IBM and AMD have released Pelagos, an open-source, GPU-accelerated framework for evaluating quantum error-correction decoders at scale. The practical point: researchers can use AMD or NVIDIA GPUs to run demanding decoder tests, with the companies reporting strong scaling across eight AMD GPUs.
AMD Watch analysis
What happened
Pelagos is designed to help researchers simulate the error patterns used to assess how well a quantum error-correction decoder performs. Those evaluations can require billions of samples, making classical computing resources an important part of developing fault-tolerant quantum systems.
HPCwire reports that the framework uses Triton and supports both AMD and NVIDIA GPUs. In accompanying research, IBM and AMD describe tests on AMD Instinct MI300X and MI355X accelerators, where the GPU decoding stage achieved speedups of up to roughly 7.5 times when scaling from one GPU to eight. They also cite earlier results of up to 97 times the throughput of a single CPU core for decoder-evaluation workloads.
Why it matters
Quantum error correction is not only a problem for quantum hardware. Researchers need to test decoding methods against large numbers of simulated errors, and that workload depends on conventional computing too. Pelagos gives researchers an open-source framework for that work, while the reported multi-GPU results offer a concrete example of how accelerator systems might help run it at scale.
The figures describe particular tests, not a universal speed advantage for every decoder or system. Still, this is more than a partnership announcement: there is a named tool, published scaling work and an invitation for researchers to use and contribute to the framework.
Our read
This is a useful piece of infrastructure for a field where progress depends on the less glamorous business of testing whether error correction works. Open source and support for GPUs from both AMD and NVIDIA make Pelagos easier to explore beyond a single hardware camp. The next question is whether other research groups find it useful for their own decoders and workloads.
What to watch
- Whether researchers outside IBM and AMD adopt Pelagos or contribute new decoders.
- How performance varies across different codes, decoder algorithms and GPU systems.
- Whether the framework supports larger, more realistic evaluation campaigns.
Discussion spark: Should open-source, GPU-accelerated tools such as Pelagos become the standard way to compare quantum decoders, or are independent benchmarks needed before researchers rely on their results?
Sources and evidence
- IBM and AMD Accelerate Quantum Error Correction Decoding at Scale – HPCwire (17 November 2026, 17:11 UTC)
Independent WittyWires tracker for public updates about AMD. Not affiliated with or endorsed by AMD; this is not an official account.