macOS GPU Driver Engineer, Performance

Santa Clara Valley (Cupertino), California, United States
Software and Services

Summary

Posted: Nov 12, 2018
Weekly Hours: 40
Role Number: 113318586
The team is now seeking a top-notch and dedicated GPU driver Engineer who is passionate about creating great innovative products, and has a strong track record of building high performance and production quality software on an aggressive schedule.

Key Qualifications

  • The ideal candidate will have 3 plus years with the following skills and experience or a fresh graduate with interest in graphics.
  • Excellent programming skills and knowledge of C/C++.
  • User mode or kernel mode device driver development experience.
  • Excellent debugging skills (lldb, gdb) and capability to dive into different layers of the driver and macOS stack.
  • Experience with 3D Graphics and any GPU programming APIs like Metal, OpenGL, OpenCL, Vulkan, D3D.
  • Experience with driver, low-level programming or operating system development.
  • Excellent analytical and problem solving skills.
  • Strong communication and a good team player.

Description

You will work on the driver team to implement performance-focused Metal features for current and next generation macOS platforms. As part of this team you'll work closely with the HW Architecture team to provide feedback on performance of HW roadmap as well as cross-platform architecture comparison across products. You'll debug performance issues across the GPU stack, identify and develop performance optimizations within the GPU driver and across the stack. This role works with the Metal Ecosystem team to collaborate with external developers on best practices with the Metal API for optimized performance in their apps. Finally, you'll develop internal tools for performance analysis.

Education & Experience

BS/MS CS or similar technical degree

Additional Requirements

  • Knowledge of GPU architectures and 3D Graphics pipeline
  • Experience analyzing performance of 3D applications
  • Knowledge of performance strengths and deficiencies of different HW architectures is a plus.