Open Hardware - Reaching Out

[jonathan]

The OSHW community is currently bigger than XMOS. Much bigger. I wouldn't worry about you awarding yourself your own logo or taking over in any way...

And I agree that it would be great to get others to participate, but:

1) No-one has the same kind of technology as XMOS and therefore can have the same kind of potential relationship with OSHW
2) Someone has to go first. If not XMOS, then XMOS can't be and won't be a leader.

I don't think the XMOS community can take the kind of lead required to get these other "providers" interested, so again, if not XMOS, who else?

What's your ideal situation here Dan? Happy to talk to a few of the guys involved and see whether we can make it happen, but I view this as an area XMOS can be a leader, not just another name on a long list of chip suppliers. Unless XMOS wants that leadership role, it's not worth the effort.

[dan]

The OSHW community is currently bigger than XMOS. Much bigger. I wouldn't worry about you awarding yourself your own logo or taking over in any way...

And I agree that it would be great to get others to participate, but:

1) No-one has the same kind of technology as XMOS and therefore can have the same kind of potential relationship with OSHW

Perhaps we should discuss in more detail what that relationship might be - because possibly we are talking at cross purposes here. Obviously our silicon IP (e.g. the G4, L1 etc) is not open. Our many software components, our 'software defined silicon' is open and becoming more so every day but this is actually software, not hardware.

So for example, we wouldn't use the logo on our XC based stuff, because it is not a tangible item. We would however use it on our modular hardware system, possibly on some or all of our dev kits and other xmos based efforts by the community.

Would you agree?

Of course because the creation of XC based components uses simple free tools, unlike FPGAs, it makes the platform very accessible for open hardware but we can't actually position that in specific terms vis-a-vis the OSHW definition.

2) Someone has to go first. If not XMOS, then XMOS can't be and won't be a leader.

This is where I am confused by what you mean by a leader. I see that the likes of sparkfun and arduino are involved already and from what I suggested above I'm not sure how what we bring is different - we are offering open source hardware designs just like they are. So in what sense would we be a leader?

Would it be in the sense that we would be the 'leading (possibly only) semiconductor vendor member of the OSHW community'? Is that what you mean?

Sorry if this seems a little specific and pedantic but I'm trying to pin down the scope of what you are advocating here. I'm not against it but I do want to understand what you think it is and what you think it is not?

[Heater]

jonathan

The OSHW community is currently bigger than XMOS. Much bigger.

What exactly do you mean by that?

There many facets of OSHW. At least:

1) The thousands or millions of hobbyists and tinkerers building stuff at home. They have always been there, my father was building radios for himself in the early 1930s for example. Only now they have the huge power of the internet binding them together.

2) A growing industry of companies supplying the needs of those tinkerers. The SparkFuns and the GadgetGangsters. Despite open sourcing their designs they are making money supplying kits and boards and components etc.

3) The guys who make a lot of this possible. ATMEL, MicroChip, They make the silicon. They DON'T open source any of their key products. I wonder how much input they have to the open source communities. When you are selling billions of dollars worth of micro-controllers to industry are you really concerned about the OSHW tinkerers?

I still think it's a paradox. XMOS and co. could well benefit from being adopted by the OSHW suppliers and surrounding community of tinkerers. Think ATMEL and Arduino. But it all seems a bit hypocritical to wave the flag for open source hardware whilst keeping your key IP under lock and key.

[Heater]

When I say "tinkerers" I'm not intending to mean that as a put down in anyway. Just by way of contrast to large commercial users.

Now, there is the thing. I believe it is vital for XMOS to provide "software defined hardware" components and other components in an open source manner and with a license the encourages anyone and everyone to use the devices. No matter if the resulting products are commercial or not, or if they are open source or not.

[Folknology]

Well here is something opinion based.

I think right now Xmos offers the closest thing to pure microprocessing open hardware. Obvious;y there is a caveat here in that Xmos isn't (that I am aware of) opening their XS1 design so that is of course closed. but pretty much everything else is open or will be opened according to Xmos intentions.

Contrast this with something popular like ARM MCUs, from any of the major vendors. First off like Xmos the ARM core is closed. but worse still it has terrible licenses with its software, check out the CMSIS terms which are supposedly open. Next add into that the proprietary peripherals added by the major vendors, most of this hardware is closed and proprietary as are the software drivers for them. Worse still many have closed development tools, although many can of course be used with GCC chains. There is also a great deal of specialisation with the peripherals than make compatibility between vendors near impossible from a code base or modularity POV. Thus ARM based MCUs would appear a great deal less open than Xmos XS1 based products and theoretically would be further from the opensource ideals. You can draw similar comparisons from the other major MCU playes like Atmel, Microchip, TI etc..

