Virtually every smartphone available today has a processor based on ARM architecture, and ARM-based chips are also widely used in IoT, automotive, and other applications.
But one of the biggest players in the ARM-based processor space has just announced it’s teaming up with other chip makers in hopes of “advancing adoption of RISC-V globally by enabling next-generation hardware development.”
Qualcomm says says it’s working with NXP Semiconductors, Infineon Technologies, Nordic Semiconductor, and Bosch to “jointly invest” in a new company that will be based on Germany.
According to Qualcomm, the new company “will be a single source to enable compatible RISC-V based products, provide reference architectures, and help establish solutions widely used in the industry.”
At launch the company will focus on solutions for the automotive industry, but eventually the goal is to develop products for IoT and mobile devices as well. In other words, Qualcomm, which makes the chips that power many of the best Android phones, may eventually transition (at least partially) to making RISC-V chips.
There’s a lot to unpack here. The first thing is that ARM processors are currently way ahead of RISC-V chips when it comes to performance-per-watt, software support, and an overall ecosystem. So companies making RISC-V chips have some catching up to do before we see flagship-class phones with RISC-V chips rather than ARM-based processors.
But there are a number of reasons that companies like Qualcomm may be looking for alternatives to ARM. As Ars Technica points out:
- ARM’s parent company has been looking for ways to get more money out of ARM. After the planned sale of ARM to NVIDIA fell through, there have ben indications that ARM will start charging higher license fees from customers like Qualcomm.
- RISC-V architecture, meanwhile, is an open standard instruction set that’s available royalty-free. In other words, companies or individuals can design all the RISC-V chips they want without paying a penny to RISC-V International, the organization behind the chip architecture.
So investing in RISC-V is a way for chip makers to save money while hedging their bets in case ARM raises prices, starts cracking down on licensing terms in some other way, or just flat out goes belly up.
We have already seen some interesting developments in the RISC-V space in the past few years, with companies like SiFive developing ever-more-powerful processors and dev boards, including the HiFive Pro P550 that was designed in partnership with Intel, as well as a growing number of affordable Raspberry Pi-like single-board computers with RISC-V chips starting to pop up, including the BeagleV-Ahead, Sipeed Lichee Pi 4A, and boards from Milk-V and MangoPi, among others.
Pine64 is even selling a PineTab-V tablet with a RISC-V processor.
It’s too early to say whether RISC-V could one day have the kind of market dominance that ARM currently enjoys. But with a growing number of mainstream chip makers expressing interest, it’s not as hard to imagine as it once was.
keep the qualxom out
I’m very suspicious that qualcomm is very interested in Risc-V. I see them making Risc-V a closed-sourced, don’t-know-what-you’re-really-getting-spy-chip-hidden-inside kind of chip. Just like how Microsoft “loves Linux” so much. I just don’t trust it.
RISC-V is an architecture, not a chip. Maybe try using terms right if you want to not look like a complete idiot.
One option you might consider is to read what they intended. They didn’t say that Qualcomm would completely change the architecture, but how they could build a chip from that architecture. So, in their theory, the product would in fact be a chip, not an architecture. What they suggest, while I don’t think it will happen, is possible. Acting like it is incorrectly stated when it is a valid sentence which could occur doesn’t help argue against it.
HAHAHAHA, right.
I agree to some degree. I think many of the people getting excited about RISC-V are misplacing their hopes.
Lots of people seem to think that RISC-V is going to be a new frontier of computer products that unshackle people from the chains of untrustworthy tech companies.
RISC-V’s apparent “open” nature isn’t necessarily for the user’s benefit. Its the companies who actually build these chips that are being given freedom. At the end of the day, these are all multi-billion dollar companies that are just trying to make more money.
Sure, we’re going to have some cool hobbyist products using RISC-V, but that’s not what’s going to be paying the bills for these chips getting made. The focus will always be money, and we aren’t being guaranteed anything.
These five biggies will do all the heavy lifting. When they are done with the work, Pooh Bear will get up off the cushy sofa, pick up the result, copy it, jack up the price, and get slaves to build stuff with it that gets sold back to them, and spies on them too. Idiots.
🙂 This is excellent news. I’d been waiting to see what Qualcomm were going to do. They expressed an interest in RISC-V in 2017, had RISC-V cores in their processors by 2019, and given the souring of the relationship with Arm, I’m surprised that they’re only just making this move now.
And I think RISC-V will catch up to ARM both in terms of hardware and software quicker than you think. See where it is in 2 years.
I could call it good news if they were willing to commit to a firmware standard that includes unlocked bootloaders, but realistically, RISC-V is going to be just as locked down as ARM if not moreso.
We’re already beginning to see published designs – https://github.com/openhwgroup/cv32e40p – and we’re going to see more performant open source processors going forward.
That’s going to start putting hitherto unheard-of pressure on design companies.
I highly doubt RISC-V is going to catch-up to ARM anywhere from the 2W to 40W power range. That’s for single-core, multithread, and overall efficiency.
Even if they get close theoretically, this won’t translate to real-world conditions. Software does take a decent time to optimise.
Anyone paying attention to the industry would know that ARMs resurgence was not a surprise. They’ve always been innovative in the ULP front, but they managed to push the envelope massively with the likes of the Cortex-A9, Cortex-A73, and the upcoming Cortex-X4. In contrast, RISC-V has been impressive but for different reasons.
As others have said, just because the core architecture is built upon open-source, does not mean the overall hardware won’t be riddled with locked-down proprietary additions, and the firmware being less open in nature by the end of it.
Some people have compared ARM vs RISC-V as something like a Dictatorship versus Communism. But it’s probably more akin to Aristocracy versus Anarchy. In the market of ARM processors we have providers from: ARM, Apple, Qualcomm, Samsung, Nvidia, and a few niche players (server market). In RISC-V there will be major players, smaller players, and individuals who can stamp and print their own processors in their own homes/service, but there wont be much interoperability and instead a lot of fragmentation around.
I have worked extensively with ARM based SOC’ from the outset, and was very impressed with the clarity of concept, lead by ARM Ltd in England.
By contrast, the European have been chronically late to adopt new technologies, and standards, to understand the importance of the related software and firmware techniques, grotesquely thinking they can buy it, rather than having real mastery there.
Will the European ever wake up ?
I welcome an alliance based around Bosh, and the automotive…
A change in thinking will take some time, and the laggards will not become tigers overnight.
Andre Gompel
good comment but ignores quick turn around times required in our ai designed chips future. think about it this way, you would a main processor / soc die and that would remain largely unchanged for third party software compatibility but that staid die, with just one layer of abstraction, there would be rapidly evolving chiplets / tiles
I can append on that that CPU architecture does not matter for a very long time already. What does is private licensed IP cores. You want efficient hardware video decoding/encoding? Only a dozen companies or so have enough expertise. Efficient iGPU design for specific power targets/expected scene types/resolution/API? Same story. Attempts on building open source GPU like MIAOW can give a general idea of the chasm between commercial solutions and open source ones.
Most of the moder CPU are co-processors and accelerators, just like old game consoles used hardware sprite pipelines and texturing units, while CPU were pretty weak: Z80, 68k or alike.
There are still niches for the RISC-V though, next to MIPS and SPARC.