The OpenPOWER ISA EULA Draft Published - Generous For Libre Hardware

[Luke_Wolf]

where's the source code? where is the HDL? where are the Board Support Packages allowing people to program these milions of devices?

sorry to have to point this out. proprietary, secret anr closed systems, no matter how many hundreds of millions of units are sold, do not make a thriving open user and developer community.

Sorry to have to point out but the kind of "open user and developer community" you want for hardware is a pipe dream until such time as chip fabrication technology becomes accessible to consumers in their own homes, and thus you can actually have a developer community as opposed to just large corporations who can pay fabrication plants millions in order to produce chips.

[lkcl]

https://forums.raptorcs.com

allo mr mad scientist, nice to see you around. will make sure to crossref that forum on our wiki.

[lkcl]

hardware is a pipe dream until such time as chip fabrication technology becomes accessible to consumers in their own homes,

DARPA has relatively recently dropped something like a hundred million dollar budget on the table, to do exactly that. in the meantime the average student and home developer can buy a $99 ECP5 FPGA board.

for Libre teams - those willing to do something different, radical and out of the box, it turns out that anonymous companies, constrained by their influence in areas they have a conflict of interest, are finding creative and indirect ways to sponsor them.

all it takes, then, is a willingness to commit and to believe, and, if you don't believe, do it anyway.

[bridgman]

Sorry to have to point out but the kind of "open user and developer community" you want for hardware is a pipe dream until such time as chip fabrication technology becomes accessible to consumers in their own homes, and thus you can actually have a developer community as opposed to just large corporations who can pay fabrication plants millions in order to produce chips.

Not sure I agree here... there are a lot of options available today that don't require chip fab factories at home. Multi-project wafer programs like MOSIS have been around for 30+ years, although their focus is more on university class projects than on individual hobbyists. Minimum orders are in the thousands of dollars rather than millions even today though.

https://www.mosis.com/

FPGA boards also represent an immediately accessable option, as lkcl pointed out.

[Luke_Wolf]

Not sure I agree here... there are a lot of options available today that don't require chip fab factories at home. Multi-project wafer programs like MOSIS have been around for 30+ years, although their focus is more on university class projects than on individual hobbyists. Minimum orders are in the thousands of dollars rather than millions even today though.

https://www.mosis.com/

Doesn't change the fact that there's an extreme barrier to entry to fabbing your own hardware, a cost that only companies can really stand to bear. There's a reason there's been no serious efforts to do this.

FPGA boards also represent an immediately accessable option, as lkcl pointed out.

Don't be absurd. FPGAs and virtual processors are cute, but let's not play pretend that they're actually useful in these kinds of endeavours because their performance is absolute garbage. Nobody wants a GPU that can only manage 25FPS @ 720p, which are their own numbers at how terrible this thing really is. If there is no user interest because it is so garbage then consequently there will be no independent developer interest in any way that can be considered meaningful. Which means we come back to needing real ASICs, which are prohibitively expensive for even software developers to afford getting produced... which means we get back to this underlying problem that user driven open hardware rather than corporate driven open hardware (POWER, SPARC, RISC-V) is nothing more, and nothing less than a pipe dream that is unlikely to catch on in any meaningful sense. Though to be clear I don't think this is a bad thing any more than the fact that the Linux Kernel is really driven by corporations rather than individuals, just to a far less extreme.

[lkcl]

focus is more on university class projects than on individual hobbyists. Minimum orders are in the thousands of dollars rather than millions even today though.

https://www.mosis.com/

FPGA boards also represent an immediately accessable option, as lkcl pointed out.

DARPA was looking to bring the cost down of custom ASIC development to around i think it was USD 600 per million transistors.

no mask charges (or, they are amortised, like in shuttle services) and that's nanometre geometries.

the project is going extremely well. there was a talk about it by someone who used to work for Esperanto, i apologise i forget his name.

their idea is to FULLY automate the entirety of chip development, entirely as libre licensed tools.

with a USD 150 million budget they are getting an awful lot of traction.

in the meantime thanks to that anonymous sponsor we have access to two 180nm tapeouts at a cost of only USD 600 per sq mm, one will go on a test Cell Library ASIC in March, the other in October we are scrambling to get a single core basic chip done *and* P&Routed by then.

[SystemCrasher]

I'd say moaning anything about "EULA" is very bad start if someone is anyhow serous about being open and libre. "EULA" doesn't sounds "opensource friendly" at all and usually implies quite unpleasant treatment. As for RISC-V vs microcontrollers... there're already some! Chinese GigaDevices company did a rather funny trick, they basically replaced ARM cortex M core with RISC-V core, while more or less retaining STM32-like peripherals. A very funny combo any day - and whatever but it makes reasonable MCU, capable of nearly anything ARM version has been capable of. Of course it boasts RISC-V "compressed" instruction set similar to thumb2, etc. So it seems ARM soon would have far harder times. And all these "us vs china" trade wars would definitely speed this up.

[SavageX]

RISCV is very popular in proprietary systems where the augmentations, firmware, modified toolchains etc never see the light of day. Western Digital, Trinamic, NVIDIA, they are all using RISCV... *internally*...

