Intel Invests In Open-Source RISC-V Processors With a Billion Dollars In New Chip Foundries (zdnet.com) 41
An anonymous reader quotes a report from ZDNet: RISC-V International, the global open hardware standards organization has announced that Intel has joined RISC-V at the Premier membership level. Let that sink in for a minute. Intel, which has made billions from its closed-source, complex instruction set computer (CISC) x86 processors, is joining forces with RISC-V, the open-source reduced instruction set computer (RISC) CPU group. What next? Dogs and cats living together!? Dr. David Patterson, co-creator of RISC-V helped create it to be an open lingua franca for computer chips, a set of instructions that would be used by all chipmakers and owned by none. Today, Patterson said, "I'm delighted that Intel, the company that pioneered the microprocessor 50 years ago, is now a member of RISC-V International."
Why? Because Intel sees a future in which ARM, x86, and RISC-V all play major roles. In particular, Intel has already seen strong demand for more RISC-V intellectual property (IP) and chip offerings. Intel's not just giving this idea lip service. Intel also announced a new $1 billion fund to support early-stage foundry startups. Together Intel Capital and Intel Foundry Services (IFS) will prioritize investments in chip IP, software tools, innovative chip architectures, and advanced packaging technologies. Randhir Thakur, IFS President, said this new program will focus on two key strategic industry points: Enabling modular products with an open chiplet platform and supporting design approaches that leverage multiple instruction set architectures including and spanning x86, Arm, and RISC-V.
As part of these initiatives, IFS will sponsor an open-source software development platform. This will provide IP for all three of the leading ISAs chip architectures. RISC-V has always been about providing open modular building blocks. Together Intel and RISC-V will expand the RISC-V ecosystem and help drive its commercialization. [...] Intel is already offering RISC-V chips: It's Nios V processors based on RISC-V. Moving ahead Intel hopes its new RISC-V investment will speed up RISC-V's development. Calista Redmond, RISC-V International's CEO, sees these moves as recognizing that "massive investment in open source has the power to change the course of history." Redmond went on to say that open collaboration with RISC-V has "ignited a profound shift in the semiconductor industry, and this partnership will accelerate innovation in open computing. RISC-V welcomes Intel and looks forward to our collective expansion and the commercial adoption of RISC-V across compute workloads and industries, growing RISC-V everywhere."
Why? Because Intel sees a future in which ARM, x86, and RISC-V all play major roles. In particular, Intel has already seen strong demand for more RISC-V intellectual property (IP) and chip offerings. Intel's not just giving this idea lip service. Intel also announced a new $1 billion fund to support early-stage foundry startups. Together Intel Capital and Intel Foundry Services (IFS) will prioritize investments in chip IP, software tools, innovative chip architectures, and advanced packaging technologies. Randhir Thakur, IFS President, said this new program will focus on two key strategic industry points: Enabling modular products with an open chiplet platform and supporting design approaches that leverage multiple instruction set architectures including and spanning x86, Arm, and RISC-V.
As part of these initiatives, IFS will sponsor an open-source software development platform. This will provide IP for all three of the leading ISAs chip architectures. RISC-V has always been about providing open modular building blocks. Together Intel and RISC-V will expand the RISC-V ecosystem and help drive its commercialization. [...] Intel is already offering RISC-V chips: It's Nios V processors based on RISC-V. Moving ahead Intel hopes its new RISC-V investment will speed up RISC-V's development. Calista Redmond, RISC-V International's CEO, sees these moves as recognizing that "massive investment in open source has the power to change the course of history." Redmond went on to say that open collaboration with RISC-V has "ignited a profound shift in the semiconductor industry, and this partnership will accelerate innovation in open computing. RISC-V welcomes Intel and looks forward to our collective expansion and the commercial adoption of RISC-V across compute workloads and industries, growing RISC-V everywhere."
Why? Because they already failed at ARM (Score:4, Informative)
No shit, sherlock. We still have MIPS around too, and hell the goddamned Z80 still crops up now and again. And PPC, and probably a whole bunch of architectures I can't think of right now. More importantly Intel already tried and failed with ARM and they don't have to spend any money to license RISC-V for their next try at low power.
Comment removed (Score:5, Funny)
Re: (Score:2)
Nah, they probably only want to claim to have the fastest* RISC-V CPU on the planet.
* at more than double the power requirements of the nearest competitor, but they're not going to show that part on the marketing material.
Re: (Score:3)
Nah, they probably only want to claim to have the fastest* RISC-V CPU on the planet.
* at more than double the power requirements of the nearest competitor, but they're not going to show that part on the marketing material.
I find it very odd that we put power restrictions on pretty much anything going for "fastest ever".
