtune-Profiler-platform-profiler/vpp-data/

  • Windows: In the Platform Profiler server startup dialog, click the "Browse Directory" button. Select "C:\ProgramData\Intel\VTune Profiler Platform Profiler\data" and then click "Start" to start the sever.

PLATFORM PROFILER IS UNABLE TO UPLOAD RESULTS OR VIEW RESULT DIRECTORIES AND FILES WITH GOOGLE CHROME* VERSION 80 OR LATER (DXET-2045)
Users of Google Chrome* version 80 or later are unable to browse Platform Profiler collected results or upload new result files. Clicking the “View Results” on the Platform Profiler home page displays an empty page.

To work around this problem:

  • In Google Chrome*,
    • Navigate to the “New Results” page.
    • Press F12 to open the Chrome Developer tools.
    • Click and hold the “refresh” toolbar button.
    • Select “Empty Cache and Hard Reload” from the menu.
  • Alternatively, you can also use Mozilla Firefox* or Microsoft* Edge. Platform Profiler is fully functional with Mozilla Firefox* and Microsoft* Edge. However, some some degradation in user experience (e.g. topology diagrams are not fully rendered) is likely.

SIZE LIMITATIONS WHEN UPLOADING DATA COLLECTION RESULTS TO THE PLATFORM PROFILER SERVER (DXET-1200)

The Platform Profiler server currently supports data collection results up to 1GB in size (compressed).

CPU TIME IN EXCESS OF 100% SHOWN FOR INDIVIDUAL THREADS (VASP-21841)
For data collected using the sampling driver on both Windows OS and Linux OS, the timeline may show CPU Time larger than 100% for individual threads, typically for the duration of one sampling interval at a time. This issue is caused by the thread migrating between logical CPUs.

As a workaround, you can switch to hardware sampling with stacks on both Windows OS and Linux OS or use driverless per-process sampling on Linux OS.

PYTHON CODE PROFILING LIMITATION: WHEN COLLECTING DATA IN ATTACH TO PROCESS MODE, PYTHON USER CODE IS NOT SHOWN IN VTUNE SOURCE VIEW FOR PYTHON* 3.6 VERSION AND HIGHER.
The limitation is valid starting from Python* 3.6.

APS TOOL FROM VTUNE PROFILER 2021 PACKAGE LOSES APPLICATION ARGUMENTS

You might observe a problem that Application Performance Snapshot (APS) tool loses application arguments wile starting a data collection.

The issue affects the following versions: Intel® VTune™ Profiler 2020 Update 3, Intel® VTune™ Profiler 2021 Beta Update 10, and future Intel® VTune™ Profiler 2021.

As a workaround you can create a small shell script where to place a command line of the application with parameters and then launch the script from APS tool.

VTUNE PROFILER’S GPU OFFLOAD OR GPU COMPUTE/MEDIA HOTSPOTS ANALYSIS MAY FAIL ON COLLECTING DATA AT ANDROID TARGET DEVICE (VASP-22213)
With GPU Offload or GPU Compute/Media Hotspots analysis types for a target Android device you need to disable the ‘Trace GPU programming APIs’ option. Otherwise the collection will fail with error: ‘Unknown option: --gpu-kernels-to-profile=*#1#1#4294967295’. With command line interface use the ‘collect-programming-api‘ knob to disable this option, for example:

vtune -collect gpu-offload -knob collect-programming-api=false …

CREATING OF MANY ITT API INSTANCES CAN LEAD TO ERROR REPORT THROUGH __ITT_REPORT_ERROR() AND/OR ABSENCE OF ITT API DATA.
To avoid this issue consider using latest ITT API from open source repository: https://github.com/intel-innersource/applications.analyzers.components.userapi

Windows* Specific Limitations

HW SAMPLING COLLECTION CONFLICT DUE TO WINDOWS DEFENDER VIRUS PROTECTION (DOC-10966)
The hardware sampling based collections (hotspots, uarch exploration, memory access, etc.) won't work correctly if Windows Defender Virus Protection is enabled as it is using the PMU counters for virus detection. In order to load the drivers, the user should also disable Windows Defender.

Steps to disable Windows Defender Real-time Virus Protection:

  1. Search for "Windows Security" and click to open the app.
  2. Click on "Virus & threat protection".
  3. Under the section, click the "Manage settings" option.
  4. Turn off the "Real-time protection" toggle switch to disable Microsoft Defender Antivirus.

