[Internship] Innovative HPC Performance Analysis
❓ About SiPearl…
Founded in 2019 with the support of the European Union 🇪🇺, SiPearl embodies Europe's dream of mastering the technological heart of its supercomputers: the microprocessor.
SiPearl is building Rhea, the high-performance, low-power European microprocessor dedicated to supercomputing and AI inference.
This new generation of microprocessors will first target EuroHPC Joint Undertaking ecosystem, which is deploying world-class supercomputing infrastructures in Europe for solving major challenges in medical research, artificial intelligence, security, energy management and climate with a reduced carbon footprint.
SiPearl is working in close collaboration with its 30 partners from the European Processor Initiative (EPI) consortium - leading names from the scientific community, supercomputing centers and industry - which are its stakeholders, future clients and end-users.
SiPearl employs more than 190 people in:
- France (Maisons-Laffitte, Grenoble, Massy, Sophia Antipolis),
- Germany (Duisburg)
- Spain (Barcelona)
- and Italy (Bologna)
SiPearl is part of French Tech Next 120 programme 2024 class.
💻 About the Internship
In this role, you will work closely with Olivier Deprez, our System & Software Architecture Team Leader
🎯 What a regular day at the job might look like
Recent ARM arch based HPC CPUs are based on coherent mesh network (CMN) to support communication between the Cores, the memory, and the other units (Storage, PCIe,…). This mesh network is the principal physical link between those units, transporting data with high bandwidth. On the top of the CMN, a transport layer called CHI is used to transmit flits of data from one point to the other. The CMN is a critical part when considering performances on HPC context.
Performance Analysis is an important topic in HPC because it is the first step to perform when optimizing and accelerating a specific program/application for a specific architecture. Different methodologies and technics exist for Performance Analysis, and there are often based on specialized kind of hardware unit called (Performance Monitoring Unit) Core PMU. The Core PMUs are monitoring events of the CPU cores itself. Counting and aggregating those events over multiple cores gives a first level of analysis methodologies. For example, top-down methodology is based on Core PMUs. The CMN itself is also equipped with different kind of PMUs, however, no clear methodology exists to exploit CMN PMUs.
The goal of the internship is to develop a methodology for CMN.
🔍 What would make you succeed in this role
- A good understanding of modern CPU architecture
- Good learning capacity
- Good writing skills
- Familiar with Linux dev environment & tools
- Familiar with essential dev tools like gcc, git, make, cmake
- Familiar with C language
- Fluency in English is a prerequisite for success in this role, as effective communication in English is essential for collaboration and ensuring the seamless execution of responsibilities.
➕ Experience in the following is a plus
- Knowledges on ARM architecture
- Knowledges on HPC
- Knowledges on Benchmarking
- Knowledges on CMN
- Knowledges on performance optimization & profiling
- Knowledges on ARM System & Software stack
- Knowledges on ARM Applicative CPU architecture
Dear candidate, even if you consider you do not fulfill all the qualifications mentioned above, please still apply and share with us why you believe you would be a good fit.
Recruitment process
✔️Discovery interview with our Talent Acquisition Partner (30 min)
✔️ AssessFirst personality test - no need to worry, there is no wrong or right answer; our goal is to see beyond your resume (45')
✔️ 1 Technical interview (1h)
💡 Benefits and conditions :
- Contract: Internship agreement (End of Studies)
- Benefits: Meal vouchers (60% covered by SiPearl); 3 days paid leave days during your Internship 🌴
- Work model: On-Site
- Duration: from 4 to 6 months
- 📍 Office: This role is open for Maisons-Laffitte or Massy (Paris metropolitan region)
- Awesome activities such as: Hackathons, Training Challenges, Quarterly Kick-off sessions, team events, Company events and much more
Are you curious to learn more about us?
- Follow us on LinkedIn to stay updated with our latest news
- Get updated about our recruiting activities: Sipearl careers webpage
- Get to know a bit more about our team culture: Welcome to the jungle
At SiPearl, we are dedicated to building a diverse and inclusive workplace that thrives on the strength of varied perspectives and backgrounds. We recruit talent based on merit, experience, and alignment with our company's goals and values.
- Department
- Software
- Role
- Architecture
- Locations
- Maisons-Laffitte, Massy
- Remote status
- Hybrid
About SiPearl
Founded in 2019 with the support of the European Union, SiPearl embodies Europe's dream of mastering the technological heart of its supercomputers: the microprocessor.
SiPearl is building Rhea, the high-performance, low-power European microprocessor dedicated to supercomputing and AI inference.
This new generation of microprocessors will first target EuroHPC Joint Undertaking ecosystem, which is deploying world-class supercomputing infrastructures in Europe for solving major challenges in medical research, artificial intelligence, security, energy management and climate with a reduced carbon footprint.
SiPearl is working in close collaboration with its 30 partners from the European Processor Initiative (EPI) consortium - leading names from the scientific community, supercomputing centers and industry - which are its stakeholders, future clients and end-users.
SiPearl employs more than 200 people in:
- France (Maisons-Laffitte, Grenoble, Massy, Sophia Antipolis),
- Germany (Duisburg)
- Spain (Barcelona)
- and Italy (Bologna)
SiPearl is part of French Tech 120 program 2024 class.
[Internship] Innovative HPC Performance Analysis
Loading application form