RISC-V indeed seems to be a popular choice right now for deeply embedded systems, which means those are usually not user-programmable. There are open and closed RISC-V implementations - and I expect the same to be true for OpenPOWER.

https://github.com/antonblanchard/microwatt

anton blanchard's microcontroller would tend to suggest otherwise! anton i hear wanted to learn VHDL. hey, why not?

It's always nice to see open implementations. Sadly, for the project you kindly referred to, I don't see resource stats (e.g., FPGA utilization, which gives a rough hint at implementation complexity). However, whether OpenPOWER is suitable for microcontrollers IMO depends on how the ISA will be split into "must implement" and "optional" parts. Is a FPU mandatory? Is a vector unit mandatory? A MMU? Depending on what is needed to be compliant, microcontrollers might be within reach - or not. POWER is not exactly "slim" if implemented to full spec.

On a *long term* basis, we can see that the RISC-V Foundation is just yet another example of "Fake Oern Source", where by contrast because IBM takes this very seriously, they are willing, with Hugh's help, to take the time to get it right.

While your problems with the RISC-V Foundation are well-documented (it appears to be non-trivial to propose new extensions that weren't already on the Foundation's agenda), in general the RISC-V Foundation seems to develop, ratify and publish specifications that can be implemented without undue restrictions. I'm eagerly awaiting OpenPOWER to first get to the same state of "getting things done" and eventually surpass it. Until then the jury is still out.

[lkcl]

RISC-V indeed seems to be a popular choice right now for deeply embedded systems, which means those are usually not user-programmable. There are open and closed RISC-V implementations - and I expect the same to be true for OpenPOWER.




It's always nice to see open implementations. Sadly, for the project you kindly referred to, I don't see resource stats (e.g., FPGA utilization, which gives a rough hint at implementation complexity).

yes, it's very much an experiment on anton's part.

However, whether OpenPOWER is suitable for microcontrollers IMO depends on how the ISA will be split into "must implement" and "optional" parts. Is a FPU mandatory? Is a vector unit mandatory? A MMU? Depending on what is needed to be compliant, microcontrollers might be within reach - or not. POWER is not exactly "slim" if implemented to full spec.

these are really good questions, which genuinely need answers. RISC-V went to the trouble of creating "embedded" platform and "unix" platform, to make it easy and clear.

While your problems with the RISC-V Foundation are well-documented (it appears to be non-trivial to propose new extensions that weren't already on the Foundation's agenda),

i'm really pleased to hear that people are "getting it". that it took my own personal money and that of NLNet (which is charitable donations) - really pisses me off. the ******s who include patterson telling me "i should quit For The Benefit Of The RISC-V Community" which was deeply shocking and absolutely inappropriate - have a hell of a lot to answer for.

some day i will publish the full message patterson sent me. or perhaps i will be required to do so as part of an EU anti-trust investigation.

yes, you're right: getting extensions into RISC-V is not just non-trivial, it's absolute hell. MIT had to fight tooth and nail for six months just to get a three-paragraph chapter on "TSO" into the spec. the person i spoke to gave me the impression that it was deeply and disturbingly unpleasant.

and that's a "respected university".

in general the RISC-V Foundation seems to develop, ratify and publish specifications that can be implemented without undue restrictions.

yes. as long as you "do what the arrogant fascists dictate" (EDIT: NOTE TO READERS: DEFINITION 2 ON URBANDICTIONARY. not the proper noun which starts capital F), you're fine. if you question their "right to tell you what to do, in the way that they tell you it must be done", that's where the problems start.

Trademark Law is *specifically* designed - very very clearly - to make the penalties for this kind of unacceptable behaviour very harsh. the problem comes in that even *beginning* to pursue a Trademark Invalidation lawsuit is a waste of funds, and becomes a "Pyrrhic victory" regardless.

i.e. if we quotes win quotes a Trademark invalidation lawsuit against the RISC-V Foundation, it destroys the entire *purpose* of the Trademark.

this is why i had a quiet word with someone from the EU Commission, at FOSDEM, because they're better equipped to take the RISC-V Foundation - and its members - to task, over the fact that they're running into anti-trust laws.

people don't believe me, they think it's a joke that i call the RISC-V Foundation a cartel: it most definitely is not a joke. the person i spoke to took it very seriously when i pointed out that we'd been excluded from "innovation" due to the unique combination of "Libre", "transparency" and "business objectives".

I'm eagerly awaiting OpenPOWER to first get to the same state of "getting things done" and eventually surpass it. Until then the jury is still out.

me too. well, luckily, the "trail has been blazed". OpenPOWER can pretty much cookie-cut what the RISC-V Foundation has done (Compliance Suites, WGs, Platform Specs), except do it in a properly inclusive and transparent fashion.

[SystemCrasher]

IMO depends on how the ISA will be split into "must implement" and "optional" parts

Most funny part about that vs RISC-V is that GD32V MCUs even managed to implemented "full" 32-bit RISC-V specs rather than limited 16-regs-only crippled "embedded" version. Ofc using "compressed instructions" extension and so on. So it seems RISC-V ISA isnt bad about stuff like that. I wonder if someone can provide any comparison for power in this regard.