דרושים » הנדסה » Senior CPU Architect, Google Cloud

Contribute to CPU front-end designs, emphasizing micro-architecture and RTL design for next generation CPU.
Propose performance enhancing micro-architecture features with efficiency in mind. Work with architects and performance teams for trade-off studies.
Deliver designs that meet power, performance, and area (PPA) goals with production quality.
Interpret techniques into design constructs and languages in order to provide guidance to and participate in the performance modeling effort.

Contribute to CPU frontend designs, emphasizing on microarchitecture and RTL design for the next generation CPU.
Propose performance enhancing microarchitecture features with efficiency in mind. Work with architects and performance teams for trade-off studies. Communicate pros and cons of microarchitecture enhancements. Facilitate final decision making.
Deliver designs meeting Power, performance, and area (PPA) goals with production quality.
Become familiar with techniques for at least one processor functional block. Interpret the techniques into design constructs and languages in order to provide guidance to and participate in the performance modeling effort.

Design, develop, test, deploy, maintain, and improve Central Processing Unit (CPU) software modeling and other software tools.
Manage project priorities, deadlines, and deliverables.
Collaborate with hardware and software CPU architecture teams, SOC performance modeling team, and other Google Software teams.

Define architecture and micro-architecture features, write specifications, and understand implementation tradeoffs (e.g., performance, power, frequency, etc.).
Define the CPU block level design document (e.g., interface protocol, block diagrams, transaction level flow, control registers, pipelines, etc.).
Perform RTL development process (e.g., coding and debug in Verilog, SystemVerilog, or VHDL), function/performance simulation debug, and Lint/CDC/FeV/PowerIntent checks.
Contribute to the SoC level integration, and participate in synthesis, timing/power closure, and silicon bring-up.
Participate in test plan and coverage analysis of the block and SoC-level verification.

We're proud to be our engineers' engineers and love voiding warranties by taking things apart so we can rebuild them. We keep our networks up and running, ensuring our users have the best and fastest experience possible.

Responsibilities
Design, develop, test, deploy, maintain, and improve CPU software modeling and other software tools.
Manage individual project priorities, deadlines, and deliverables.
Collaborate with hardware and software CPU architecture teams, SOC performance modeling team, and other Google Software teams.

Develop methodology of SoC roll up of power reduction, projection, configuration, test plan and tools.
Work intact with Architecture, Frontend and Backend teams driving power reduction features (both logic and circuit).
Support power delivery and packaging teams with simulations and scenario definitions.
Work intact with Architecture and DV to define power scenarios and tests, debug, and integrate into the flow.
Track power goals are met throughout execution.

Collaborate with Architecture, Design and Verification teams to develop new product bring-up, validation, characterization, qualification strategies and manufacturing test solutions for new High Performance Computing (HPC) products in advanced process technologies.
Verify test solutions on pre-silicon models (simulation or emulation) and develop ATE test modules, DC tests, binning, production and characterization flows.
Develop and validate test programs on Automated Test Equipment (ATE) platforms for new product integration (NPI) in preparation for high volume manufacturing (HVM), working with ATE vendors.
Support product, volume data analysis of screening and characterization data, test time and yield improvements, assess test escapees and Return Materials/Merchandise Authorizations (RMAs), localize failures, implement containment measures and partner with design, manufacturing, quality and reliability teams to root cause and implement corrective actions.
Develop tools, flows and methodologies to improve and automate the testing.

Develop and execute comprehensive validation plans for Google's custom silicon, covering functional, performance, power, and reliability aspects.
Design and build scalable validation test infrastructure, including hardware setups, software frameworks, and automation tools on Emulation and/or FPGA platforms.
Lead the debugging and resolution of complex silicon issues, collaborating with cross-functional teams such as design, architecture, software, and firmware.
Analyze validation data to identify trends, root causes, and opportunities for improvement in silicon quality and reliability.
Build and mentor a high-performing team of silicon validation engineers, fostering a culture of collaboration, innovation, and technical excellence.

We're proud to be our engineers' engineers and love voiding warranties by taking things apart so we can rebuild them. We keep our networks up and running, ensuring our users have the best and fastest experience possible.
Responsibilities
Define the System on a Chip (SoC)/block level design document such as interface protocol, block diagram, transaction flow, pipeline, etc.
Perform Register-Transfer Level (RTL) development (e.g., coding and debug in Verilog, SystemVerilog), function/performance simulation debug and Lint/CDC/FV/UPF checks.
Participate in synthesis, timing/power closure and Application-specific integrated circuit (ASIC) silicon bring-up.
Participate in test plan and coverage analysis of the block and SoC level verification.
Communicate and work with multi-disciplined and multi-site teams.

We are seeking an experienced Runtime Software Architect to lead the design and development of an advanced runtime framework tailored for our servers. This role involves analyzing existing runtime systems and machine learning runtime frameworks such as ONNX Runtime onnxruntime.ai, TensorFlow Serving, PyTorch's TorchServe, Apache TVM tvm.apache.org, and Glow github.com, to inform the creation of a new framework that efficiently leverages low-level hardware components, including DSPs, video decoders, ARM cores, and DLAs. The ideal candidate will possess a deep understanding of these technologies and a proven track record in architecting software that harmonizes with complex hardware infrastructures.

Key Responsibilities:
Framework Analysis: Conduct comprehensive evaluations of existing runtime systems and ML runtime frameworks, such as ONNX Runtime, TensorFlow Serving, TorchServe, Apache TVM, VLLM, TRT-LLM, and Glow, to identify their strengths, weaknesses, and applicability to our objectives.
Architecture Design: Develop a robust runtime framework optimized for our servers, ensuring seamless integration with hardware components such as DSPs, video & audio decoders, ARM cores, and DLAs.
Hardware Integration: Collaborate with hardware engineers to adapt and optimize software components, maximizing performance and efficiency across various hardware modules.
Performance Optimization: Implement strategies to enhance system performance, including reducing latency, increasing throughput, and improving resource utilization.
Cross-Functional Collaboration: Work closely with cross-functional teams to align software architecture with business goals and technological advancements.