Due to the security risk, users should disable Windows Defender at their own discretion.

VTUNE 2020 U1 SHOWS EMPTY WINDOW WHEN A NEW RESULT IS OPENED AFTER SYSTEM AWAKE (VASP-19132)
Open VTune 2020 U1 GUI, open a result, put system on sleep and login back. In VTune GUI, if a new result is opened then it shows an empty window. Workaround for this issue is to restart VTune GUI.

JIT WRITER API CANNOT HANDLE TOO LONG FILE NAMES ON WINDOWS* (VASP-484)
This issue happens when file name of JIT trace file is longer than 260 symbols. VTune™ Profiler is unable to generate a proper result when the result directory has such a long name.

THREAD STACK SIZE LIMITATION (VASP-1808)
VTune™ Profiler may crash with the following error message:
Error: failed to create a sampling thread: not enough storage is available to process this command.

This happens when the profiled application has big reserved and committed thread stack size (/STACK:reserve[,commit] command line switch of link.exe). The suggested workaround is reducing the reserved/committed thread stack size to profile the target.

INCORRECT TIMING RESULTS WHEN RUNNING ON A 32-BIT VIRTUAL MACHINE
VTune™ Profiler may fail to collect correct timing data when running on a virtual machine with problematic virtualization of time stamp counters. In this case the VTune™ Profiler throws a warning message:
Warning: Cannot load data file '<path_to_a_trace_file>.trace' (syncAcquiredHandler: timestamps aren't ascended!)

UNINSTALLING LIMITATION: PIN.EXE STAYS RUNNING AFTER DETACHING.
Once attached to a target, VTune™ Profiler cannot be uninstalled until execution on the target is over. The cause is that pin.exe keeps working after detaching from the target and exits only after the profiled application/process execution finishes.

SECOND ATTACH TO THE SAME APPLICATION SHOULD PRINT AN ERROR AND EXIT IMMEDIATELY
VTune™ Profiler allows you to run an analysis while the previous one is in progress. Howeverm it does not store any data from the second analysis run.

TRUNCATED .NET* MODULE NAMES MAY BE DISPLAYED IN RESULTS VIEW (VASP-3772)
When viewing results collected for a .NET application, you may observe truncated .NET module names. Ensure that the system was rebooted after installing the .NET application and before profiling with the VTune™ Profiler.

.NET MODULES LOADED DYNAMICALLY VIA REFLECTION API ARE DISPLAYED AS "UNKNOWN" HOTSPOTS (VASP-1945)
If dynamic .NET modules (e.g. loaded via Assembly.LoadFrom) are used in target application, VTune™ Profiler will show them as "unknown" functions and modules in hotspots list.

ATTACHMENT AS ADMINISTRATOR TO A PROCESS THAT LAUNCHED BY THE 'SYSTEM' USER FAILS (VASP-476)
You may use a utility from http://technet.microsoft.com/en-us/sysinternals/bb897553 to profile a system service (for example, w3wp.exe-based code) from the command line using vtune.exe that is located in <product_install_dir>/bin64. Do the following:

  1. Configure the w3wp service to run with the permissions you use to log in: Open IIS Manager, right-click an application pool you are using and set “Process model:Identity” to the account under which w3wp needs to be run.
  2. Run the w3wp service and make sure you run it with proper credentials and remember its PID.
  3. Start data collection: psexec -i 0 /path/to/vtune.exe -c=hotspots -r /path/for/your/data_dir --target-pid=PID
  4. Run your workload.
  5. Stop data collection: psexec -i 0 /path/to/vtune.exe -command detach -r /path/for/your/data_dir
  6. Open /path/for/your/data_dir in GUI: <product_install_dir>/bin64/vtune-gui /path/to/your/data_dir

LIMITED SUPPORT OF WINDOWS STORE* APPLICATIONS (VASP-531)
VTune™ Profiler supports attaching to Windows Store* applications, but not launching them. Only hardware event-based analysis types are supported.

VTUNE™ PROFILER ANALYSIS MAY FAIL ON MACHINE WITH VERDASYS DIGITAL GUARDIAN* SOFTWARE INSTALLED. (VASP-2701)
User mode sampling and tracing based analysis types (Hotspots, Threading) run from GUI may fail if Verdasys Digital Guardian* software is installed on the machine. However, it is still possible to run the collection from the command line.

