<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Performance Engineering :: Category :: Documentation for HPC</title><link>https://docs.hpc.gwdg.de/categories/performance-engineering/index.html</link><description/><generator>Hugo</generator><language>en</language><atom:link href="https://docs.hpc.gwdg.de/categories/performance-engineering/index.xml" rel="self" type="application/rss+xml"/><item><title>NVTOP &amp; Nsight</title><link>https://docs.hpc.gwdg.de/software_stacks/applications/nsight-systems/index.html</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://docs.hpc.gwdg.de/software_stacks/applications/nsight-systems/index.html</guid><description>NVTOP In HPC environments, GPU-intensive workloads often require real-time monitoring of GPU utilization. NVTOP (Neat Videocard TOP) is a Linux command-line tool, based on htop, that interactively reports GPU usage. On the cluster, the spider command can be used to check available versions. Loading nvtop also requires gcc/13.2.0 or gcc/13.2.0-nvptx:
module load gcc/13.2.0 module load nvtop Info NVTOP monitors GPU activity on compute nodes, so that it must be run on a node equipped with an active GPU. For information on allocating GPU nodes, refer to the cluster’s GPU usage documentation.</description></item><item><title>Score-P</title><link>https://docs.hpc.gwdg.de/software_stacks/applications/score-p/index.html</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://docs.hpc.gwdg.de/software_stacks/applications/score-p/index.html</guid><description>Scalable Performance Measurement Infrastructure for Parallel Codes (Score-P) is a performance-measurement framework for C, C++, and Fortran applications. It targets applications that use multi-processing (MPI, SHMEM), thread parallelism (OpenMP, Pthreads), or accelerator models (HIP, CUDA, OpenCL, OpenACC, OpenMP-offload), as well as combinations of these. A Python interface is also available for tracing and profiling Python applications.
Score-P currently supports only MPI or SHMEM for multiprocessing. Other multiprocessing approaches (e.g. Torch distributed methods) cannot be traced. It also does not support importlib.reload().</description></item><item><title>Vampir</title><link>https://docs.hpc.gwdg.de/software_stacks/applications/vampir/index.html</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://docs.hpc.gwdg.de/software_stacks/applications/vampir/index.html</guid><description>Vampir is a visualization tool for handling and analyzing instrumented and sampled event traces generated by Score-P. It provides powerful performance-analysis capabilities for highly parallel applications.
Vampir is available as a commercial product since 1996, requiring purchase for “standard” or “professional” licenses directly from vampir.eu. However, it is usable as a module on the cluster of the GWDG.
To use Vampir remotely, X11 forwarding must be enabled when connecting to a node:</description></item></channel></rss>