OK I create some slides adressing it and cutting it into the video
Cheers -lev
On Tuesday, January 18, 2022 6:32:42 PM WET Ferenc Éger wrote:
Hello David, Hello List,
Maybe this message is late, but anyway. I had some time to review the slides in-depth. At the general sense, I feel that the technology steps and the machines are a bit over-represented, or more precisely, other aspects of the project are under-represented. The other aspects of the project (roadmaps, LS standard cell library and compiler, SOC and MCU plans, 555, that is, what we did in the previous ~~3-4 years) should also be shown. Without this, it feels a bit like a Christmas wishlist with a BTC address.
I also have the following specific remarks:
First, let's, in thought, take apart the "Goal" section: we assume that there are two kinds of audience there: big players and small players, with the delimiting factor is whether one can/'t afford to set up one's own production (CMP: 6kUSD, diffusion furnace 5kUSD, CVD, sputter, RIE, litho, cleanbox/cleanroom, etc. Altogether let's say 50kUSD. Plus the means of getting HF, TMAH, H2O2, H2SO4 and their friends. I mean, legally.). For the big players, the question is this:
"I am a fab owner or a fabless designer with existing contracts to fabs. Why should I spend 50kUSD on developing a common process? What is, among the listed problems, that I cannot solve for myself otherwise without it? Supply shortage? The EU right now bombards established european fabs with EUR 145 bn for process upgrades and capacity expansion. Transparency, hidden backdoors, vendors spying on consumers? No problem, we big-tech and governments are the beneficiaries of this. Just let's have the key. Vendor lock-in? Monopolistic playoff of overpricing? As a fab, why shall I be sad about that? As a customer, I don't care: as long as it affects me and my competitors equally, I can charge the overprice to my end-users without risk. Are IP laws, export control, etc. weaponized? That's we meant them for... " As for NDAs hindering cooperation, that's not really true, at least not on this scale. It is more-or-less usual that European companies, behind the NDA wall, co-developing processes, or cross-licensing them (e.g. I know an instance, where an European automotive company I cannot name here cooperated with an european semiconductor manufacturer to develop a process for power transistors, on the condition that then both will cross-license it and devices designed on it and manufacture them in their own fabs). Not to mention the approach of big players to open-source: "We outsource the sub-problems to the open-source community while holding key elements back, let them solve it for gratis, then we siphon back the results and recombine it with the held-back part into a commercial product for our profit."
The question of the small player is this: "I am a small startup or a hobbyist or a researcher with inadequate funding or institutional background or just a casual user who want to invoke freedom #1 (the right to modify) on an open-source CPU in my laptop, or etc. I have, and presumably will have, no provision for getting the needed devices or the authorization to procure and use the needed materials, or I am a HDL simulation guru with no experience/expertise in physical manufacturing. Setting up manufacturing for myself is simply infeasible and/or overkill. My problem is not vendor lock-in, but vendor lock-out: I don't want to order one million pieces plus a maskset for 500k USD, so everyone slams the door at me. So what is my benefit from a common technology? Why shall I invest financially or otherwise? Semiconductor manufacturers won't implement it as eliminating vendor lock-in is against their business model, and even if vertically integrated companies implement it in their own fabs, they will reserve it for in-house production of their products, as it is now. The key problem is that 1) manufacturing is not flexible enough, 2) even if a more flexible AND cost-effective means would exist, no-one would offer it to the general public."
In my opinion, both of these, but at least one of them, must be addressed somehow in the slides and the talk, and not waiting for the audience to ask these questions.
Second, I would remove the wiki's link, at least for now and emphasize the Mumble and the mailinglist. Reason: the wiki wasn't touched since it was set up, the last commit in the git repo is 9 months old, and the website is still stalled with 2018 news. If I would ask for donations, I would call attention to what we do and not what we don't.
Finally, a remark that reaches beyond the scope of the talk topic, about the entire BTC donation thing. Rectifying it is not possible nor expected in the remaining time, but I consider it unavoidable on the long term. Also this is a rather sensitive issue, so no insult to anyone is intended.
Back in 2020, almost 2 years ago, it was planned that a foundation is to be established in the EU. The project applied for an NLNet fund, which seemed to be granted at first, but it did not materialize. The apparent problem is that they didn't want to transfer to a private bank account. It has to be understood that the way western economy works is that there is a swarm of bigger and smaller economic entities, and everyone wants to get money out of everyone else in all conceivable ways (and expects everyone else to do so). The dynamics of this swarm is governed by bilateral legally binding and enforceable contracts, either explicit (written) or implicit (buying a loaf of bread in the corner shop) that guarantees safety of investment for both parties. Someone will pay money to someone else only if there are concrete terms the payer expects in return, and the payer is assured by law or contract that the other party will deliver or the money can be sued back. Similarly, the vendor will give a product away only if he is assured that the buyer will pay for it (i.e. the buyer has adequate funds to buy the item and the payment can be enforced). The way that John Privateperson Doe meets the CEO, CTO or C-whatever-O of a company in private in a coffee shop because He Has A Lucrative Business Idea (R) (C) (TM) simply does not work. Of course, this does not entirely rule out donations. But when someone donates, one expects that the donation goes to a place that is accountable. If you get a brochure of an animal welfare or whatever group with a registered foundation's bank account, you may consider a donation. If a never-seen-before guy on the street unsolicitedly goes to you and says "I am raising funds for the local animal shelter, I accept cash (or BTC) only, I give no receipt", you just go to the other side of the street. Indeed, all significant open-source projects (Mozilla, the Linux kernel, Debian, VLC) have some legally tangible entity behind themselves, and not just by coincidence. Would-be sponsors and suppliers are willing to either invest, donate or deal with an entity that they can contract with, that gives them assurance of how the money will be used (via a contract or the organization's statute), and that in case of trouble, liability can be enforced (e.g. LLCs in Hungary have a publicly defined sum to which their liability is limited that can be checked by anyone, and that sum is required to be always kept deposited in a bank by law). I, however, have the observation that we have an aversion to any sort of legally binding structure (registering a foundation, approaching others "via the front door" and not via personal contacts, accepting government or NGO funds, ...). On the long term, in my opinion, it will be unavoidable to set up such an arrangement, if we want anyone we expect to do business with us to take us serious.
Regards,
Ferenc
On 1/15/22 08:25, Hagen SANKOWSKI wrote:
Hello List!
This is our weekly announcement for our next Mumble Sessions on Sunday
2022-01-16 @ 18:00 UTC.
Please join us as usual at our Mumble Server murmur.libresilicon.com at Port 64738, the Channel is IC.
We like to follow-up our topics from mumble sessions before.
Regards, Hagen. _______________________________________________ Libresilicon-developers mailing list Libresilicon-developers@list.libresilicon.com https://list.libresilicon.com/mailman/listinfo/libresilicon-developers
Libresilicon-developers mailing list Libresilicon-developers@list.libresilicon.com https://list.libresilicon.com/mailman/listinfo/libresilicon-developers