You can also use EBS collection with stacks, either from the GUI or command line.

ANALYZED APPLICATION MAY CRASH DURING ANALYSIS IF "SYMANTEC ENDPOINT PROTECTION APPLICATION AND DEVICE CONTROL" SOFTWARE IS INSTALLED (VASP-2286)
The issue may occur with user mode sampling and tracing based analysis types (Hotspots, Threading). Ways to overcome the problem:

  • If user mode sampling and tracing based analysis is needed exclude the analyzed application from the list of monitored applications of "Symantec Endpoint Protection Application and Device Control" software.
  • Use Hotspots analysis with Hardware Event based Sampling and Collect stacks option turned on.

DOCUMENTATION CANNOT BE OPENED IN MICROSOFT INTERNET EXPLORER* 10 ON WINDOWS SERVER* 2012 (DOC-204)
To fix the error, modify security settings in Microsoft Internet Explorer* 10. From Tools > Internet Options > Security, add “about:internet” to the list of trusted sites.

CHILD PROCESS ANALYSIS DOES NOT WORK FOR EBS ANALYSIS TYPES IF PROCESS IS RUN FROM SCRIPT USING START COMMAND (DOC-231)
If a process is started from a batch file by the start command, it does not have original command interpreter as a parent process. So the new process cannot be identified as a child and not profiled. This is true for all Hardware event-based analyses. Possible workarounds:

  • Use user-mode sampling and tracing based analysis types instead
  • Use call command in a batch file instead of start
  • Use "attach to process" collection mode to connect directly to the process of interest

RESULTS MAY CONTAIN PROCESS NAME CUT ON THE 14TH CHARACTER (VASP-2702)
VTune™ Profiler may truncate a lengthy process name at the 14th character in results collected with EBS analysis types with stacks. For example, a result should contain find_hotspots.exe, but contains only find_hotspots.. This is an operating system limitation.

WARNING "UNSIGNED DRIVER" DURING INSTALLATION FOR SEPDRV AND VTSS DRIVERS ON WINDOWS* 7 AND WINDOWS SERVER* 2008 R2 (VASP-234)
VTune™ Profiler hardware event-based sampling drivers (sepdrv.sys and vtss.sys) are signed with digital SHA-2 certificate key for compliance with Windows 10 requirements.

To install the drivers on Windows* 7 and Windows Server* 2008 R2 operating systems, add functionality for the SHA-2 hashing algorithm to the systems by applying Microsoft Security Update 2033929: https://technet.microsoft.com/en-us/library/security/3033929.

STARTING PLATFORM PROFILER CAUSES ERROR: "THE SERVER WAS STARTED FROM AN UNEXPECTED LOCATION" (DXET-2063)
When starting the Platform Profiler server, the server may fail to start and report the following error:

“The server was started from an unexpected location. Please stop this server instance before trying to start the server again”.

This error can happen when the server was previously started from the Windows Command Prompt using the "vpp-server start" command.

To work around this problem:

  • Stop these processes:
    • influx
    • <vtune install dir>/vpp/server/dist/3rdparty/vpp-python/python.exe
  • Restart the Platform Profiler server.


Note that starting the Platform Profiler server from the VTune user interface does not result in this error.

LONG FINALIZATION TIME IF WINDOWS BINARIES ARE BUILT WITH THE “/OPT:ICF” LINKER OPTION (VASP-21607)
Debugging information generated for large (Several hundreds of MBs) binaries built with the “/OPT:ICF” option can contain many overlapping symbols. That can cause two following problems:

  • Long finalization of VTune results (because VTune is not optimized to handle such symbols);
  • Impossibility to distinguish certain symbols because they have the same address range in debug info.

A workaround is to remove the “/OPT:ICF” from linker options.

VTUNE™ PROFILER 2021 TOOLBAR BY DEFAULT IS HIDDEN IN MICROSOFT* VISUAL STUDIO 2019 IDE (VASP-22133)
To enable the toolbar right-click on the Visual Studio toolbar area and check “Intel VTune Profiler”. The VTune Profiler toolbar will appear, but unfortunately, the extension toolbar needs to be enabled on each launch of the IDE.

Alternatively, you can launch VTune Profiler using Menu->Tools->Intel VTune Profiler or from a project context menu.

