Vampir meets International Supercomputing Conference 2013

Vampir meets ISC '13

We would like to welcome you at our Booth #645 at ISC '13 from June 16th to June 20th 2013 in the city of Leipzig. Among other activities, we will showcase our recent improvements in performance tracking and visualizing on the road to exascale.

Vampir 8.1 will be released at ISC '13. This release focuses on extending the graphical presentation of performance data and adds support for the collaborative performance monitor Score-P 1.2, while maintaining compatibility with earlier VampirTrace releases. Among other features this release introduces hierarchical process folding, new peer-to-peer communication metrics, and a fully re-designed comparison mode for multiple program runs.

Score-P is the primary code instrumentation and run-time measurement framework for Vampir 8 and also works natively with Scalasca, TAU, and Periscope. The latest version 1.2 provides a number of improvements and the following new features: generic event model for remote memory access (RMA), metric plugins for custom user data from arbitrary sources, snapshots (OTF2) that enable selective processing of performance data, and support for Cray systems.

Additional information material

The Vampir performance analysis tool is available in three editions to meet the needs of different business
sizes and application development goals. The editions provide different features sets to support
application performance optimization in workstation-type scenarios as well as in extreme scale scenarios.

Vampir Editions Vampir’s visual performance analysis provides a convenient way to gain insight into parallel program
behavior. To generate its input event traces it relies on the powerful Open Source toolkit Score-P for code
instrumentation and run-time monitoring.

Vampir’s visual performance analysis provides a convenient way to gain insight into parallel program
behavior. To generate its input event traces it relies on the powerful Open Source toolkit Score-P for code
instrumentation and run-time monitoring.

Vampir Tool Chain

Performance data originating from large-scale parallel applications can easily reach volumes in the
order of hundreds of gigabytes. Interactive visualization and analysis of such data volumes require a
parallel and highly scalable analysis engine. The Vampir tool chain addresses this issue with a parallel
analysis component called VampirServer. Developers can interact with VampirServer from their desktop
using the Vampir GUI as a performance data browser.

Vampir Scalable
Trace Analyse

Analysis of event data with Vampir is an effective approach to optimize the performance of parallel
applications.
Collecting the performance data in a scalable and efficient fashion is a highly challenging
task. With the new Score-P scalable performance monitor we support a convenient measurement infrastructure
for recording fine-grained performance events, with special focus on parallel applications.

Vampir Monitoring
with Score-P The Marmot Umpire Scalable Tool (MUST) is
a runtime error detection tool that automatically
detects non standard compliant use of
MPI.

The Marmot Umpire Scalable Tool (MUST) is
a runtime error detection tool that automatically
detects non standard compliant use of
MPI.

MUST flyer

Our events on International Supercomputing Conference 2013

Time	Room	Title
Tue, 18th June 10:00 AM - 11:00 AM	M01 Messehaus	BoF 08: An Integrated Software Defined Network Architecture for Applications Relying on Huge, Geographically Dispersed Datasets
Wed, 19th June 12:00 PM - 01:00 PM	Hall 4 CCL	BoF 16: Execution Analysis & Optimization of Parallel Applications

Screenshots

> <