Description: MemtestG80 and MemtestCL are a software-based testers to test for "soft errors" in GPU memory or logic for NVIDIA CUDA-enabled or OpenCL-enabled (of any manufacturer) GPUs. They use a variety of proven test patterns (some custom and some based on Memtest86) to verify the correct operation of GPU memory and logic. They are useful tools to ensure that given GPUs do not produce "silent errors" which may corrupt the results of a computation without triggering an overt error.
Haque IS and Pande VS. Hard Data on Soft Errors: A Large-Scale Assessment of Real-World Error Rates in GPGPU. In Proceedings of 10th IEEE/ACM International Conference on Cluster, Cloud, and Grid Computing (CCGrid 2010), pp 691-696.
Available Downloads and Their Potential Uses: This project offers prebuilt binaries as well as source code for the CUDA-based MemtestG80 tester (NVIDIA-only) and OpenCL-based MemtestCL (vendor-agnostic).
Purpose/Synopsis: MemtestG80 and MemtestCL are software-based tester to test for "soft errors" in GPU memory or logic for GPUs supporting CUDA or OpenCL.
Audience: Users (scientific and general) of NVIDIA/ATI GPUs interested in verifying the correct operation of their hardware.
Long Term Goals and Related Uses: 1) MemtestG80/CL allows end-users to verify the correct operation of their hardware under their own environmental conditions.
2) GPU software developers can integrate the MemtestG80/CL code into their own projects prior to distribution as an added self-test mechanism on target machines.
Keywords: gpu, memory, reliability, testing
(Links display other projects with that keyword anywhere in the project.)