Jeff Wittich, Chief Product Officer
Using OnSpecta’s Deep Learning Software (DLS) optimization engine, we can deliver performance increases with higher performance, lower power and better predictability than ever before. As a result, our customers will achieve significantly higher AI performance from the Ampere® Altra® family of cloud native processors deployed in cloud and edge infrastructures.
Yes, a 4X Increase!
For Ampere, this acquisition brings us one step closer to our vision to deliver a truly differentiated cloud native computing platform for our customers in both cloud and edge deployments – including inferencing which is already a significant portion of the workloads deployed on general purpose compute in the cloud. We recognize that our customers need to manage the costs and the growing requirements for AI inference tasks in both centralized and edge-based infrastructure scenarios. To solve these challenges, DLS is a seamless binary drop-in library for many popular AI frameworks and will accelerate AI inference significantly on Ampere Altra delivering instant speed-up which will lower costs and improve responsiveness. DLS also enables the use of the Altra-native FP16 data format that can double performance over FP32 formats without significant accuracy loss or model retraining in addition to the speed-up achieved by the inherent operational improvements of the DLS engine!
The acquisition was covered by leading journalists:
- – According to Tobias Mann from SDx Central “Ampere Computing’s acquisition of OnSpecta last week was a warning shot to Inteland Nvidia as the Arm-CPU vendor turns its attention to AI inference.”
- – The Next Platform stated that “Ampere computing buys an AI inference performance leap.”
- – According to ZDNet, “OnSpecta’s deep learning expertise will help Ampere deliver a better platform for inference task – processing.”
A powerful combination
The OnSpecta DLS AI optimization engine can deliver significant CPU-based performance enhancements on commonly used machine learning frameworks like TensorFlow, ONNX, and PyTorch. This acquisition will bring these exciting new capabilities to Ampere and its Ampere Altra processor family. Last year, we started shipping Ampere Altra, which we have scaled out to the Ampere Altra Max processor this year. By applying the DLS optimization layer to its industry leading 128 cores, Ampere will deliver the most scalable AI inference platform for the cloud.
We look forward to seeing our customers benefit from this acquisition as they begin developing and rolling out exciting new products and applications that leverage all that the OnSpecta DLS AI inference engine can deliver.
We welcome OnSpecta into our growing and dynamic family. AI inference has never looked better, or should we say faster and more accessible.