Samsung GPU Performance Architect Memory System 

United States, Texas, Austin 

37531129

Role and Responsibilities

As a GPU Performance Architect (Memory System), you will work as part of the GPU Architecture team where you will drive the modeling and analysis of memory-system features for a highly efficient mobile GPU. You have a curious mindset that thrives on navigating the unknown through innovation and continuous learning. You will contribute towards current and future plans strategy.

We believe in connecting your area of expertise with the right level and functional discipline that can empower you to grow. You will have the unique opportunity to explore and contribute to the graphics pipeline, while broadening your knowledge in different aspects of GPU development.

• You work closely with the architecture, SW, and design teams to understand the architecture and micro-architecture of the GPU memory system including cache hierarchy, bus unit, interfaces with the GPU core and rest of the SoC.
• You develop complex GPU performance models to help define micro-architectural features and implementation optimizations of next-generation GPUs, in areas of memory system.
• You develop tools and methodology to correlate and validate performance models, triage and fix performance issues, and identify bottlenecks and propose solutions to improve GPU performance.
• You drive high-level performance analysis on complex workloads, with the goal to optimize the GPU with rest of the Memory System IP (Interconnect, Last-level Cache, Memory Controller).

Skills and Qualifications

• 10+ years of experience with a Bachelor’s Degree in Computer Science/Engineering, or 8+ years of experience with a Master’s Degree, or 6+ years of experience with a PhD
• 5+ years of experience in architectural modeling and performance analysis, either with analytical models or cycle-accurate simulators
• Experience with CPU, GPU, or Memory systems microarchitecture
• Good programming and/or scripting skills (Python is a plus)
• Good understanding of power/performance trade-offs
• Good understanding of VLSI concepts, HW design
• Good written and verbal communication skills

• Performance modeling or workload analysis experience, in areas of memory subsystems
• Knowledge of Core Pipelines (CPU, GPU or Accelerators)
• Directly related experience with correlation of RTL with performance model
• Experience in integrating internal/external IP models on a full system-level framework.

U.S. Export Control

This position requires the ability to access information subject to U.S. export control restrictions. Applicants must have the ability to access export-controlled information or be eligible to receive a government authorization to access export-controlled information.