IBM and AMD have announced the launch of Pelagos, a new open-source framework designed to accelerate the evaluation of decoders used in quantum error correction. This launch is an advanced step in the ongoing collaboration between the two companies and the Oak Ridge National Laboratory (ORNL) to build infrastructure that supports fault-tolerant quantum computing.
The Challenge: The Computational Burden of Error Correction
Quantum computing is inherently prone to errors, making "quantum error correction" a fundamental pillar for making these systems useful and practical. This process relies on monitoring what are known as "syndromes" that identify where an error has occurred without compromising the quantum information, after which a "decoder" determines the appropriate correction.
As research moves toward building larger and more complex quantum systems, the need to simulate billions of error patterns to test algorithm efficiency increases, which requires massive classical computing capabilities. This is where the need for tools capable of processing this massive data at high speed arises, which is the role Pelagos was designed to fulfill.
Technical Capabilities of the Pelagos Framework
The Pelagos framework (which derives its name from the Greek word for "open sea") relies on advanced acceleration techniques to bridge the gap between quantum computing and high-performance classical computing (HPC). Its key features include:
- GPU Acceleration: The framework was built using the Triton programming framework, allowing it to operate efficiently on both AMD and NVIDIA GPU environments.
- Superior Performance: Initial studies have shown that Pelagos can achieve a throughput 97 times higher than a single CPU core, a performance comparable to hundreds of traditional processor cores.
- Scaling Across Multiple Systems: Tests have demonstrated the framework's ability to scale efficiently across multiple GPU systems, with AMD Instinct™ MI300X and MI355X processors achieving up to 7.5 times acceleration when using eight GPUs compared to a single unit.
Why is this development technically significant?
The dilemma of quantum computing is not limited to developing quantum hardware; it extends to the classical infrastructure responsible for managing these systems. This development illustrates that achieving stable quantum computing requires tight integration between quantum processors and the high-performance computing (HPC) systems that manage decoding operations and data coordination.
By making Pelagos open-source, IBM and AMD aim to lower barriers for researchers and developers, allowing for broader participation in the development of error correction algorithms. This approach supports the vision of "quantum-centric supercomputing," where classical and quantum processors work side-by-side in an environment optimized to reduce latency and increase reliability.