Slashdot Mirror
'Is It Time For Open Processors?' (lwn.net)
Linux kernel developer (and LWN.net co-founder) Jonathan Corbet recently posted an essay with a tantalizing title: "Is it time for open processors?" He cited several "serious initiatives", including the OpenPOWER effort, OpenSPARC, and OpenRISC, adding that "much of the momentum" appears to be with the RISC-V architecture. An anonymous reader quotes LWN.net:
The [RISC-V] project is primarily focused on the instruction-set architecture, rather than on specific implementations, but free hardware designs do exist. Western Digital recently announced that it will be using RISC-V processors in its storage products, a decision that could lead to the shipment of RISC-V by the billion. There is a development kit available for those who would like to play with this processor and a number of designs for cores are available... RISC-V seems to have quite a bit of commercial support behind it -- the RISC-V Foundation has a long list of members. It seems likely that this architecture will continue to progress for some time.
Here's some of the reasons that Corbet argues open souce hardware "would certainly offer some benefits, but it would be no panacea."
Here's some of the reasons that Corbet argues open souce hardware "would certainly offer some benefits, but it would be no panacea."
- "While compilers can be had for free, the same is not true of chip fabrication facilities, especially the expensive fabs needed to create high-end processors... It will never be as easy or as cheap as typing 'make'..."
- "Without some way of verifying underlying design of an actual piece of hardware, we'll never really know if a given chip implements the design that we're told it does..."
- "Even if RISC-V becomes successful in the marketplace, chances are good that the processors we can actually buy will not come with freely licensed designs..."
- "Finally, even if we end up with entirely open processors, that will not bring an end to vulnerabilities at that level. We have a free kernel, but the kernel vulnerabilities come just the same. Open hardware may give us more confidence in the long term that we can retain control of our systems, but it is certainly not a magic wand that will wave our problems away."
"None of this should prevent us from trying to bring more openness and freedom to the design of our hardware, though. Once upon a time, creating a free operating system seemed like an insurmountably difficult task, but we have done it, multiple times over. Moving away from proprietary hardware designs may be one of our best chances for keeping our freedom; it would be foolish not to try."
There are several dozen teams designing RISC-V implementations. And many ASICs have RISC-V cores buried in them today. With a handful of designs being open.
The main barrier for ordinary people and software developers to have a proper R5 workstation is for there to be a market for such a chip. Right now the market is driven by the needs of ASICs, and that's not really what people are asking for when they say an "Open" processor.
“Common sense is not so common.” — Voltaire
That is definitely a way of looking at it. The other way to look at it is that somewhat-wealthy organisations already do invest significantly to other open projects [ not just/only open-source projects ], because it benefits them to do so.
One online article notes 16nm Finfet fab entry cost at $80M, 66 mask steps. You would need a very wealthy patron.
I believe you need to go back and re-read the REASON for the law.
The idea for it started from "lazy journalism" - which this is none of. This is a vetted technical person actually asking a technical question to the community at large. The technical question is followed up with detailed analysis on why such a question is being asked, and the ramifications of the decision if it were to be made.
Not even that. There is actual expense involved. You will not be downloading a Risc-V or any other processor core and then going to a 3d printer to print it. That will never happen. The technology for 3D-printing right now couldn't even 3d-print a tube for an analog computer.
What people are using right now are FPGA's which cost 100x more than the chip core they are capable of emulating. Most of these FPGA's that are in affordable range can barely emulate an 8-bit computer. So unless you want to sacrifice the performance of an Intel i7 in favor of an Z80, no I think this is barking entirely up the wrong tree.
We have to be resigned to the fact that there will never be "free cpu"'s at all. A chip fab will not spend a billion dollars on a facility, and then produce everything for free. That's insane.
