DSN Member Profile: Myrtle.AI

Myrtle.AI

Overview

Myrtle.ai optimizes inference workloads for neural networks deployed on Intel FPGAs in data centers and in embedded applications. With a focus on Recommendation Models and RNNs, Myrtle.ai is able to deliver high throughput in latency-constrained applications such as recommendation systems, speech transcription, natural language processing, speech synthesis, time series analysis and trading fraud detection. This enables businesses to reduce costs and offer enhanced, scalable services to their customers. Myrtle.ai owns the speech transcription workload for MLPerf, a machine learning benchmarking initiative driven by the industry’s leaders.

Available Design Services

Design Services Device Family Country/Region Supported
Design Verification
Embedded Design: ARM* Drivers, ARM* Applications
Software Design: Software Drivers, Intel Xeon, Board Support Packages
Hardware Design: Specification Development, FPGA, Timing Closure
IP Integration
Contract Manufacturing
Stratix® V, Intel® Stratix® 10, Intel® Arria® 10 EMEA: Sweden, Netherlands, Spain, Ireland, Austria, Italy, Germany, France, Israel, Switzerland, Denmark, United Kingdom, Finland, Belgium, Norway
Americas: Canada, United States

Additional Services and Features

‧    Delivering lowest latency, highest performance, lowest power

‧    Best TCO compared to software-only and GPGPU alternatives

‧    Support for Caffee, MxNet, TensorFlow, Pytorch via ONNX

Contact Information

50 St Andrews Street
Cambridge, United Kingdom
Tel:(44) 1223967248
E-mail:hello@myrtle.ai
Website:https://www.myrtle.ai