myrtle.ai's VOLLO sets new STAC-ML records for Gradient Boosting Tree Inference
At the STAC Summit in London, myrtle.ai announced that its VOLLO® inference accelerator has set new benchmarks in gradient boosting tree (GBT) model inference under the STAC-ML Markets framework. The system achieved a p99 latency of less than 2 microseconds and throughput rates up to 50 million operations per second on an AMD Alveo V80LL card within a Blackcore server. These results represent more than a 30% reduction in latency compared to previous benchmarks, with throughput increased by at least five times. The performance improvements enable firms to deploy larger and more accurate machine learning models without compromising speed or market responsiveness. This is particularly critical for electronic trading where milliseconds can impact returns significantly. VOLLO has already been deployed successfully in live trading environments across thousands of hours and is now also being adopted beyond finance, including telecommunications, cybersecurity, and defense sectors due to its flexibility.
Tickers: $AMD
Topics: AI low latency computing machine learning inference STAC benchmarks