Where we can potentially break the ceiling on proprietary designs lies more in breaking the CPU up into pieces that can first be designed using multiple FPGA units, and once those designs are proven to work, license those out pieces out to whoever is willing to finance the production of the ASIC's that must not be modified from the open design. If you allow third parties to modify it, then we're back to proprietary designs again. This is one of the quite literal best use cases for GPL-like provisions. Any ASIC released, must be released with everything required to reproduce the ASIC.
A monolithic design similar to the i7 in scale, will simple not be possible without someone like Apple adopting it, and nobody else out there is willing to fight the inertia to do such a thing. Even Nintendo has moved away from RISC to ARM.
You can buy RISC-V asic's currently https://www.sifive.com/posts/2017/10/04/sifive-launches-first-risc-v-based-cpu-core-with-linux-support/
But you're looking at a chip that is so slow, that you would not be sticking this in servers or desktops, it would be relegated to hobby devices.
Let's assume for a moment we had a rousing speech from the ghost of John F. Kennedy saying that this community should commit itself, to achieving the goal, before this decade is out, of creating an open processor, and installing it safely in a computer. And Jeff Bezos thought it was a good idea and committed to writing a blank check to make it happen.
And enough of the the few thousand people in the world who can ground-up design a processor have willingly donated their time to the effort, and have made a perfect, error-free processor with very little physical testing, and one or two of the few-dozen-at-best CPU fab plants in the world have committed their time to retool their assembly lines to decrease the output of Intel and AMD and ARM and Qualcomm chips to make a few hundred thousand of this OpenProc. Also, we're assuming that all of this is done such that there are zero patent infringements from the existing guys, and thus at no point are there any lawsuits from Intel or AMD.
We're already comfortably in 'not happening' territory, but let's keep going.
These CPUs need to fit into motherboards somewhere, right? I mean, the implication here is that we're looking for desktop and server chips. They're not going to work in standard Intel or AMD sockets, I'm assuming...so on the heels of designing an open processor, we need an open motherboard to fit it (which again, avoids any and all litigation as it's being designed). Somewhere in that process, we also end up with an OpenNIC and an OpenSoundBlaster and OpenSATA and FreeUSB and FreePCIe et al. Also, someone codes a ground-up open UEFI or BIOS or something that interacts with all of this hardware properly and without issue or conflict, because any issue faced in this scenario becomes the biggest possible nightmare to test. Also, Foxconn agrees to produce this MagicMobo alongside standard, more profitable units.
Now, we've got all that hardware and can get to a boot device. What are we booting? Linux successfully compiled for this barely tested hardware using a compiler that assumes all the specs are, in fact, working as intended? Okay, great! now let's get some more software on it, because a full Linux distro, even something as relatively-simple as DSL or Puppy is going to require all of its software to be recompiled, so it's yet another race to start porting over applications, with some applications never leaving x86 due to a lack of developer interest.
Everyone, everywhere, ever, has willingly done their part to support this new architecture. Now, to convince people to use it. Who, exactly, would that be? Some software developers and hobbyists, I mean sure, but then who? End users, even tech savvy ones, are going to be wary of an architecture where the best case scenario is a subset of standard Linux software, to say nothing about the countless Windows and OSX titles, niche hardware, and lack of laptop iterations.
Maybe if it were heavily optimized for database loads it might have a bit of a niche there, but now you have to have someone's name on it. Who is going to be the OEM to sell these machines? Companies aren't buying motherboards and rackmount cases to start using these as servers; Dell or HP or Lenovo will have to get on board, which is rough when Intel has been their poison of choice for so long.
So, in summary, even if everyone volunteers everything they need at every step of the way, what is the expectation? A niche market at best, which will always be treated as a second class citizen, and whose selling point is the great sacrifice made to bring it into existence.
Designing the architecture and logic is fraction of the engineering effort necessary to design and build a modern high end microprocessor.
In addition, a high end processor needs a complicated motherboard to run it, with high speed memory, and various peripheral I/O systems, driven by separate ASICs, or integrated in the CPU. A desktop PC motherboard is a very complex design, which is only made affordable by huge volumes.