GPU Compiler Performance Lead @ Qualcomm

Responsibilities

• Profile trending GPU benchmarks and applications (games, HPC, and AI applications)

• Use external and internal tools to analyze profiling result and identify performance bottlenecks

• Summarize profiling result and give presentations to customer and internal teams

• Propose improvements in compilers and GPU architecture to tackle identified bottlenecks

• Propose modifications to applications for better performance on Qualcomm GPUs

• Characterize GPU application trend and architectural requirement

• Share performance analysis and tuning experience within compiler team

Qualifications

• A BS / MS / PhD degree in Computer Science or Electrical Engineering

• 7+years of relevant experience.

• Compiler knowledge

• Understanding of computer architecture (CPU, memory, data layout, etc.) and performance tradeoffs

• Experience with C / C++, scripting languages (e.g. Python)

• Understanding of parallel computing on multi-core CPU, GPU, or heterogeneous systems

• Good communication skills and teamwork spirit, reliable and self-motivated

Plus

• Compiler development experience (proprietary or open source)

• Graphics programming (OpenGL, DirectX, or others) or OpenCL or CUDA application development

• Development of system-level performance tool

• Extensive experience in high-performance (parallel) computing

• Experience / understanding of profiling / debugging tools and hardware performance counters for parallel applications on multicore or many core architectures

• Experience with machine learning / deep learning tools (scikit learn, tensorflow, or others)

• Experience with benchmarking and performance tuning for parallel applications