To put that in the typical car analogy, how many Lamborghinis only have 100HP, and come with 1-liter engines that get 50MPG due to peer pressure?
Exactly.
Re:Why? Because they already failed at ARM (Score:4, Insightful)
I find it very odd that we put power restrictions on pretty much anything going for "fastest ever".
To put that in the typical car analogy, how many Lamborghinis only have 100HP, and come with 1-liter engines that get 50MPG due to peer pressure?
Exactly.
It's not odd at all. To continue this car analogy (this being Slashdot, after all), we have multiple categories of vehicle speed records. The "fastest ever" street legal vehicle is a separate category from the "fastest ever" top fuel dragster, and the top fuel dragster is a separate category from the "rockets with wheels" category that sets land speed records.
Nvidia holds the record in the unlimited power category for vector processing and AMD holds the record in the same unlimited power category for scalar processing. Apple holds one of the power efficiency records for scalar processing, leaving Intel out in the cold. It would be a very Intel thing to do to blow out the power budget in order to produce a "fastest ever" RISC-V processor. Intel is good at missing the point like that.
Personally I wouldn't even mind. RISC-V has been languishing for years with its promise largely undeveloped. If the RISC-V consortium could claw its way up to at least parity with ARM and then give us a platform spec, which ARM doesn't do, it would be a big step up for low power computing everywhere.
Embrace and profit (Score:5, Interesting)
"Embrace and extinguish" refers, of course, to late 1990s Microsoft. Today, Microsoft makes more money from Linux machines than they do selling Windows.
Microsoft divides their businesses into three categories when they do their financials:
Azure
O365
Gaming and other
Windows is part of "other" now. Not because Microsoft isn't doing well. Microsoft is making Anton of money. Much of it by selling Linux instances on Azure. Microsoft is doing well because Nadella fully understood that the way MS made money in 1998 isn't the way they'll make money in 2020. About a year ago, they did a PR for a kernel patch to let them run their money maker, Azure, fully atop Linux, so they could get rid of Windows on the hypervisor layer.
It may be Intel leaders are realizing the way Intel made money in 1998 might not be the way they make money in 2028.
Re: (Score:3)
"Embrace and extinguish" refers, of course, to late 1990s Microsoft. Today, Microsoft makes more money from Linux machines than they do selling Windows.
This is the dumbest shit I've ever heard, microsoft and intel have been slowly working together to turn the PC into a locked down mobile device to finally end piracy. You should really check out trustedcomputinggroup. The last 23 years of game theft on the PC and hacking game software from ubisoft, and the big rebrand of PC games as mmo's to steal their networking code to deny game ownership and dispossess the masses. The idea that microsoft has given up on embrace, extend, extinguish, is NAIVE at best a
Re: (Score:2, Insightful)
Today, Microsoft makes more money from Linux machines than they do selling Windows.
This is the dumbest shit I've ever heard, microsoft and intel have been slowly working together to turn the PC into a locked down mobile device to finally end piracy
Who cares about the PC? Microsoft doesn't.
Windows 10 only just recently surpassed 1 billion installations.
Most of those were free upgrade licenses. OEM licensing boils down to a few dollars per license.
Azure has surpassed 200 billion active instances monthly.
Each and every one has a running cost, a sum of CPU usage, RAM usage, and data storage.
Azure instances are reoccurring costs on a monthly basis.
Windows is a one time payment at best.
It is their Azure and Xbox divisions that bring in most of their inco
Re: (Score:2)
Who cares about the PC? Microsoft doesn't.
https://tifca.com/wp-content/u... [tifca.com]
Re: Embrace and profit (Score:2)
"This is the dumbest shit I've ever heard, microsoft and intel have been slowly working together to turn the PC into a locked down mobile device to finally end piracy."
HAHAHA etc
First, end piracy? GLWT. Second, no, the purpose is to control "your" PC, and make it theirs.
Re: (Score:2)
Most likely for internal use in their own products.
Remember the Intel Management Engine? It's got its own CPU that is a proprietary Intel design. It's basically a complete separate computer inside your computer, that allows the system to be managed even when turned off. It supports remote desktop so you can access the BIOS screens over the network.
Intel has to develop that CPU, and develop tools for it like compilers. Since RISC-V is getting support and active development they might as well switch to that n
Re: (Score:2)
The ME is based on a cut down x86 core, which already has tools like compilers available for it...
Re: (Score:2)
True. It used to be ARC, then they moved to Intel Quark which is x86 based. It's totally separate from their high performance x86 stuff though. Limited instruction set, designed for power efficiency. It's a lot of effort to keep it current.
Re: (Score:2)
Keep it current with what? It is not a performance part.
Now, if it keeping it currently with the needs of new processors, then how does switching to RISC-V help with that at all?
