One radical new company is aiming to once again alter the meaning of speed and efficiency in computing by giving central processing units (CPUs) a 100-fold boost in performance. Called Flow Computing, the newly formed Finnish innovator has just gone public with its parallel processing technology. Backed by $4.3 million in funding from Nordic venture capital investors, the Helsinki-based startup is likely to alter the face of computing worldwide.
In an age where incremental gains of nanometres in semiconductor and microchip design are trumpeted, Flow Computing envisions the kind of leap that makes people feel they are living in the next century. A leap all the more notable because the technology can be shoehorned into any CPU design, and because its sleek simplicity could enable the next era of computing: CPU 2.0.
‘It’s the most serious bottleneck in computing today,’ says Timo Valtonen, the co-founder and CEO of Flow Computing, whose startup wants to do away with it completely. Flow’s Parallel Performance Unit (PPU) promises 10 times the performance of existing CPUs. Flow’s co-founder and CEO Timo Valtonen shows off the startup’s first product, a telecoms network processor implementing their PPU architecture. Photo courtesy Flow Computing. Valtonen argues that CPU performance has stagnated over the past decade, while current technologies struggle to keep up with massive data volumes – a contradiction to Moore’s law.
What makes this exciting is that the newly designed architecture uniquely reinvents the CPU’s capability by mating it to Flow’s Parallel Processing Unit (PPU), ensuring the performance uplift that was previously a concept; which jump is not only a theoretical proof of concept but a very tangible leap in the evolution of the workloads of present and future computing needs (instantly recognisable in intense computation workloads such as AI, in addition to edge and cloud computing).
More than that, its universal compatibility, where the same technology can work across the entire CPU range, from mobile devices to PCs, and up the food chain to the world’s supercomputers, giving board-level, performance-boosting fail-safes to every tier of the CPU market – this is something entirely new. What is also new is how backwards software compatibility is achieved. Existing software can take advantage of this newfound power mostly untouched, with little or no changes needed.
The knock-on quantum boosts of Flow’s PPU also affect the CPU’s environment. Supercharging the CPU would cascade its effects to connected devices and peripherals – not only would their performance indirectly improve, but the synergistic rise in total system-performance would also amplify.
Flow’s work is just getting started. The team has already begun conversations with some of the world’s largest semiconductor companies. It would take too long to document all the technical details here, but I’ll say that much more will be revealed in the second half of 2024. Precisely, when, yet another paradigm shift for CPU performance will be sometime next year.
The company, whose investors include VTT Technical Research Center of Finland, isn’t just the pride of the Finnish tech industry, it’s poised to lead a global revolution in semiconductor design. The excitement in the tech community – among journalists, industry veterans, and investors – reflects the magnitude of Flow’s innovation and how it could change the level playing field not only for device architecture but everywhere computing performance is critical.
On the cusp of what could be the most significant revolution in CPU architecture and performance, Flow Computing’s technology opens the way for a whole new paradigm of computing. Flow is not just improving a technology. It is changing the way in which we think about CPUs because the technology allows us to redefine the very metrics by which we judge computational performance and efficiency. With Flow, there isn’t just an edge. There is a new kind of edge: an edge in which there is no limit on what is possible.
© 2024 UC Technology Inc . All Rights Reserved.