Rtl Power Analysis Engineer

AMD AMD · Semiconductors · Austin, TX · Engineering

This role focuses on Register-Transfer-Level (RTL) power analysis and reduction for AMD's Infinity Fabric team, working on a wide range of products including AI accelerators, servers, laptops, and gaming consoles. The engineer will collaborate with system architects and designers to optimize power consumption, perform analysis using power tools, and design power management algorithms. While the role leverages AI tools for automation, its core function is in hardware design and power optimization, not AI model development.

What you'd actually do

  1. Work within the Infinity Fabric power team to drive power reduction including estimation, analysis, and optimization, flow setup, and methodology improvements.
  2. Collaborate with system architects and designers, define use cases, identify and prototype power and perf-per-watt optimizations.
  3. Run performance models and power tools, write scripts, analyze data, and evaluate tradeoffs against project goals.
  4. Power management algorithm design and implementation.
  5. Leverage latest AI tools and flows to drive automation and analysis

Skills

Required

  • RTL power reduction
  • ASIC power and low-power design techniques
  • power analysis and optimization tools
  • RTL design, simulation, and synthesis
  • performance, analyzing perf/power trade-offs
  • Scripting language, e.g. Python
  • analysis and problem-solving skills
  • well organized
  • self-driven
  • Excellent communication skills

Nice to have

  • good arch/uArch background
  • Knowledge in NOCs, memory controllers, SOC blocks
  • AI tools and flows

What the JD emphasized

  • strong interest and skill set in RTL power reduction
  • strong interest building up that context to drive effective power improvements
  • Strong experience in ASIC power and low-power design techniques
  • Strong experience in power analysis and optimization tools
  • Strong experience in RTL design, simulation, and synthesis
  • Experienced in performance, analyzing perf/power trade-offs