Pearl Research Labs
A company building AI inference infrastructure and Pearl, a blockchain whose proof-of-useful-work mining is a by-product of GPU matrix multiplication.
Pearl Research Labs describes itself as building infrastructure and algorithms for AI inference 1. It develops the Pearl blockchain, an instance of proofs of useful work:
- Protocol specification. Pearl specifies a proof-of-useful-work protocol in which the unit of mining work is FP8 matrix multiplication on GPUs 2.
- Integer whitepaper. Its earlier whitepaper presents mining as a side-effect of the matrix multiplications in AI training and inference workloads 3.
- Open-source network. Pearl publishes a monorepo with a reference full node, a vLLM-based GPU miner and a proof-of-work circuit and verifier 4. Its README says the mining follows the proof of useful work from arbitrary matrix multiplication proposed by Komargodski and Weinstein 4 5.
- Reproducible GPU arithmetic. Two Pearl researchers, with two Stanford co-authors, published Hawkeye at MLSys 2026. It re-executes on a CPU, bit for bit, the matrix multiplications that NVIDIA Ampere, Hopper and Ada Lovelace GPUs perform in FP16, BF16 and FP8 6; see Deterministic and bit-exact inference.
On this page
Implementations
Implementations this organization develops.
- A blockchain whose mining is designed to be a by-product of GPU matrix multiplications in AI workloads, with public node and miner code.
Related records
Mechanisms and implementations whose records cite or describe this organization's work.
- Cryptographic evidence that a given amount of matrix-multiplication work was completed, proposed as one input to accounting for spare capacity on declared hardware.
Publications
Sources this organization authored or published.
- AE. Badash et al. (2026). Hawkeye: Reproducing GPU-Level Non-Determinism. Proceedings of Machine Learning and Systems 8 (MLSys 2026). Source recordCited by Deterministic and bit-exact inference; Pearl proof-of-useful-work blockchain; Numerical nondeterminism; Pearl Research Labs
- BPearl Research Team (2026). Pearl Floating Point Scheme Specification. Pearl Research Labs. Source recordCited by Proofs of useful work for capacity accounting; Pearl proof-of-useful-work blockchain; Proof of (useful) work; Pearl Research Labs
- BPearl Research Labs (2026). Pearl INT Whitepaper. Pearl Research Labs. Source recordCited by Proofs of useful work for capacity accounting; Pearl proof-of-useful-work blockchain; Pearl Research Labs
- BPearl Research Labs homepage. Pearl Research Labs. Source recordCited by Pearl Research Labs
- BPearl Research Labs (2026). pearl: Monorepo for the Pearl network. GitHub. Source recordCited by Proofs of useful work for capacity accounting; Pearl proof-of-useful-work blockchain; Pearl Research Labs
- CPearl Research Labs (2026). Proof of Useful Work from the Ground Up. Pearl Research Labs. Source recordCited by Pearl proof-of-useful-work blockchain
Sources
- BPearl Research Labs homepage. Pearl Research Labs. Source recordSupports: infrastructure and algorithms for AI inference
- BPearl Research Team (2026). Pearl Floating Point Scheme Specification. Pearl Research Labs. Source recordSupports: FP8 proof-of-useful-work protocol specification
- BPearl Research Labs (2026). Pearl INT Whitepaper. Pearl Research Labs. Source recordSupports: integer matrix-multiplication whitepaper; mining as a side-effect of AI workloads
- BPearl Research Labs (2026). pearl: Monorepo for the Pearl network. GitHub. Source recordSupports: public monorepo: full node, GPU miner, proof-of-work circuit and verifier; README cites the Komargodski-Weinstein proposal
- BI. Komargodski & O. Weinstein (2025). Proofs of Useful Work from Arbitrary Matrix Multiplication. arXiv. Source recordSupports: proof of useful work from arbitrary matrix multiplication (Komargodski and Weinstein)
- AE. Badash et al. (2026). Hawkeye: Reproducing GPU-Level Non-Determinism. Proceedings of Machine Learning and Systems 8 (MLSys 2026). Source recordSupports: Hawkeye (MLSys 2026): two of four authors at Pearl Research Labs; bit-exact CPU replay of GPU matrix multiplication