Transprecision computing lets processors match floating-point precision to application needs. FPnew demonstrates how a configurable, open-source FPU can support multiple formats, SIMD operations, and energy-proportional execution in RISC-V systems.
Articles Tagged: energy-efficient computing
Error-Efficient Computing: Trading Precision for Efficiency
Error-efficient computing reduces the effort spent preventing errors when an application can tolerate limited deviations. For automation teams, the approach offers a way to assess potential efficiency gains while protecting control functions and data that require strict correctness.
Energy-Efficient Computing: Research With Industrial Relevance
Research in energy-efficient computing spans chip design, thermal management, approximate hardware and machine-learning optimization. Sherief Reda’s work connects these areas, offering technical themes relevant to engineers building and managing increasingly compute-intensive industrial systems.
Ultra-Low-Power Sensor Systems for Industrial IoT
Research in ultra-low-power VLSI, adaptive computing and millimeter-scale sensor nodes points to design principles relevant to industrial IoT: manage energy carefully, account for variation and bring useful processing closer to the point of sensing.
Responsible AI in Industrial Automation: Key Lessons
NVIDIA’s 2019 CSR report connects technology development with governance, energy use, supply-chain practices and workforce priorities. Its approach offers useful considerations for industrial organizations assessing how to develop and deploy AI and automation responsibly.
Semiconductor Technology Priorities for Industrial Automation
Industrial automation depends on chips that can sense, process, store and communicate information reliably. A semiconductor industry roadmap identifies five research priorities with implications for smarter machines, connected systems, security and energy use.