Application Scenarios

Build file

• The Find function in CMake parsing depends on the GCC compilation result. If the tool runs in a Kunpeng environment, CMake cannot successfully parse or detect x86-specific compilation options such as -m64.
• Makefile supports the shell commands in $(shell shell_command) format, but does not support the build configuration files using shell scripts.

C/C++ code analysis

If the custom macro header files, third-party header files, or some system header files are missing, some macros to be modified may not be identified and accurate modification suggestions cannot be provided.

Intrinsic functions

The tool cannot provide accurate replacement suggestions for more than 130 intrinsic functions related to the Intel coprocessors.

Assembly code

• The tool cannot provide absolutely accurate modification suggestions in any of the following scenarios:
  • When a function contains parameters, a message will be displayed indicating that the function conversion may fail due to a special structure.
  • When parameters are transferred to a stack, the parameter type is integer by default.
• Manual conversion is prompted in any of the following scenarios:
  • The goto, call, and ret statements are used.
  • A segment register or a physical register without a value is used.
  • Global symbols are accessed.
  • The bit width of the input and output variables is greater than 128 bits.
  • The number of input and output global variables is greater than 21.
  • The bit width of instructions is different from that of variables.
  • The hybrid Intel and AT&T assembly format is used.
  • A syntax error exists in the assembly.
    NOTE:

    For details about the cases, see Exception Scenarios and Common Syntax Errors in Inline Assembly.

Byte alignment

If dependency files are unavailable when header files are used in recursion, the analysis accuracy decreases.

Comparison analysis

Currently the supported analysis tasks are overall analysis and hotspot function analysis.

HPC cluster check

Password-free login must be enabled for each node in the MPI cluster. For configuration items that do not comply with the best practice of the Kunpeng platform, the tool provides optimization suggestions.

HPC application analysis

During OpenMP data collection, the kernel parameters /proc/sys/kernel/kptr_restrict and /proc/sys/kernel/perf_event_paranoid are enabled to collect call graph data and PMU events. After the collection is complete, the two kernel parameters are restored to their original values.

Microarchitecture analysis

You must have the root permission to perform the following operations.

1. If the configuration of Paranoid is incorrect, set the Paranoid variable to -1. For example, in CentOS and openEuler, run the following command:

   1	echo -1 > /proc/sys/kernel/perf_event_paranoid

2. If a message is displayed indicating that data collection fails and the OS performance monitor is not enabled, run the following command to enable it:

   1	echo 0 > /proc/sys/kernel/nmi_watchdog

Memory access analysis

This function is available on openEuler and CentOS 7.6 with the Statistical Profiling Extension (SPE) feature. The supported openEuler kernel versions are 4.19 and later and the supported CentOS 7.6 kernel versions are 4.14.0-115.el7a.0.1, 4.14.0-115.2.2.el7a, 4.14.0-115.5.1.el7a, 4.14.0-115.6.1.el7a, 4.14.0-115.7.1.el7a, 4.14.0-115.8.2.el7a, and 4.14.0-115.10.1.el7a.

• Miss event analysis

  This function is not available in VM and container environments.

• False sharing analysis

  This function is not available in VM environments.

• NUMA refined analysis

  This function is not available in VM environments.

I/O analysis

The system kernel must support ftrace collection.

Process/Thread performance analysis

If the sysstat version is earlier than 11.0.0, the process/thread performance analysis result may contain abnormal data.

Resource scheduling analysis

You are advised to use an OS whose kernel version is 4.19 or later to run resource scheduling analysis tasks.

Roofline analysis

Set /proc/sys/kernel/perf_event_paranoid to 0 or a smaller value.

Lock and wait analysis

The environment must support the extended Berkeley Packet Filter (eBPF) configuration.

Real-time profiling

• The JDK version must be JDK 8 or later.
• If the server does not support hotspot analysis, the Create Hotspot Analysis button is unavailable.
• Currently, Concurrent Mark Sweep (CMS) and Garbage-First (G1) garbage collector logs can be parsed. Java 8 supports CMS and G1, whereas Java 9 and later versions support only G1.
• Each user can perform real-time profiling on a maximum of 10 processes at the same time.

Sampling profiling

The JDK must be BiSheng JDK 8.302 or later or OpenJDK 9 or later.

Memory usage diagnosis

• This function is not available in container environments.
• The environment must support the eBPF configuration.

Memory overwriting diagnosis

The application must be compiled using GCC 4.9 or later.

Network I/O diagnosis

The OS must have a kernel version ranging from 4.15.0 to 5.4.0, and UOS 20 is not supported.

Storage I/O diagnosis

Do not use logical disks or existing files for pressure tests.