VTUNE™ PROFILER 2021 IS NOT SUPPORTING A REMOTE COLLECTION WITH VIA PSEXEC, SSH, POWERSHELL UTILITIES ARE NOT SUPPORTED ON WINDOWS PLATFORMS ( VASP-22269)
As a workaround, you can use a Windows Remote Desktop session and collect VTune profiles as on a local machine.

VTUNE™ PROFILER IS NOT ABLE TO RESOLVE DEBUG INFORMATION WHEN AN APPLICATION IS COMPILED WITH THE /DEBUG:FASTLINK OPTION. (VASP-3093)

SOMETIMES VTUNE PROFILER HANGS ON PROCESSING .PDB FILES BUILT WITH /DEBUG:FASTLINK OPTION
To fix this issue, please use the /DEBUG:FULL option instead. For more information, see https://docs.microsoft.com/en-us/cpp/build/reference/debug-generate-debug-info

VTUNE™ PROFILER IS REPORTING AN ERROR WHEN CONFIGURING AN ANALYSIS WITH A PROJECT CREATED OR RESULTS STORED IN A BOX DRIVE FOLDER (E.G. C:\USERS\<USER NAME>\BOX ON WINDOWS) (VASP-22392)
A discussion of the problem with Box Drive can be found here: https://support.box.com/hc/en-us/community/posts/1500001146222-Directory-to-Box-drive-is-an-empty-symlink-

A workaround would be to place a project and result files in local folders or other shared drives.

A DPC++ APPLICATION COMPILED WITH OPTIONS THAT PRESERVE DEBUG INFORMATION MAY BE SIGNIFICANTLY SLOWED DOWN WHILE RUNNING ON CPU
To work around this issue, explicitly set an environment variable CL_CONFIG_CPU_NO_DBG_CBK=1 before running the application or profiling it with VTune Profiler.

VTUNE PROFILER DOES NOT SUPPORT DEBUG INFORMATION FOR INLINE FUNCTIONS ON WINDOWS (VASP-24541)
VTune Profiler is unable to read debug information for inline functions when it contains in a PDB file. Consequently, when you profile applications on Windows systems, you may not see source line information for inline functions.

To work around this issue, before you profile an application with VTune Profiler, set the

CL_CONFIG_USE_NATIVE_DEBUGGER environment variable to 0

(CL_CONFIG_USE_NATIVE_DEBUGGER=0)

Linux* Specific Limitations

RISK OF DISABLING INDIRECT BRANCH TRACKING(IBT) FOR VTUNE DRIVER MODE COLLECTION (DOC-10643)
The Indirect Branch Tracking hardware feature is momentarily disabled (while loading the drivers) in VTune driver collection mode if the feature is supported and enabled in the hardware and Linux operating system. There is no security impact expected during performance profiling after the drivers are loaded.

CORE DUMP WITH SEP 5.38 ON SOME FreeBSD SERVER CONFIGURATIONS WITH FreeBSD-12.2 AND FreeBSD-12.2-p14 (SEP-5505)

This issue may be caused by sep driver installation failed.

If “kldload sep pax” fails, retry with “kldload sep/sep.ko pax/pax.ko”. 

If “vtune-vars.sh” configures PATH and LD_LIBRARY_PATH to use 32-bit pathnames on a 64-bit system, run “uname -m” and if the output is “amd64”, modify vtune-vars.sh to compare the “uname -m” output from “x86_64” to “amd64” and rerun the “vtune-vars.sh” script.

SOME FEATURES REQUIRE INSTALLING AS ROOT
Most of Intel® VTune™ Profiler profiling features work with a non-root install. Many features work on either a genuine Intel processor or a compatible processor.

Some advanced features that use event-based sampling require the latest OS kernel or sampling driver to be installed. Intel Atom® processors also require this driver for analysis.

To install the driver on a system with a genuine Intel processor, launch the installer as root or ask your system administrator to install the driver later. For information on building and setting up the drivers, see https://www.intel.com/content/www/us/en/develop/documentation/vtune-help/top/installation/sep-driver.html.

SLES 15 MACHINE REBOOTS WHEN STARTING INPUT/OUTPUT ANALYSIS WITH BOTH PERF/SEP (VASP-14205)
When running Input and Output analysis on SUSE* Enterprise Linux (SLES) 15, consider switching to Custom Analysis and turn off page faults collection. This helps prevent a system crash that may happen when kernel trace points for page faults are turned on.

