Performance best practices
Power Systems performance claims and proof points
Built to scale data-intensive workloads and optimized for performance, IBM Power Systems deliver superior price-performance over x86 competitors.
Benchmarking linear models of machine learning (ML) frameworks – Snap ML versus cuML
Machine learning (ML) has been defined as the field of study that gives computers the capability to learn without being explicitly programmed. This blog shows the speedup of Snap ML in comparison with cuML for the Epsilon, Higgs, Taxi and Price Prediction preprocessed data sets.
DeepLabv3+ image segmentation model with TFLMSv2
This blog showcases the advantages of using IBM’s WMLCE 1.6.1 TensorFlow Large Model Support (TFLMS) on DeepLabv3+ model and performs a competitive comparison to highlight IBM POWER9 processor’s NVLink 2.0 advantages while training large neural networks.
DeepLabv3+ image segmentation model with PyTorch LMS
PyTorch with IBM’s WML-CE 1.6.1 comes with LMS to enable large PyTorch models and in this blog, we capture the benefits of using PyTorch LMS on DeepLabv3+ along with the PASCAL Visual Object Classes (VOC) 2012 data set.
A guide to GATK4 best practice pipeline performance and optimization on the IBM OpenPOWER system
GATK4 is an open source toolkit frequently used by most genomic research and clinical analyses. The high-performance data and analytics (HPDA) solution, based on IBM® OpenPOWER and IBM Spectrum® computing, dramatically accelerates the analysis workloads.
Analyzing performance using perf annotate
In this white paper, we compare the performance of an OpenPOWER-based cluster containing the IBM Power Systems S812LC and a cluster containing x86 systems.
Best practices for Java and IBM WebSphere Application Server (WAS) on IBM POWER9
This article describes best practices to achieve performance from applications running in the Liberty profile of IBM WebSphere Application Server (WAS) on IBM Power System S9xx and L922 systems. Acme Air, running in IBM Cloud Private is used as a case study to demonstrate the benefits and the application of the best practices.
Low latency tuning tips for Linux on Power
This article is intended to document the “state of the art” for low latency tuning, specifically targeted at Linux running on IBM Power Systems.
Application tuning tips for Linux on Power
This article is intended to document the “state of the art” for application tuning, specifically targeted at applications running on Linux on IBM Power Systems.
CPU Frequency Scaling on IBM Power Systems Running Linux
To measure and control CPU frequency and freq/voltage scaling on Linux on Power, the first step is to understand what firmware mode the system is running in, OPAL or PowerVM.
SMT and cgroup cpusets
If you are running Ubuntu, there is a cgroup hotplug issue to consider when changing SMT modes. If the system was previously set to a lower SMT mode and a user changes the system to a higher SMT mode, this cgroup hotplug issue may prevent tasks from running on those CPUs that were brought online to switch the processor cores to a higher SMT mode.