The AI revolution is steadily bringing waves of innovative technologies and practices to many sectors of the modern economy. AI inference-based tasks such as image classification are having transformative impacts in such diverse areas as health care, transportation, retail sales, natural resource management, and manufacturing. If an organization processes visual data as part of any of their key workflows, it's likely that AI inference workloads could help them extract additional value from that data in new ways, process more data, and do it all in less time. Image classification technology can open the door to exciting new efficiencies, ways of doing business, and revenue centers, but because inference workloads require serious computing power, it’s vitally important to have server hardware with enough muscle to do the job well.
To see how well a Stratus ztC Endurance server could handle AI inference-based image classification tasks, we used a ResNet-50 image classification workload to assess the throughput and latency performance of a Stratus ztC Endurance 7100 server at three levels of precision—FP32, Bfloat16, and INT8. FP32 precision reflects a level of classification accuracy that’s appropriate for tasks such as medical imaging or semiconductor manufacturing. Bfloat16 precision typically takes less time and memory space than FP32, but it still works well for tasks like weather modeling. INT8 precision provides the lowest level of precision, but it’s the fastest of the three and is suitable for scenarios where speed is critical, such as object recognition in transportation contexts.
Across all three precision levels, we found that the Endurance 7100 server offered strong throughput and low latency performance for CPU-based inference work, showing that it’s a viable platform for a variety of AI image classification tasks. This means that whether your organization’s work requires extreme precision or values speed, the Stratus ztC Endurance 7100 is a strong platform that may meet your AI image classification needs.
To see more about our Stratus ztC Endurance 7100 AI inference performance tests, check out the report below.
Principled Technologies is more than a name: Those two words power all we do. Our principles are our north star, determining the way we work with you, treat our staff, and run our business. And in every area, technologies drive our business, inspire us to innovate, and remind us that new approaches are always possible.