PROXY-RELATED ERRORS WHEN STARTING THE PLATFORM PROFILER SERVER (DXET-1338)
On some systems, starting the server generates the following error:

Malformed Proxy variables. Please set your proxy variables with "http" scheme and restart the PPE server :[Errno socket error] [Errno 10060] A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond

To work around this problem:

  1. Remove the following proxy-related environment variables and start the server again: http_proxy, https_proxy, noproxy
  2. Start the server using the following commands:

           $ cd <vtune install directory>/vpp/server

           $ dist/3rdparty/vpp-python/python -m scripts.start --skip-proxy-check

SECURITY-ENHANCED LINUX* IS NOT SUPPORTED
Security-enhanced Linux settings (SELinux) are currently not supported by the VTune™ Profiler and need to be either disabled or set to permissive for a successful tool suite installation. If your Linux distribution has SELinux enabled the following error message will be issued by the installer:

Your system is protected with Security-enhanced Linux (SELinux). We currently support only "Permissive" mode, which is not found on the system. To rectify this issue, you may either disable SELinux by - setting the line "SELINUX=disabled" in your /etc/sysconfig/selinux file - adding "selinux=0" kernel argument in lilo.conf or grub.conf files or make SELinux mode adjustment by - setting the line "SELINUX=permissive" in your /etc/sysconfig/selinux file or ask your system administrator to make SELinux mode adjustment. You may need to reboot your system after changing the system parameters. More information about SELinux can be found at http://www.nsa.gov/selinux/

VTUNE™ PROFILER USER MODE COLLECTION MAY FAIL TO RUN ON UBUNTU* (VASP-3352)
VTune™ Profiler may fail to collect data for Hotspots and Threading analysis types on the Ubuntu operating system. Once a collection is started, this message appears in the output:

Failed to start profiling because the scope of ptrace() system call application is limited. To enable profiling, please set /proc/sys/kernel/yama/ptrace_scope to 0. See the Release Notes for instructions on enabling it permanently.

To work around this problem for the current session, set contents of the /proc/sys/kernel/yama/ptrace_scope sysctl file to 0.

To make this change permanent, set kernel.yama.ptrace_scope value to 0 at /etc/sysctl.d/10-ptrace.conf file using root permissions and reboot the machine.

THREADING ANALYSIS MAY PROVIDE INCORRECT WAIT TIME DATA IF LINUX FUTEXES ARE USED
In certain cases, custom synchronization constructions based on Linux futexes are not detected by the VTune™ Profiler. System libraries may also use such constructions. To ensure Wait time correctness in the Threading analysis, use syscall(SYS_futex, ...) API in your code.

VTUNE™ PROFILER MAY CAUSE KVM GUEST VIRTUAL MACHINE TO CRASH (VASP-3806)
In some cases, when you run Event-Based Sampling analysis on a host machine, a guest KVM virtual machine may experience kernel panic. This happens only if precise events are used, for example, with Microarchitecture Exploration (formerly, General Exploration) analysis type. Upgrade KVM to a newer version; there is a KVM patch that fixes the problem.

SYMBOLS ARE NOT RESOLVED FOR BINARIES REMAPPED WITH HUGTLBFS UTILITY (VASP-707)
hugetlbfs remaps segments from executable file to its temporary file system or to anonymous memory regions. VTune™ Profiler can’t map those segments to the original executable.

The workaround for this problem is to rename the original executable to the one that hugetlb exposes, i.e. the module name that you see in VTune™ Profiler results.

my_exe (original executable) → my_exe_32_2 (file name on hugetlb file system in case of sharing between processes)

my_exe (original executable) → libhugetlbfs.tmp.<random hash> (memory region name in case of anonymous mappings)

RUNNING OTHER PMU-BASED TOOLS SIMULTANEOUSLY WITH HARDWARE EVENT-BASED SAMPLING COLLECTION IS NOT SUPPORTED
The Sampling Driver requires exclusive use of the PMU resources. Running multiple PMU based tools will cause incorrect results and/or system instability.

HOTSPOTS AND THREADING ANALYSIS TYPES MAY NOT WORK ON EXECUTABLES THAT DO NOT DEPEND ON THE LIBPTHREAD.SO.0 LIBRARY
There is currently a limitation in the product regarding profiling application targets where the executable does not depend on the libpthread.so.0 library. The message "Link libpthread.so to the application statically and restart profiling" appears when profiling an application where program image does not depend on libpthread.so.0 but then it dlopen()s a shared library which does depend on libpthread.so.0. The collector is not able to follow the program execution and module load/unload so the collection results are likely to be misleading. A workaround is to set "LD_PRELOAD=libpthread.so.0" before running the collection.

