Flow devises method to speed up central processing units by 100 times without rewriting all software

Finnish start-up Flow Computing has developed a computational unit called the Parallel Processing Unit (PPU), which it claims can vastly improve the performance of any central processing unit (CPU). The purpose of the PPU is to significantly increase processing power, potentially allowing for speed boosts of up to 100 times.

Despite the transformations of central processors since the era of vacuum tubes and punch cards, their basic limitations persist. CPUs are inherently single-threaded, meaning they execute tasks sequentially, one at a time. While they can perform billions of jobs every second using multiple cores and threads, the CPU’s bottleneck nature remains a cause for concern. Timo Valtonen, Co-Founder and CEO of Flow, vows to change this to improve overall computation efficiency.

Even with current rapid command processing capabilities, serial instruction execution comes with considerable wastage, as one task must finish before the next task commences. Flow claims that its PPU tackles this limitation effectively. A CPU will continue to execute one task at a time, but the PPU will manage task streams on a nanosecond scale, resulting in faster task transfers in and out of the processor, reports TechCrunch.

Crucially, the PPU facilitates efficient task distribution among cores at high speed, boosting computation without significant adjustments to existing program codes. The PPU neither escalates the clock frequency nor burdens the system in ways that could cause additional heating or increased power usage. Flow’s technology, successfully demonstrated on FPGA platforms, is now in talks with chip manufacturers to integrate the PPU into future processors. This could potentially result in revolutionary leaps in performance.

Analysts believe that one of the main obstacles Flow Computing may face is chip manufacturers’ reluctance to risk established development strategies for an untested technology. Flow Computing has recently raised 4 million euros from venture investors for further development. It is now the start-up’s task to convince the industry of the merits of its technology in satisfying the growing demand for computational power in the field of artificial intelligence.