Sr Staff Engineer , Silicon Platform Software

What You’ll Do

The Advanced Controller Development (ACD) Lab is part of Samsung’s Memory Business Unit, the industry's all-time DRAM and NAND Flash leader both in technology as well as in volume. ACD’s vision is to solve critical bottlenecks in Cloud & Data center, Automotive, and even edge-devices by developing new approaches via memory/storage controllers, high-speed interfaces, and advanced architectures. We are an integral part of Samsung’s paramount R&D innovation engine. We work closely with cross-disciplinary development teams to bring feature innovation to product roadmaps. Come and join the team that is creating the new paradigm shift from a traditional CPU-centric to Memory-centric computation architectures needed to support emerging machine learning applications, data analytics, and edge computing.

• Contribute to the new CPU architecture and microarchitecture definitions with a partnership with micro-architects, hardware and software teams to research innovative ideas
• Assess these ideas by using the performance analysis infrastructure you and your team develop
• Profile the workloads and analyze their key performance characteristics, conduct performance analysis and debug performance bottlenecks
• Build processor performance analysis infrastructure, which includes preparing workloads, developing performance models, porting a design onto a hardware emulator, and conducting performance and workload analysis.
• Work with the RTL team to correlate the performance model with the RTL.
• Collaborate with key micro-architects to develop and assess microarchitecture ideas.
• Select appropriate industry standard processor benchmarks for goal setting. Define and produce custom in-house benchmark suites using 1st party applications.
• Participate in PPA (Performance, Power, Area) tradeoff analysis for architecture and microarchitecture features. Communicate analysis results in both qualitative and quantitative fashion to support major decisions
• Perform system analysis and optimize actual workloads (Data search, Video application, ML workloads, etc.) on various computation IPs and chipsets while keeping up on the power constraints
• Optimize to realize the best in class sustained data processing performance (perf/Watt) on Linux kernel and custom FW platform.
• Use various tools to identify performance bottlenecks in the system and identify related optimizations such as vectorization, hardware acceleration, etc.
• On the system/kernel side you will work towards optimizing the power saving algorithms for the underlying SoC architecture.
• Collaborate and participate in design/code reviews with various cross-functional teams
• A successful candidate is self-driven and motivated to explore across the software stack - userspace/kernel on Linux and/or acceleration engines (its sw architecture/threading model etc.)
• Identify and influence SoC HW recommendations for real life workloads collaborating with HW teams.
• Invent and file patents on technical solutions to the problems identified

What You Bring

• BS + 15 years experience, MS + 13 year of experience, PhD + 10 years experience in Electrical Engineering, Computer Science or related.
• Experience in any of programming languages such as C, C++, Java, Python.
• Knowledge of CPU/SoC architecture
• Self-motivated problem-solver with an ability to work well in a team.
• Direct experience with performance modeling, performance analysis, and workload characterization.
• Knowledge in system software components, such as Linux, drivers, and runtime.
• Debugging skills on multi-processor environments using tracing tools such as systrace/ftrace/perf
• Strong analytical skills
• Experience in SIMD programming
• Experience optimizing benchmarks, user experience and memory on Linux and FW
• Deep understanding of PPA (performance, power, and area) trade-offs.
• Familiarity with digital design and verification flows.
• Deep understanding of flash memory controller architecture, and related IP design, verification, SOC implementation process and tools.
• Understanding of the emerging technologies (CXL, Computation in memory and storage, Ethernet-attached SSD etc.) in server memory and storage systems 
• Experience with performance modeling of embedded systems is beneficial.

#LI-BG1