OPENING RESULTS SIGNIFICANTLY DELAYED ON LINUX SYSTEMS CONFIGURED WITH "LARGE PAGES"
To work around this issue make sure that you have hugectl utility and libhugetlbfs library on the system. VTune™ Profiler can be started this way:
hugectrl --heap <vtune install dir>/bin64/vtune-gui
LD_PRELOAD=libhugetlbfs.so <vtune install dir>/bin64/vtune-gui

HELP TOC IS TRUNCATED IN SOME LINUX BROWSERS
When using certain HTML browsers to view the VTune™ Profiler offline documentation, if the Contents tab width appears truncated and hides some of the topic titles:

  • Click the Index (or Search) button
  • Click the Contents button

COLLECTION FAILS WITH ERROR "NO CONFIG_PERF_EVENTS=Y KERNEL SUPPORT CONFIGURED" (DOC-241)

Driverless EBS collection on Linux can fail with the error: "No CONFIG_PERF_EVENTS=y kernel support configured" even if the kernel config has been set. This is a known Linux kernel issue that is resolved in kernel version 4.1. There is a simple patch that provides a workaround: http://halobates.de/ivb-allow-mem-load-uops

USER TASKS, EVENTS AND FRAMES MARKED WITH ITT APIS MAY NOT BE SHOWN IF TARGET APPLICATION IS FINISHED BY CTRL+C (VASP-1367)
Use another method to stop application, or VTune™ Profiler stop command.

WHEN INSTALLING ON A XEN* VIRTUAL MACHINE, IT MAY HAPPEN THAT INSTALLATION SCRIPT EXITS CLAIMING CPU IS UNSUPPORTED
To skip the check for CPU model, add --ignore-cpu parameter to install script command line:
./install.sh --ignore-cpu

VTUNE™ PROFILER HELP DOES NOT OPEN FROM GUI ON SOME LINUX SYSTEMS (VASP-3322)
On some Linux* systems, like Wind River Linux, default ways to open web browser may not work. You may export the environment variable BROWSER (before starting VTune™ Profiler) with the path to the web browser to get VTune™ Profiler help working.

For example:
$ export BROWSER=/usr/bin/firefox

VTUNE™ PROFILER USER APIS ARE NOT SUPPORTED FOR STATIC BINARIES (VASP-2287)
To use ITT and JIT APIs with software-based collectors (Hotspots, Threading), you must link their application dynamically. Otherwise ITT and JIT API notifications will be missing in the result.

VTUNE™ PROFILER TARGET EXECUTABLES DO NOT WORK ON YOCTO PROJECT X64 DUE TO DIFFERENT PATHS TO LD (VASP-8531)
Error message: "-sh: ./amplxe-perrunss: No such file or directory". Usually ld is located in /lib64/ld-linux-x86-64.so.2 but on Yocto x64 it is located in /lib/ld-linux-x86-64.so.2.

The workaround is to create /lib64/ld-linux-x86-64.so.2 as a symlink to /lib/ld-linux-x86-64.so.2.

EVENT-BASED SAMPLING DRIVER FAILS TO BUILD ON YOCTO PROJECT 1.6 INTEL X86_64 BSPS DUE TO MISSING CONFIG_COMPAT #DEFINE IN LINUX KERNEL HEADERS (VASP-1558)
Sampling driver fails to build on Yocto x86_64 BSPs images that do not have “CONFIG_COMPAT=y” setting (https://bugzilla.yoctoproject.org/show_bug.cgi?id=6777)

To resolve the issue, you need to re-build the image with the option manually.

