With up to 160 SIMD units per processor on Altra® and 256 units on Altra Max, computational workloads that are built to scale perform well on Ampere® Altra®. Computational workloads are divied up here in several categories to allow quick review of various high-performance compute applications.
AI Inference, the practical application workhorse of modern Artifical Intelligence infrastructure
Video Transcode workloads are the backbone of CDN networks and social media sites across the internet. Find all aspects of the Video delivery stack for content distribution and processing here.
Box86 enables running x86 (Linux) applications on non-x86 Linux systems. Integrated with DynaRec (dynamic recompiler) and by using native library box86 enable high performance on GPU based applictions. (Source from https://github.com/ptitSeb/box86). As Ampere Altra and Altra Max can support both 32 and 64 applications, Box86 are natively supported on Ampere platforms.
Remote rendering, simulation and visualization are important for big dataset, computational demanding simulations, etc. The applications include CFD (aerodynamic design, weather prediction, etc), centralized computer aided systems (AR/VR), big data set (like BIM). Together with partners, Ampere has enabled various GPUs on Ampere Altra and Altra Max platforms. Servers support large RAM, SSD, multiple GPUs, and multiple CPUs are available for large data set with demanding simulation tasks.