Tokyo Institute of Technology Chooses TotalView for Debugging Large Scale Simulations

Rogue Wave Software announced that the Tokyo Institute of Technology has selected TotalView for debugging on their latest supercomputer, the TSUBAME2. Tokyo Tech is Japan's largest public research university dedicated to science and technology, founded in 1881, with three undergraduate schools, six graduate schools, and numerous research labs and facilities.

The computing services of TSUBAME2, in the Global Scientific Information and Computing Center, are used by a wide community: Tokyo Tech's students, researchers, and staff; external collaborators; and even the private sector. Each of the 1408 nodes on the system includes three NVIDIA Tesla M2050 GPUs, and the system's users need to be able to debug their host programs as well as CUDA on GPU simultaneously. Tokyo Tech's search for a debugger that would provide such capabilities led them to TotalView; its seamless handling of both the CPU and CUDA debugging was a key factor in their decision. TotalView offered a unique combination of capabilities for pinpointing and fixing hard-to-reproduce bugs, memory leaks, and performance issues related to GPU development. Since each node of TSUBAME2 boasts three Tesla cards, Tokyo Tech benefits from TotalView's unique ability to simultaneously debug parallel applications using more than one GPU device per node across the entire cluster.

"We take pride in providing Tokyo Tech with the complete solution for the debugging needs of the HPC communities they serve," said Brian Pierce, Rogue Wave CEO. "TotalView, with its interactive analysis and debugging capabilities, simplifies development for CUDA and MPI applications, and frees up researchers and scientists at the Institute to focus on their subjects."

TotalView is designed for developer productivity, simplifying and shortening the process of developing, debugging, and optimizing complex code. TotalView provides the visibility into threads and GPU kernels that is so critical in tracking down and resolving elusive software problems.

"Being able to debug with TotalView will help us in the development of large scale simulation programs such as material science, computational fluid dynamics, bioscience, and environmental science simulation," says Dr. Satoshi Matsuoka, Professor, Global Scientific Information and Computing Center, Tokyo Institute of Technology. "We view TotalView as a vital component of the environment that enables us to produce our high-level parallel and heterogeneous processing applications."