2024 Wrapped – Redefining CPU Performance

As we wrap up 2024, it’s clear this has been a pivotal year for Flow —one defined by our disruptive innovation taking shape, establishing ourselves in the industry, and maintaining a steadfast focus on redefining CPU performance. From our spin-off from VTT Technical Research Centre of Finland in 1Q/2024 to strategic collaborations with RISC-V and building touchpoints in the industry, we’ve taken critical steps toward ushering in a new era in CPU performance. Let’s revisit the key moments that shaped our journey this year.

Q1: Our disruptive innovation takes place

Flow was founded during the first quarter of 2024 with a singular mission: to address the bottlenecks in traditional CPU performance. On March 1st, 2024 our founding team—Jussi, Martti, and Timo—became the first employees of Flow, with 110% sole focus on the company’s future success. Over the last few decades, CPUs have seen only incremental improvements, often limited by sequential architecture, scalability issues, and energy inefficiencies.

Transforming CPU performance with the PPU

The PPU reimagines CPU architecture by delivering scalable performance boosts:

• ≥2X boost for existing software with parallelizable workloads, dynamically optimized by the PPU compiler.
• ≤100X boost for applications explicitly designed for the PPU, leveraging parallel algorithms and shared memory interactions.

With full backward compatibility, an immediate ‘out-of-the-box’ CPU performance boost is unlocked through simple recompilation. The PPU’s seamless integration with all major instruction set architectures (RISC-V, x86, ARM) enables effortless scalability from mobile devices to supercomputers.

Amin Vahdat, VP/GM of Machine Learning, Systems, and Cloud AI at Google Cloud, explains the importance of continued CPU innovation:

"While our investments in compute accelerators have transformed our customers’ capabilities, general-purpose compute (=CPU) is and will remain a critical portion of our customers’ workloads. Analytics, information retrieval, and ML training and serving all require a huge amount of compute power. Customers and users who wish to maximize performance, reduce infrastructure costs, and meet sustainability goals have found that the rate of CPU improvements has slowed recently. Amdahl’s Law suggests that as accelerators continue to improve, general-purpose compute (=CPU) will dominate the cost and limit the capability of our infrastructure unless we make commensurate investments to keep up."

(Source: Google Cloud Blog, April 9, 2024)

The PPU directly addresses these pain points, unlocking untapped potential across industries and enabling significantly improved CPU performance. By enabling high-performance general-purpose parallel processing, the PPU empowers industries to meet the growing demands of modern workloads efficiently and effectively.

Validating PPU technology with industry partners

Throughout 2024, Flow has collaborated with several industry leaders to validate improvements in CPU performance and gather valuable feedback for the development of PPU Commercial Beta. These insights have been instrumental in shaping the PPU’s role and positioning as an accelerator for general-purpose parallel computing performance. 

A growing team, united by innovation

2024 has been a year of dynamic growth and team building at Flow. Our team has expanded to 10 skilled professionals, representing 5 nationalities residing in 5 different European countries, with a rich diversity of expertise. This multicultural collaboration fuels our ability to drive innovation and redefine CPU performance for a global audience.

By fostering a culture of creativity and inclusivity, Flow is laying the foundation for transformative advancements in parallel processing technology—one breakthrough at a time.

Throughout 2024: Big steps in PPU development

Flow’s team has been actively developing PPU architecture, performance model, and compiler toolchain throughout the year. These advancements are instrumental in shaping PPU’s scalability, efficiency, and integration, ensuring it delivers its promise on transformative improvements in general-purpose computing. 

June: Emerging from stealth mode

In June 2024, Flow officially emerged from stealth, unveiling its transformative PPU and securing a €4M funding round led by Nordic investors. This milestone underscored Flow’s mission to address one of computing’s greatest challenges: the stagnation of CPU performance improvements.

Our licensable PPU architecture seamlessly integrates parallel processing cores into any CPU design, delivering practical and scalable solutions for industries that demand high-performance computing. From cloud computing to industrial IoT, the PPU is poised to redefine what’s possible in general-purpose compute power.

August: Showcasing PPU at Hot Chips

In August 2024, Flow participated in Hot Chips 2024, presenting the PPU in this annual  symposium on high-performance chips. Our founders demonstrated the design principles behind the PPU and highlighted the results achieved with the PPU proof-of-concept. This event sparked several valuable conversations and fostered close connections across the industry.

September-December: Engaging through the Flow Podcast

From September to December, we released a 6-episode podcast series featuring Dr. Jörg Keller, offering an in-depth look at the PPU’s potential. Topics included:

• Simplifying parallelization with PPU cores for greater efficiency
• Enabling performance improvements for both legacy and modern software
• Reducing latency and improving energy efficiency in demanding applications

The series also highlighted Flow’s ongoing work in developing its architecture and performance model, emphasizing the integration of PPU technology alongside CPUs to unlock new levels of computing performance.

October: Strategic collaboration with RISC-V International

In October, Flow announced its strategic membership with RISC-V International and revealed that we are initially working on a RISC-V version of the PPU. This collaboration underscores Flow's commitment to open innovation and to creating scalable solutions that enhance CPU performance for diverse applications while advancing our architecture and performance model.

November: Connecting with the world at Slush 2024

In November, our CEO and co-founder, Timo Valtonen, represented Flow at Slush 2024, a world-renowned startup event. The event provided a platform to share Flow’s vision, build connections, and engage in inspiring discussions with industry leaders and potential investors. Slush highlighted the growing global interest in the transformative potential of PPU technology and confirmed that our planned development steps are spot-on.

Looking ahead to 2025

As we work to finalize the commercial configuration of our technology, we plan to launch the PPU Performance Model at the end of H1 2025. This milestone will mark a significant step toward delivering scalable, practical solutions for industries that demand high performance and efficiency. We will share progress updates along the journey towards the PPU Performance Model.

Thank you for an incredible year

We are deeply grateful to our investors, partners, team members, and supporters for believing in our vision. Together, we are building a future where computing performance is not just incremental—it’s revolutionary and parallel. :-)

Here’s to 2025 and the continued evolution of parallel processing with Flow!

Discover what’s next

Join us in shaping the future of parallel processing. Follow us on LinkedIn for the latest updates, industry insights, and innovations.

Contact us to learn more about Flow and PPU.

Email: info@flow-computing.com.