Again, you make no sense.
Re: (Score:1)
Again, you make no sense
Be gentle with our resident shill - his is an uphill battle.
Re: (Score:2)
Yes, it's a Quark core - which is also an actively developed and supported product in its own right.
The instruction set is that of the original pentium, it's not current and not intended to be. It's basically an early 90s pentium fabbed on a more modern process with the resulting reduction in power usage and higher clock rates.
The ongoing effort is going to be pretty minimal, and compiler/os support for this architecture is very mature already. I'm sure sales of quark cores on their own more than pay for wh
Re: (Score:2)
"Intel has to develop that CPU, and develop tools for it like compilers. Since RISC-V is getting support and active development they might as well switch to that now, reducing their costs significantly."
But by your own admission, Intel has already developed a "proprietary Intel design", so how can it be that "switching to that now" will reduce costs "significantly"?
Sounds good, but isn't.
Re: (Score:3)
A lot of the low-end (microcontroller world) seems to have consolidated around variants of ARM, though there are outliers. (8-bit MCUs aren't ARM, but 32-bit ones are now nearly as cheap. And ESP32 was Xtensa, at least originally.) ARM has also been an IP block for use as a processor to embed in other peoples' ASICs and FPGAs for a long time.
I can see RISC-V being well positioned to go after ARM in these spaces. No one wants to have to pay licensing fees for someone else's IP when they need something to emb
Re: (Score:3)
I can see RISC-V being well positioned to go after ARM in these spaces. No one wants to have to pay licensing fees for someone else's IP when they need something to embed in their products
RISC-V needs a lot of work and development before it can really complete with ARM on anything other than price. The RISC-V implementation I have experience with, SiFive, is a real mess compared to comparable ARM offerings. It's got an interrupt mechanism only a mother can love.
Re: (Score:2)
32 bits are far from "nearly as cheap" as 8-bit.
A cheap 32-bit SoC is now less than a dollar. But a cheap 8-bit SoC is now less than a cent. Look e.g. at Padauk devices.
Re: (Score:2)
"I like the ESP32 - good cost, performance, capabilities, and newer ones have RISC-V cores. It's a nice bit of kit - very versatile. I don't like that's Chinese origin."
They have one RISC-V variant. It is not their newest one, it is their cheapest and lowest performance one. Also, while it may be a "nice bit of kit", in no way is it as nice as their other "bits of kit". Not that you are likely to know the difference.
I cannot stand misleading, ignorant bullshit like this. Nothing but a disingenuous effor
Re: Why? Because they already failed at ARM (Score:2)
Windows has demonstrated the same flexibility repeatedly, it has supported several non-PC architectures... PPC, ARM, Alpha. Far as I know only parts of the system ever ran on ARM but all of it at the time ran on the others.
Re: (Score:3)
I wouldn't say Intel "tried and failed" with ARM, but if that were true then they will fail if they try with RISC-V as well. After all, it wasn't Intel's inability to make an ARM processor. If they can fix that with RISC-V then they could fix it with ARM.
Also, there is no reason to accept at face value the biased claim you cited. Intel may invest in RISC-V but there is no reason to believe it's because Intel "sees a future" where RISC-V "plays a major role". It may be hedging and nothing more.
Furthermor
Re: (Score:2)
Interestingly, AMD is doing well with ARM. Seems Intel is lacking engineering capabilities...
Wow, can't wait (Score:5, Interesting)
So, will Intel start porting Minux to RISC-V ?
https://www.zdnet.com/article/... [zdnet.com]
Re: (Score:1)
I understand they want to port TempleOS to RISC-V
Could be Gelsinger influence (Score:2)
And yet (Score:2)
They can't get their current chips out the door so manufacturers can produce PCs. But sure, why not, add more to your plate when you haven't finished what you already have.
Keep the failures rolling.
Intel became overspecialized (Score:3)
Foundries have become market drivers and their new strategy is not just about Risc-V, it's about the shift to modular IP in the CPU marketplace. They realized they had to change their basic game plan and they are putting their money behind their game plan.
Re: (Score:1)
...Intel never got on the GPU train.
That's not fair; they were going for efficiency. [youtube.com]
Re: (Score:2)
Modular IP sounds like a recipe for fragmentation.
This is good (Score:2)
If we want RISC-V to compete with the big dog chips something like this was inevitable. [youtube.com] Intel was also the likely big player to jump in considering their lack of ARM development. There is no "good" company of any large size that could jump in, it's a pick your poison type of thing.
Intellectual Property (Score:3)
Is this early-stage foundry enough for ya Intel? (Score:1)
Sounds RISKy (Score:2)
Staying in the foundry biz requires more than x86 (Score:2)
Embrace, Extend, Extinguish (Score:1)