The next best alternative is soft cores on FPGA. However the industry is rife with proprietary closed IP and very proprietary tools, the score card for the FPGA vendors looks much worse than Xmos. There are of course exceptions such as Lattice with there open source core, but they still have issues around tools. OpenCores community again offer both open source cores and peripherals as well as wishbone for connecting them, but even here the tools have historically been weak. There is some light at the end of the FPGA tunnels as most of the tool issues have now been overcome and as such opensource tooling is beginning to break through outside of the vendors. But here again as far as businesses go Xmos is a stronger and a better aligned candidate or at least could be if they wanted to pursue it.

What would Xmos do?
Well Xmos need to get an opensource story across to OSHW folk, they have already made excellent start with both their tools, technology and soft peripherals with favourable OS licenses on github. Xmos could do more on the tools front and supporting software like the flash and platform stuff, they could also get the XC toolset out there in full. But Xmos need to make more of what they already have in these communities, they could present a stunning case IMHO. I think they could as Jonathan stated 'lead the way' where other vendors just cannot tread.

just my $0.02
regards
Al

[dan]

What I am trying to tease out here is how the wider XMOS open source story relates to the bit of the OSHW definition that states: "tangible items which require commitment of physical resources" (or TIWRCPR for short).

I agree with the thrust of what Folknology and Jonathan are getting at, but most of that is not TIWRCPR.

So if we are to have a story for OSHW, lets talk more about TIWRCPR specifically.

[Interactive_Matter]

Excuse my totally pathetic, wrongly used Hiphopicism: Word Up @folknology

Regarding tangible Items: I think publishing dev kits (if applicable - like XK-1A, XC-1A, XC-2, XC-3) open source is a good idea - and it basically is published open source. Publish the design files and add the right license and you are done. I cannot and do not want to decide if this is applicable to all XMOS baords.

I think spreading the word of mouth and working on success stories (OSH helping to convince real paying customer, customer later contributes to OSHW) would be a great step forward. And yes speaking to the OSHW guys too.

I thin the success story can help to sell more stuff in the end.

[Folknology]

I think the short term tangibles are an important symbol but are really just the tip of the iceberg for the bigger opportunity represented by OSHW. Yes Xmos should make as many kits OSHW license compatible, many are 90% of the way there anyhow, just make some minor tweaks, get approval and add the logo on boards and packaging simple.

But the bigger opportunity here is the long term investment in opensource hardware and software. The more developers that get behind the Xmos opensource story, the greater Xmos's future in OSHW arena will be. In many cases this is more about engaging with the larger OSHW community, businesses and events.

regards
Al

[jonathan]

Hmm... OK, so here's basically the crux, tried to cut it as simple and short as possible.

OSHW guys talk about tangibles. By this they basically (currently) mean boards. At a certain granularity, boards stop becoming open. Typically - as Al said - tools, peripherals, etc.

XMOS changes that. Peripherals and tools are open. XMOS moves the boundary of openness. That is why it can be a leader. No-one else can do this.

With respect to what it should do - in my view:

1) Embrace the current status quo - by openly providing (and branding) devkits - what OSHW currently understands as OSHW - in order to get an engagement. Possibly even sponsor stuff - to validate this engagement.

2) Educate the OSHW peeps to the fact that XMOS actually moves this OSHW boundary inside the chip itself.

I think this is a big story - honestly. I wouldn't be suggesting it otherwise... But in the end, the big picture has to start with a recognition of the small picture and it does have to be XMOS-driven.

And personally, I'm deeply curious about how far openness can go. More than curious in fact.

So, embrace and educate.

[dan]

OK thanks Jonathan, that was useful to clarify exactly what your thinking is. We will discuss this internally to arrive at a decision around point (2). In principle there is no reason no to go ahead with (1) immediately, but I think it's worth waiting a little longer until :

* we have developed a view/policy on the big picture
* we have something a little more fully formed on the modular hardware.

But don't worry, I don't think closing these two items is going to take very long. I'll report back when we have something more to share. Meanwhile lets try and finish of the HW slices definition.

Dan