Microsoft's Project Brainwave wants to make AI super fastAt Microsoft Build, Microsoft reveals how Project Brainwave will use fast, flexible chips to make AI available in real time.
I'm really thrilled to announce the preview of project Brainwave, project Brainwave is a distributed real time AI that today is working with FBGAs from Intel to bring the best in class, real time AI, low latency AI capabilities and inferencing capabilities. It has 5x lower hardware latency than TPU for real time AI. And this is gonna be available in preview for developers. And we're also gonna extend it all the way to Edge. We're works with our system partners to take FPGA and make it available Wherever Azure is available. And we have many customers who have these high precision and high throughput needs who are working with us. And one such customer is Jabil. And so let's roll the video to show you how they using FBGA for realtime AI already. [MUSIC] People tend to think of Jabil as a contract manufacturer. We are more than that. We are a manufacturing solutions provider. We've been talking with Microsoft about Project Brainwave. The benefit of the FPGA is the ability to deploy our machine learning models at the edge at a lower cost than GPUs but at a higher performance capability than CPUs. The early test result from Microsoft has shown FPGA's Capable of predicting 550 images a second comparing with a CPU cluster who is 40 images a second. [MUSIC] The potential for Project Brainwave is substantial in allowing us to move forward at scale for machine learning across Jabil, very exciting. [MUSIC]