Change kernel configuration according to http://www.yoctoproject.org/docs/1.6.1/kernel-dev/kernel-dev.html#changing-the-configuration :

  1. Create the compat.cfg file in meta-yocto-bsp/recipes-kernel/linux/files/ directory with content: CONFIG_IA32_EMULATION=y CONFIG_COMPAT=y
  2. Add the following strings to meta-yocto-bsp/recipes-kernel/linux/linux-yocto_3.10.bbappend: FILESEXTRAPATHS_prepend := "${THISDIR}/files:" SRC_URI += file://compat.cfg
  3. Clean and build a modified kernel according to http://www.yoctoproject.org/docs/1.6.1/dev-manual/dev-manual.html#patching-the-kernel and http://www.yoctoproject.org/docs/1.6.1/dev-manual/dev-manual.html#build-the-modified-qemu-kernel-image

           $ bitbake linux-yocto -c cleansstate

           $ bitbake -k linux-yocto

  1. Build image:

           $ bitbake core-image-sato

COLLECTING PERFORMANCE DATA FOR INTEL® OPTANE™ PERSISTENT MEMORY (DXET-1844)
If you are using an account with root privileges and you try to collect data on a system enabled with Intel® Optane™ Persistent Memory, this warning appears:

User needs sudo privilege to collect Persistent Memory metrics. Persistent Memory metrics will not be collected.
Use the following command to start data collection with elevated privileges:
sudo <vtune install dir>/vpp/collector/vpp-collect start

Use the following command to stop data collection:
sudo <vtune install dir>/vpp/collector/vpp-collect stop

NOTE: elevated privileges are required to stop a data collection session that was started with elevated privileges.

VTUNE COLLECTION CAN SOMETIMES FAIL REPORTING LIBSTDC++.SO.6 LIBRARY NOT FOUND (VASP-21417)
Due to a known Linux loader bug, VTune collection can sometimes fail when a user had libraries in LD_PRELOAD environment prior to launching VTune collection. The example of how the bug can manifest when having libzet_intercept.so in LD_PRELOAD:

/opt/intel/vtune_profiler_2021.1.0/bin64/pin: /opt/intel/vtune_profiler_2021.1.0/bin64/../lib64/libstdc++.so.6: version `GLIBCXX_3.4.26' not found (required by /lib/x86_64-linux-gnu/libze_loader.so.1)

There are several workarounds for this problem:

  1. Clear LD_PRELOAD before launching VTune and configure LD_PRELOAD as an environment for the target application if it is required.
  2. Add libstdc++.so.6 either from VTune installation or from system location to LD_PRELOAD (see https://www.intel.com/content/www/us/en/develop/documentation/vtune-help/top/troubleshooting/error-message-cannot-collect-gpu-hardware-metrics.html that has a similar workaround)
  3. Do 'export LD_PROFILE=none' before launching collection (this could cause an insignificant performance hit for the loader, does not work in security hardened environment)

Any one of these workarounds should solve the problem.

USING A WRAPPER SCRIPT IN VTUNE PROFILER COMMAND MIGHT NOT WORK PROPERLY IF “SUDO -A SU <USER> -C “$*” IS USED IN THE SCRIPT FOR LOCAL CONNECTION. (VASP-22952)
Correct “su” commands are the following:

The command "sudo -A su <user> -c "$*" " works for Remote Linux (SSH) collection.

The command "sudo -A -C 65000 su <user> -c "$*" " works for local (or VTune Server) collection.

UNABLE TO RUN PLATFORM PROFILER ON FEDORA VERSIONS 30 OR NEWER(DXET-2342)
On Fedora versions 30 or newer, Platform Profiler can fail to start. This error message appears:
<vtune-install-dir>/vpp/server/dist/3rdparty/vpp-python/bin/python3: error while loading shared libraries: libcrypt.so.1: cannot open shared object file: No such file or directory Failed to configure Intel(R) VTune(TM) Profiler - Platform Profiler server. Intel(R) VTune(TM) Profiler - Platform Profiler is supported on Ubuntu, CentOS, Debian and RedHat based systems only.

To work around this problem, install the libxcrypt-compat package using the Fedora package manager.

PLATFORM PROFILER IS UNABLE TO COLLECT PERSISTENT MEMORY METRICS FROM SYSTEMS WITH INTEL® OPTANE™ PERSISTENT MEMORY 200 SERIES MODULES (FORMERLY BARLOW PASS) (DXET-2330)
When collecting Platform Profiler data from systems with Intel(R) Optane(TM) Persistent Memory 200 Series Modules (formerly Barlow Pass), you may see this warning:

WARNING: PMWatch was not found. Persistent Memory metrics will not be collected. For more details, see ~/vpp-collect.log" displayed

The VTune error log file contains more information:

WARNING: ipmwatch pre-requisites: Firmware version >= 01.00.00.5142 for PMON functionality 01.00.00.5127 for rest of the functionality ipmctl >= 01.00.00.3296 for PMON functionality 01.00.00.3262 for rest of the functionality ipmctl found on the system: 02.00.00.3457

ipmwatch is not guaranteed to work with versions less than the ones mentioned above.

