Skip to main content

A breakthrough technology could make CPUs 100x faster

The PPU, as imagined by Flow Computing.
Flow Computing

There are bold claims, and then, there are bold claims — Flow Computing just made the latter. According to the startup, its proprietary tech, which it refers to as a parallel processing unit (PPU), can boost the performance of any CPU by up to 100 times.

It doesn’t even have to be one of the best CPUs — Flow Computing claims that all devices in need of high-performance CPUs will find massive benefits in using the company’s PPU. There’s more to it that sounds pretty amazing, but what are the odds that it’ll actually make it to market? Let’s take a closer look.

Recommended Videos

Flow Computing only came into being in January of this year; this is a new startup built on the foundation of the VTT Technical Research Centre of Finland. The company — which appears to only employ three of its founders right now, although the FAQ doesn’t make it entirely clear — has already gathered around $4.3 million (4 million euros) in funding. Though the headcount and the age of the company are both on the shorter side, its tech sounds mighty impressive.

The company states that shifting workloads to the PPU would enable the era of new “super CPUs.” These chips would be fully backward compatible with existing software and apps, and what’s more, developers wouldn’t need to do a thing for these new PPU-infused processors to work. They would just work, and 100 times faster, at that.

In the FAQ document that the company has shared with The Verge, it said: “Parallel functionality in legacy software and applications can be accelerated by recompiling them for the PPU without any code changes. The more parallel functionality there is, the more performance boost will be subsequently gained. Our technology is also complementary in nature — while it boosts the CPU, all other connected units (such as matrix units, vector units, NPUs, and GPUs) will indirectly benefit from the performance of PPU and get a boost from the more capable CPU.”

The architecture of the PPU.
Flow Computing

Add to that the fact that Flow Computing also believes that its PPUs could bring this kind of performance boost even to smaller devices, such as smartphones and watches, and we’ve got something that sounds like a real hit on our hands. However, the problem is that Flow Computing hasn’t built such a chip yet, so these claims are difficult to verify.

Instead of tackling this project by itself, the young startup hopes to team up with the likes of Intel, AMD, Arm, Qualcomm, and Nvidia, as well as smaller CPU startups in order to “co-develop the future era of advanced CPU computing.” The startup is targeting a global market and planning to license its technology all around the world. “[The] PPU is totally independent of instruction set design, so it can be used in any CISC or RISC CPU and it can be integrated into any current or pending design architecture using any process geometry,” according to the company.

Could this truly be the beginning of the future of computing? Will we start seeing CPUs that are incredibly fast, as opposed to the slow death of Moore’s Law that we’ve been witness to over the past years? We don’t have the answers, but check out the white paper and the article published by the founders if you want to decide for yourself.

Monica J. White
Monica is a computing writer at Digital Trends, focusing on PC hardware. Since joining the team in 2021, Monica has written…
Black Friday’s best PC hardware deal is still live, and you’re sleeping on it
The Ryzen 5 7600X sitting among thermal paste and RAM.

I'm not mad, just disappointed. A couple of weeks ago, I covered the insane deal that essentially allowed you to score a Ryzen 5 7600X -- still one of the best processors you can buy -- for just $105. At the time, I thought, surely, this will sell out in a matter of hours. Who would pass up on a deal this good? And yet, two weeks later to the day, the craziest deal I've seen during all of Black Friday and Cyber Monday is still live on Newegg.

Let me break down the deal again. You can get the Ryzen 5 7600X for $225, which is not a good price. However, you can get an additional $30 off by using promo code DLCDZ342, bringing the price down to $195. The kicker is that you also get a free Team Group MP44L 1TB PCIe 4.0 SSD. That's a $90 hard drive that Newegg is just throwing in with a CPU that's already available for a decent price. The fact that the deal is still live suggests either Newegg has a ton of inventory, or not enough people know about this sale.

Read more
NZXT dismisses PC rental allegations as ‘misconceptions’ while promising changes
The NZXT H7 Flow refreshed PC case showcased at Computex 2024.

NZXT founder and CEO Johnny Hou has publicly addressed growing criticism of the company’s Flex gaming PC rental program, which faced intense scrutiny after YouTube channel Gamers Nexus exposed significant flaws in its pricing and terms. In a detailed video, Gamers Nexus described the program as exploitative, pointing out that its long-term costs far outweighed the hardware's value, leaving customers locked into a financial commitment with minimal ownership options.

“I want to acknowledge that we messed up,” Hou said in a video published by the company. He also promised to address customer concerns and improve the program but offered few specifics on what changes would be implemented.

Read more
ChatGPT’s new Pro subscription will cost you $200 per month
glasses and chatgpt

Sam Altman and team kicked off the company's "12 Days of OpenAI" event Thursday with a live stream to debut the fully functional version of its 01 reasoning model, as well as a new subscription tier called ChatGPT Pro. But to gain unlimited access to these new features and capabilities, you're going to need to shell out an exorbitant $200 per month.

The 01 model, originally codenamed Project Strawberry, was first released in September as a preview, alongside a lighter-weight o1-mini model, to ChatGPT-Plus subscribers. o1, as a reasoning model, differs from standard LLMs in that it is capable of fact-checking itself before returning its generated response to the user. This helps such models reduce their propensity to hallucinate answers but comes at the cost of a longer inference period and slower response.

Read more