Software MIL X Update 114 and 115
Published: December 5, 2022
Matrox® Imaging has released two updates for MIL X—Update 114 and Update 115—which work to offload, and in many cases accelerate, inference or prediction using a MIL X deep learning classifier.
Update 114 performs the offload using an Intel® Integrated GPU (iGPU) through Intel OpenVINO with the possibility of classifying a roughly 1 Megapixel monochrome image two times faster going from the central processing unit (CPU) cores (using FP32 datatype) to the iGPU (using FP16 datatype) in the Intel Core i5-1250PE processor.
Update 115 uses an NVIDIA GPU through CUDA to handle the offload. This update can yield a two-time speed-up when classifying a roughly 1 Megapixel monochrome image going from an Intel Core i7-12700 processor (using CPU cores and FP32 datatype) to an NVIDIA GeForce RTX 3060 graphics card (using FP32 datatype).
These updates are available for download through the software update service to registered users of MIL X with a valid maintenance subscription as well as those evaluating MIL X. Please note that both updates require the presence of MIL X Version 1911 or higher with Service Pack 5 (SP5).