Quantum X Labs Inc. (Nasdaq: QXL) reported that its GPU-accelerated quantum-simulation platform achieved an approximately 16.8-fold computational acceleration compared with a CPU-only implementation, according to a press release from the Tel Aviv-based company.

The benchmark used a single machine with 64 cores and a single Nvidia T4 Tensor Core GPU, running on an AWS g4dn.16xlarge EC2 instance with IBM Qiskit Aer GPU-based quantum simulation. The same 64-vCPU machine without GPU acceleration served as the comparison baseline.

The evaluation used a real clinical gene-expression dataset comprising 11 patients and 3,531 gene-expression features. The company's proprietary algorithm, operated through its subsidiary CliniQuantum, searched for correlated three-gene configurations across 7,331,162,245 possible combinations, identifying approximately 22.2 million correlated gene triplets.

The CPU-only analysis required approximately 8.4 hours, while the GPU-accelerated version completed the same task in approximately 0.5 hours, producing identical results.

The company stated the benchmark was designed to evaluate computational performance only and did not assess the clinical validity, predictive value, or potential utility of any identified gene combinations.

"Clinical and molecular datasets can contain thousands of variables but relatively few patients, creating enormous combinatorial search spaces," said Dr. Tidhar Turgeman, head of clinical trials data analysis at Quantum X Labs. "Demonstrating the ability to evaluate more than 7.3 billion potential gene combinations on real clinical data marks an important step in our development."

Quantum X Labs said it intends to advance its algorithms toward more complex multi-feature configurations and larger biomedical datasets, and to progress from simulation toward execution on quantum computing hardware.