Arm DDT - part of the Arm Forge toolsuite (formerly Allinea DDT/MAP) for high performance software - is an intuitive graphical debugger that works at the scale you do.

Supporting the latest toolkits from NVIDIA, Arm DDT provides one of the most comprehensive debuggers of CPU and GPU applications available.

Relied on by software developers on CUDA-enabled laptops and workstations through to the world’s largest hybrid GPU/CPU supercomputers, Arm DDT is designed to help you tackle software bugs in CPU/GPU applications easily.

screen shot1


screen shot2

(click images to expand)

Key Features

  • Supports CUDA C, CUDA C++, CUDA FORTRAN and OpenACC
  • Source level debugging support for compilers from NVIDIA, Cray and the Portland Group
  • Debug, edit, build and run within one tool
  • Simultaneously debug GPU and CPU code
  • Multi-threaded, multi-process, multi-GPU and multi-node debugging
  • Browse source, set breakpoints, examine variables and control processes, threads and warps
  • Crystal clear concurrency with data and stacks from threads and processes merged to highlight outliers and exceptions
  • Explore and visualize CPU and GPU memory
  • Memory debugging to trap key CUDA and CPU memory errors such as beyond-array bound accesses and memory leaks
  • Supports all MPI implementations for multi-node clusters and supercomputers


Arm DDT is available now and supports the latest CUDA toolkits and compilers on Linux.

Free trial licenses can be downloaded from

For more information about Arm DDT and other CUDA tools: