Open Hardware - Reaching Out

[dan]

Having discussed this a bit more internally, while we agree with the sentiment outlined by Jonathan, we're inclined to keep our focus on the current projects here on xcore.com for the immediate future. We are as always, resource constrained.

I think we will pursue this avenue at some point in the not to distant future, but not right now. Personally, I also want to make more progress with the modular hardware system first in any case.

I suggest we try and close on the definition of the slicekit as our immediate priority.

[jonathan]

OK.

What is this Slicekit thing? Is XMOS going to actually make it? There seems to be more demand for the little L1-48 DIP module than interest in the Slicekit - I'd make that first if I were you. :-)

[dan]

Jonathan, see the hw_slicekit repo on github for details.

We have plans for a super low cost board using the 48 pin as well. If you have some data about relative demand please do post it, that would be useful.

Part of the motivation for the slicekit, from our side at least, is to create an XS1-L based development system to allow us to explore new markets and vertical application segments. We really ought to have a generic XS1-L based product with basic ethernet and USB connectivity, which we currently lack.

[TonyD]

We have plans for a super low cost board using the 48 pin as well. If you have some data about relative demand please do post it, that would be useful.

Part of the motivation for the slicekit, from our side at least, is to create an XS1-L based development system to allow us to explore new markets and vertical application segments. We really ought to have a generic XS1-L based product with basic ethernet and USB connectivity, which we currently lack.

I agree with Jonathan about the importance of having a small DIP module. You only have to look at the Basic Stamp DIP modules to see how simple low cost modules can be used for quick prototyping and one off builds. We would love to hear more about Xmos plans for a super low cost board :)

[dan]

Tony, can you use your powers to move this post of yours to the slicekit thread so we can continue discussion there please, it would be good to try and keep it all in one place.

[TonyD]

will do. Moved text wrt slice to https://www.xcore.com/forum/viewtopic.p ... 6&start=40

[jonathan]

What exactly is XMOS going to physically build and sell? And more importantly, what exactly is XMOS not going to build and sell?

By announcing you have "plans to build" X, Y and Z (but not actually building them) you are essentially making it very difficult for others to make a serious commitment to start building anything commercially - as XMOS could just do it themselves and swallow any market gap that exists.

There's nothing wrong with a collaborative design process to create some new boards (whether slices idea is a sensible target is a different question) but it must go alongside a clear commitment about what XMOS will/will not build. Clearly, XMOS won't make a penny from any of its devboards (or at least, it shouldn't, that's not the point of them) so it will always be in a position to create boards more cheaply than anyone in the market. That's not conducive to a commercially strong open hardware community.

Hmm I have a feeling I'm not being clear. But basically:

1) Is XMOS going to release and sell/distribute an L1-48 minimal devkit?
2) Is XMOS going to release and sell/distribute an L2 devkit?
3) Is XMOS going to release and sell/distribute the slices idea?

It doesn't matter so much whether XMOS does or does not, merely that it's clear about which decision it takes in each case. At least one of the community has put through a batch of the 48-pin DIP board - and I expect many more to if XMOS formally declare it's not making them itself.

Cheers,

Jonathan

[dan]

"By announcing you have "plans to build" X, Y and Z (but not actually building them) you are essentially making it very difficult for others to make a serious commitment to start building anything commercially - as XMOS could just do it themselves and swallow any market gap that exists."

If I gave the impression that I was announcing anything then that wasn't my intention. When I said 'build' I should have said 'develop', so my bad.

What I can say that is for all these boards we have been discussing, in general our strong preference is for them to be developed by members of the community or by established suppliers to the hobby and low volume markets.

Your point about clarity of non-competition is a good one. From our perspective what matters to us is that there is a credible supplier of these products to the community (and to us, since we would order what we need for our internal development from them). The only reason why I can see that XMOS would want to sell these boards ourselves is if there is no credible alternative supplier that comes forwards, or that no consensus on what is required emerges.

For example, we need a modular development system for our own purposes, and if no-one else besides XMOS is interested in making such a system that fits our needs obviously we'll make it ourselves and may or may not sell it as well. As far as the DIP modules are concerned, I think we can safely say we'd leave that up to the community.

Obviously we cannot make any sweeping and definitive statements right now about what we will and won't do about future boards (e.g. an L2 dev kit) but hopefully that is less of an issue now I have made our intent and preferences clear.

So the way to gain some clarity on who does what is for people and outfits who are interested in developing/building/distributing some or all of the kinds of boards we have been talking about to come forward with their proposals, which is a development we at XMOS are very much looking forwards to.

[jonathan]

Thanks for the reply, it helps.

Would it be fair to summarize that as "if you build it, then we won't"?

With respect in particular to e.g. slices - were someone else to come up with, build, sell and support an open-source modular series of development kits, presumably then XMOS wouldn't feel the need to compete with them by producing slices? Is that a fair analysis?

The same is true for the L2. If someone comes up with, builds, sells, and supports an open-source L2 devkit, presumably XMOS wouldn't feel the need to compete by producing an L2 devkit? Again, is that fair?

I ask mainly because there are specific projects and people who will be affected by this.

[dan]

Would it be fair to summarize that as "if you build it, then we won't"?

Yes I can confirm that XMOS has no intention of 'competing' with the community in the dev boards space, and we would have every intention and motivation to ensure any community hardware suppliers can have a healthy and sustainable business. XMOS is a chip company.

I think we're in alignment here, the spirit of what we want to see and what you are asking for seem to be well aligned.

With respect in particular to e.g. slices - were someone else to come up with, build, sell and support an open-source modular series of development kits, presumably then XMOS wouldn't feel the need to compete with them by producing slices? Is that a fair analysis?

My concern here is simply that there are modular systems and then there are modular systems. Low IO density vs High IO density. Power supply capabilities. Daisy chaining capabilities and so on. There may exist gaps which need to be filled, for example in the 'modular hardware' space which are not served by existing non-xmos suppliers. Obviously we'd not commit to leaving these needs unfulfilled just because an existing dev system having the word modular in its description exists.

In fact if we moved to fill such a gap, then the obvious thing to do is to arrange to fill the gap in conjunction with an existing community supplier, by agreeing that they would extend their range, perhaps with some development assistance from XMOS.

Hopefully this addresses your concerns?