Imagination GPU cleared for RISC-V CPU compatibility, licensed to chip designers It seems we're a step closer to system-on-chips containing a mix of RISC-V CPU cores and a mainstream GPU powering Linux devices and the like. Imagination Technologies' BXE-2-32 entry-to-mid-level GPU has been tested and validated to work with RISC-V-compatible CPU cores – and licensed to several companies building RISC-V …
Well, there are two costs associated with it; the manufacturing cost and the R&D cost. I don't think it's unreasonable to expect those doing the R&D to be able to recoup those costs, and also to make some money from it, to compensate them for their work. Designers have to eat too. On the flip side, it's also reasonable to suggest that they shouldn't be set for life from the licensing fees from one chip design, either, so a healthy balance has to be somewhere in the middle.
Of course, a person ought to be rewarded for their effort (skill and time and learning, etc..) I'm also pleased that you gave the flip side to the argument.
What's interesting though (at least to me), is that with an open ISA, the R&D talent seems to be hugely broad and deep - far more so than a proprietary practice could ever hope for. And the funding of the R&D seems to come from all sorts of places too.
And because creativity is encouraged, all sorts of interesting and unusual results are being produced.
Those doing the R&D are being paid by the companies they work for.
It's the companies that need to recoup the costs, and in companies, the cost of R&D is part of the product costs.
It used to be that when you bought an electric device, you also got the schematics for it. This is no longer the case, but obviously opening the source to paying customers was not a problem back then either.
Quite a bit of the R in R&D is done at universities, or subsidised by governments in other ways. Arguably that should be available to the public anyway.
A very valuable part of doing R&D is building the expertise associated with the resulting product. If you encounter a problem, or want to build on a product, who better to ask than those who developed it? And that consulting is also time that is paid for by B2B customers. Sharing the schematics only makes that easier.
Part of the success of RISC-V, I think, comes from the fact that it originated as an academic paper, and is used by universities around the world, for research and education, because it is IP-free.
As you suggest, a lot of R&D is done in conjunction with universities these days and, more and more, universities are insisting on using an open source ISA. It's exciting to see the contributions Indian, Chinese, Swiss, Finnish, Norwegian et al. universities are making, both in terms or research (like the Snitch) and with companies (as with the Shakti processor).
We've seen a lot of tech start-ups forming too - with luminaries of the industry leaving lucrative, top chip-design jobs, to pursue funding to design chips for emerging tech, using RISC-V. Next-gen data centres, automotive, ML etc... there's funding available.
And yes, there are companies with the usual business model, doing very well - Andes delivered 3Billion RISC-V chips last year alone, for example.
There's a Gold Rush around RISC-V at the moment. It'll be interesting to see what happens when the chips are mature in all sectors of the market. When there are good enough open source designs to meet people's computing needs.
What I'm suggesting is that there is demand from various quarters for RISC-V GPU, that there will be both open and proprietary offerings from different places (possibly as soon as from the end of this year), and that there are a variety of ways they will be funded.
Of course I could be wrong about any or all of that.
Given Free Software Developers work their arses off to produce drivers for closed source GPUs for no reward from the proprietary producers I think the opportunity to write a driver for something that they dont have to guess how it works would seem like a party.
Having said that I would like to think the people designing open GPUs pretty much have the driver in mind when designing the thing,
In my experience the cost of 'preserving' IP is far more than the cost of developing it. It takes two or three good engineers to design a GPU and 10 far better paid IP lawyers to patent it and then another half dozen or so to look at what everyone elses is doing in case they are infringing on your IP.
Me too!
I've wondered how much of Apple's small R&D budget (vs massive profits and margin) is Engineering and how much is IP staff and costs. IBM spends massively on IP.
Also the UPTO has been broken since the Victorian era and got worse. They make more from accepting claims than rejection and their theory is to leave it to courts regarding if it's not novel, too broad or there is prior art or obvious to anyone versed in the art. All most all patents fail on more than one of those.
This post has been deleted by its author
Interesting company - according to the great t'interweb it's a British limited company (Bris commences drooling) which is Chinese owned (Boris chokes and reaches for napkin) and fundend directly by chinese investors (doesn't rely on UK government funding? Boris' nose is back in the trough).
If this SoC really takes off, I wonder how this will square with the USA limitations on Chinese tech import/export and whether that will actually dampen its adoption in the US-centric world?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