To work around this issue,

  1. Identify the versions of libipmctl installed on the system:

          $ ls -la /lib64 | grep ipmctl

            lrwxrwxrwx 1 root root 18 Dec 18 12:59 libipmctl.so.3 -> libipmctl.so.3.1.0

            -rwxr-xr-x 1 root root 2525496 Feb 28 2020 libipmctl.so.3.1.0

            lrwxrwxrwx 1 root root 18 May 1 2020 libipmctl.so.4 -> libipmctl.so.4.0.0

            -rwxr-xr-x 1 root root 2562712 May 1 2020 libipmctl.so.4.0.0

  1. Update the link libipmctl.so.3 to point to libipmctl.so.4.0.0

           $ unlink /lib64/libipmctl.so.3

           $ ln -s /lib64/libipmctl.so.4.0.0 /lib64/libipmctl.so.3

           $ ls -l /lib64/ | grep ipmctl

             lrwxrwxrwx 1 root root 25 Apr 22 15:47 libipmctl.so.3 ->/lib64/libipmctl.so.4.0.0

             -rwxr-xr-x 1 root root 2525496 Feb 28 2020 libipmctl.so.3.1.0

             lrwxrwxrwx 1 root root 18 May 1 2020 libipmctl.so.4 -> libipmctl.so.4.0.0

             -rwxr-xr-x 1 root root 2562712 May 1 2020 libipmctl.so.4.0.0

DRIVERLESS PERF BASED COLLECTIONS ON ADL IS SUPPORTED BY LINUX KERNEL SINCE 5.15.5. (VASP-27934)
Driverless Perf based collections on ADL is supported by Linux kernel since 5.15.5. To check if your kernel supports driverless collection on ADL architecture, check for the following device files:
/sys/bus/event_source/devices/cpu_atom
/sys/bus/event_source/devices/cpu_core

macOS* Specific Limitations

NATIVE COLLECTION IS NOT AVAILABLE ON MACOS*
Intel VTune™ Profiler for macOS* supports viewing of results collected on other OSs. Native collection on macOS is not currently available.

SOME SYSTEMS DO NOT SUPPORT CONTEXT-SENSITIVE HELP
On some macOS* systems, such as macOS 10.13, VTune™ Profiler context-sensitive help is not supported.

Previous VTune™ Profiler Releases

VTune Profiler 2022 and Previous Releases:

Attributions

The license information for third party software that was used to develop Intel® VTune™ Profiler can be found in <install_dir>/licensing/third-party-programs.txt .
For the avoidance of doubt, the Intel® VTune™ Profiler is solely governed by the terms and conditions of the End User License Agreement for Intel® Software Development Product that accompanies the Intel® VTune™ Profiler.

Legal Information

No license (express or implied, by estoppel or otherwise) to any intellectual property rights is granted by this document.

Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.

This document contains information on products, services and/or processes in development. All information provided here is subject to change without notice. Contact your Intel representative to obtain the latest forecast, schedule, specifications and roadmaps.

The products and services described may contain defects or errors which may cause deviations from published specifications. Current characterized errata are available on request. No product or component can be absolutely secure.

Copies of documents which have an order number and are referenced in this document may be obtained by calling 1-800-548-4725 or by visiting www.intel.com/design/literature.html.

Intel, the Intel logo, Intel Core, Intel Xeon Phi, VTune, and Xeon are trademarks of Intel Corporation in the U.S. and/or other countries.

*Other names and brands may be claimed as the property of others.

Microsoft, Windows, and the Windows logo are trademarks, or registered trademarks of Microsoft Corporation in the United States and/or other countries.

Java is a registered trademark of Oracle and/or its affiliates.

OpenCL and the OpenCL logo are trademarks of Apple Inc. used by permission by Khronos.


Copyright 2010 - 2022 Intel Corporation.

This software and the related documents are Intel copyrighted materials, and your use of them is governed by the express license under which they were provided to you (License). Unless the License provides otherwise, you may not use, modify, copy, publish, distribute, disclose or transmit this software or the related documents without Intel's prior written permission.

This software and the related documents are provided as is, with no express or implied warranties, other than those that are expressly stated in the License.