Customer
How Zuse Institute of Berlin (ZIB) Leveraged the Debugging Capabilities of Perforce TotalView to Identify Critical Issues in Parallel Execution
" TotalView provides unique features to identify and help solve software bugs in parallel software. Its ability to handle complex debugging tasks at scale makes it an indispensable tool for our HPC environment"
Zuse Institute Berlin (ZIB) is a premier academic research organization based in Berlin, Germany. ZIB is an interdisciplinary research institute for applied mathematics and data-intensive high-performance computing. Its infrastructure supports more than 1,000 researchers across 250 distinct projects, enabling innovation in areas such as life sciences, chemistry, engineering, and earth-systems sciences.
ZIB turned to Perforce TotalView for its state-of-the-art debugging capabilities that can handle the complexities of parallel systems.
TotalView Helps ZIB:
- Debug in FORTRAN, C/C++, and Python
- Manage parallel systems
- Oversee more than 1,000 users and 250 projects
Company Snapshot
- Name: Zuse Institute Berlin (ZIB)
- Industry: Academic Research
- Location: Berlin, Germany
- Team Size: 250 total employees
- Key Focus: HPC resources in service of scientific research
Solving Parallel Debugging Challenges
ZIB operates one of the most advanced HPC systems in Germany, featuring 982 nodes with Intel CLX-AP CPUs, cutting-edge AMD Genoa and Nvidia A100 GPUs, and 20 petabytes of online storage capacity. This infrastructure is utilized by more than 1,000 researchers spanning multiple disciplines.
However, managing such a vast computational environment brings immense complexity, particularly regarding debugging parallel applications across numerous processors and threads. Debugging parallel software developed in multiple languages (FORTRAN, C/C++, and Python) requires advanced tools capable of handling the complexity of these environments.
With the help of partner SMB, ZIB identified the need for a state-of-the-art debugging solution with a graphical interface — and one that could effectively handle the scale and intricacies of ZIB’s first massively parallel HPC system, a Cray T3D.
Reducing Debugging Time for Optimal System Performance
In the search for an HPC debugging solution, Perforce TotalView stood out for its advanced debugging capabilities, including independent thread control, multi-platform support, memory debugging, and its ability to provide a seamless debugging experience across thousands of processes.
Perforce TotalView enabled ZIB developers to quickly identify and resolve software issues, significantly reducing debugging times and ensuring optimal system performance for the many code developers who rely on ZIB’s HPC infrastructure.
Metrics of Success
The integration of Perforce TotalView into ZIB's HPC system has been a game-changer, delivering remarkable improvements in software debugging and development efficiency. Key highlights of success include:
- Faster Bug Identification: ZIB developers can quickly pinpoint critical issues related to the parallel execution of software across nodes and threads.
- Streamlined Debugging: The GUI simplifies complex parallel debugging processes, allowing developers to save significant time.
- Enhanced Performance: Application errors are detected and resolved efficiently, maximizing the efficient use of ZIB’s HPC infrastructure.
Looking ahead, ZIB plans to continue leveraging TotalView's advanced capabilities to support its mission of driving scientific discovery. Regular TotalView updates ensure that ZIB’s HPC ecosystem remains at the forefront of innovation.
Debug Complex Applications With TotalView
Perforce TotalView helps research-focused organizations like Zuse Institute Berlin (ZIB) debug complex code. See for yourself how TotalView can help your organization do the same.