gpu-burn review
A focused command-line tool for timed CUDA stress workloads on NVIDIA GPUs.
Reviewed by iTechGuides Editors · Editorial team · Updated Oct 2026
gpu-burn is an open-source command-line tool for stressing NVIDIA GPUs with CUDA workloads. It suits people who want to exercise GPU computation and GPU memory, especially when they need to run a timed workload across multiple GPUs or target one by index. It is a focused option for solo users and small teams working with CUDA, rather than a general PC stress-testing suite: its documented scope does not include CPU or system-memory testing, or combined CPU-and-GPU workloads.
The workload controls make its narrow focus useful for targeted checks. A run can use a specified duration and a selected amount of GPU memory, expressed in megabytes or as a percentage of available memory. Users can choose double-precision operations or try Tensor Cores when available, and the tool can list GPUs detected by the system. These options help tailor GPU compute and memory pressure, but they do not change the tool’s scope: it is not described as a memory-error diagnostic or a broad system stability test. Choose it when CUDA workload controls are the priority; look elsewhere if you need CPU, system-memory, or combined workload coverage.
The project is free and open source, with self-hosted use and community support. The repository provides build and run instructions, including a Docker workflow, and documents a Windows port; Linux is the project’s origin platform. That means prospective users should be comfortable following build or container instructions and managing their own environment. Windows is an option, but the port framing makes Linux the clearer fit for a straightforward project choice. Select gpu-burn for targeted CUDA stress workloads across GPUs, and consider a broader PC testing tool if you need one application to cover other hardware components or a different platform scope.
gpu-burn pros and cons
- Where it wins
- Runs timed CUDA workloads across multiple GPUs or a selected GPU.
- Sets GPU memory use by megabytes or percentage.
- Can use double precision and Tensor Cores when available.
- Where it doesn't
- Does not cover CPU or system-memory stress testing.
- CUDA-specific scope limits its use beyond NVIDIA GPU workloads.
- Windows support is documented as a port; setup involves building or Docker.
gpu-burn fact sheet, pricing and score →
Advertiser disclosure: iTechGuides is reader-supported. Vendors can pay for top positions in our rankings and for a place on other products' pages, and we may earn a commission when you click some links. How we rank.
Last updated · How we research and update