What you do at AMD changes everything 
 

At AMD, we push the boundaries of what is possible.  We believe in changing the world for the better by driving innovation in high-performance computing, graphics, and visualization technologies – building blocks for gaming, immersive platforms, and the data center. 
 

Developing great technology takes more than talent: it takes amazing people who understand collaboration, respect, and who will go the “extra mile” to achieve unthinkable results.  It takes people who have the passion and desire to disrupt the status quo, push boundaries, deliver innovation, and change the world.   If you have this type of passion, we invite you to take a look at the opportunities available to come join our team.
 

Principal SoC Architect, Discrete GPU SoCs - 113062

The Role:

AMD is looking for an outstanding technical contributor to help optimize performance of next-generation Discrete GPU SoCs. As a passionate and dedicated Performance Architect in the SOC Architecture Team, you will optimize SOC performance by identifying opportunities across all parts of system architecture, spanning the application and related frameworks, driver, GPU core, memory & I/O subsystems.  You will use performance models for micro architecture trade-offs, collect and analyze performance profiling data and traces to isolate performance bottlenecks, and share the insights to help drive features into the next generation of GPUs. The ideal candidate is expected to be well informed on latest trends in GPU and CPU Systems, and be able to quickly ascertain, in a data-driven manner, how next generation GPUs and platforms should be engineered to support those needs.

Key Responsibilities:

SoC level Performance analysis for multiple SoCs in Discrete GPU product line.

Has technical time horizon of generally 6 to 18 months.

Explore and evaluate architectural design choices in Fabrics, Caches, and IPs in the SoC.

Work closely with other HW and SW architects to understand the architecture, the workloads and to propose solutions to improve/enhance performance with given SoC/IP and markets.

Architectural modeling (as needed), performance triage, deeper analysis, identifying bottlenecks and solutions for performance improvement.

Proposing, communicating, and implementing solutions to issues.

Participate in new workload definition and/or workload optimizations.

Deep engagement with Performance modelling team for new features development, developing performance analysis plan in Architecture phase, analyzing results and model correlation.

Work closely with Design and Performance Verification teams to set Performance KPIs, Performance Verification test plan reviews and triaging Performance issues identified by them.

Work closely with Post-Si teams for correlation of Pre-Silicon models and assumptions, and support them for Performance validation issues.

Position Requirements

Exceptional foundation in systems architecture, cutting across CPU or GPU, memory, storage and I/O subsystems

Experience in using Performance Modeling and Simulation Tools for CPUs, GPUs or SoC components at different abstraction levels (Functional, TLM or Cycle Accurate)

Experience analyzing CPU, GPU or System-level Micro-Architectural features to identify performance bottlenecks within different workloads

Experience with Performance monitoring tools such as Hardware performance counters and Visualization tools

Excellent communication skills (verbal, written, and presentation)

Exposure to GPU performance benchmarks and workloads are a plus

Proficiency in Unix, C++, and Scripting languages such as Python is a plus

More than 8 years relevant industry experience