IBM, AMD Release Open-Source Quantum Decoder Tool

Pelagos, a GPU-based simulation tool for quantum error-correction decoders, is now available under the Apache License 2.0.

Updated on Nov. 17, 2026 in Quantum Computing

Bold flat-color editorial illustration of an unlabeled GPU accelerator card with heatsink fins and a linked lattice pathway, evoking quantum error-correction research.
IBM and AMD released Pelagos, an open-source GPU tool for simulating quantum error-correction decoders, under the Apache License 2.0 on November 16. AI Illustration. Upload story photo >

IBM and AMD released Pelagos, an open-source GPU tool for simulating quantum error-correction decoders, on November 16. Oak Ridge National Laboratory joined IBM and AMD on November 5 in unveiling a separate quantum error-correction platform.

Why it matters

Error correction is a central obstacle to building practical quantum computers, and faster decoder simulation can speed research into how such systems will be made reliable. The release is open-source, letting researchers worldwide inspect, use and build on the tool.

Eight MI355X GPUs delivered 7.01 times the end-to-end throughput of one GPU on a two-gross code benchmark, with eight MI300X GPUs reaching 6.92 times. The benchmarks hold relay depth at R=50 and test decoder simulations, not error correction on a live quantum computer.

The players

IBM

International Business Machines is a long-running American technology company that has pursued quantum computing research and hardware development for decades.

AMD

Advanced Micro Devices is a semiconductor company known for its Instinct line of GPUs used in high-performance and AI computing.

Oak Ridge National Laboratory

Oak Ridge National Laboratory is a U.S. Department of Energy science facility in Tennessee that conducts research across energy, computing and materials.

The details

Pelagos uses Triton to run a Relay-BP decoder and processes multiple error syndromes in parallel. The tool is available under the Apache License 2.0, and the reported speedups come from benchmark tests on AMD Instinct MI300X and MI355X GPUs.

Timeline

  1. November 5: Oak Ridge National Laboratory joins IBM and AMD in unveiling a quantum error-correction platform.

  2. November 16: IBM and AMD release Pelagos.

The Tech Race

The release sits within the race among IBM, AMD and national labs to make fault-tolerant quantum computing practical, where error-correction decoding is a known bottleneck. Making decoder simulation faster and openly available lets more groups test error-correction designs before they are built into hardware.

Researchers and developers can download Pelagos now under the Apache License 2.0 and run decoder simulations on AMD Instinct GPUs. The tool does not change consumer quantum products today, and its practical value depends on whether the speedups carry over to real quantum hardware.

The takeaway

Faster decoder simulation shortens the time needed to test error-correction designs before they reach hardware. Readers tracking quantum computing should watch whether these simulation gains translate to live quantum systems.

Further reading

Explore more on Quantum Computing to see how error correction fits into the broader race to build reliable quantum machines.

Source note: This article includes information reported by TokenPost.