Myrtle.ai cuts latency in half for financial machine learning inference benchmark recording using VOLLO | For Enterprises

CAMBRIDGE, UK , April 29, 2026 /PRNewswire/ — myrtle.ai, a recognized leader in accelerating machine learning inference, today announced that its VOLLO® product-powered stack was recently audited by STAC®, a leading benchmarking organization in the financial industry.^[1] The results, announced today at the STAC Summit in London, clearly demonstrate the latency benefits of FPGA-based solutions for ML inference in financial trading and related applications.

STAC-ML (Markets) Inference is a technology benchmark standard for solutions that can be used to perform inference on real-time market data. STAC-ML Markets (Inference) is designed by quants and technologists from the world’s leading financial companies to report on the performance, resource efficiency, and quality of technology stacks that can perform inference using the models provided.

VOLLO is 2 microseconds (99^th percentile), throughput and efficiency. All three benchmark models inferred lower latency for VOLLO (99^th percentile) is higher than all previously audited systems and is half the previous record. This low and deterministic latency allows users to use more complex models and make more intelligent decisions faster than ever before, giving them a competitive advantage in trading, risk analysis, quotes, and many other trading-related activities.

VOLLO has hundreds of thousands of hours of production trading under its belt and is currently generating alpha for many of the world’s leading trading companies. These companies developed and trained a wide range of models in standard ML tool flows before compiling them into VOLLO and running them on their FPGA-based hardware platform of choice.

In the system under test, VOLLO ran on Silicom’s standard form factor FBAP4@VP18-2L0S PCIe accelerator card, equipped with an AMD Versal™ Premium series VP1802 adaptive SoC and installed in a Supermicro AS -2015CS-TNR server. AMD Versal Premium Series Adaptive SoCs offer PCIe Gen5x8 and over 3.3 million programmable LUTs, making them ideal for low-latency inference applications.

“Since VOLLO first exploited the full potential of FPGAs with the 2023 STAC benchmark, we have worked with our customers to further reduce latency, expand the types and sizes of models that VOLLO can run, and expand the range of platforms on which it can run.” said Peter Baldwin, CEO of myrtle.ai.We are pleased to collaborate with AMD, Silicom, and Supermicro on this benchmark to demonstrate how our combined technology enables ultra-low-latency AI inference in quantitative trading.”

“The future of financial markets will be shaped by AI systems that can interpret and act on data in near real-time.” said Girish Malipeddi, Director of AMD’s Data Center FPGA Business.myrtle.ai’s VOLLO, built on AMD Versal™ Premium Series Adaptive SoCs, demonstrates how advanced low-latency inference can help unlock a new generation of intelligent trading infrastructure.“

“Supermicro continues to address the broader market with the AMD systems used for this STAC-ML benchmark.Michael McNerney, Supermicro’s senior vice president of marketing and network security.Our servers address the financial services industry’s most challenging workloads, and in collaboration with our partners, we can deliver top-end performance with very low latency for machine learning workloads.”

Anders Poulsen, Vice President Solutions at Silicom Denmark:We are pleased that myrtle.ai selected Silicom’s Artena accelerator card based on AMD Versal Premium for these tests. Built around one of the largest FPGAs in the PCIe form factor, Artena is the ideal platform for VOLLO. VOLLO combined with our low-latency hardware delivers deterministic microsecond-level inference for demanding trading workloads.”

ML developers can evaluate how their models run on VOLLO without requiring FPGA tools or expertise. For more information, visit vollo.myrtle.ai or contact myrtle.ai (fintech@myrtle.ai) today.

Complete benchmark results are available in the STAC report (SUT ID MRTL260323) at http://www.STACresearch.com/MRTL260323.

About myrtle.ai

Myrtle.ai is an AI/ML software company that provides world-class inference accelerators on FPGA-based platforms from all major FPGA suppliers. myrtle.ai has extensive expertise in neural networks and has provided accelerators for applications such as fintech, wireless communications, LLM, voice processing, and recommendations.

VOLLO, VOLLO Accelerator, and the VOLLO logo are registered trademarks of myrtle.ai.

“STAC” and all STAC names are trademarks or registered trademarks of Strategic Technology Analysis Center, LLC. AMD, the AMD logo, Versal, and combinations thereof are trademarks of Advanced Micro Devices, Inc.

^[1]www.STACresearch.com/MRTL260323

Photo – https://mma.prnewswire.com/media/2965395/Myrtle_ai.jpg
Logo – https://mma.prnewswire.com/media/2965394/Myrtle_ai_Logo.jpg

View original content: https://www.prnewswire.co.uk/news-releases/myrtleai-halves-latency-in-financial-machine-learning-inference-benchmark-record-with-vollo-302753239.html