POWER8 Talos Workstation Drops Price Slightly, Long Way From Being Funded

[phoron]

Honestly speaking I admire the work performed by the company. After providing libre x86 boards they started to lurk into POWER domain, but I'm afraid this is going to fail miserably.

So throw them some money, I guess. I also like the project.

There are few tides running against them: (1) POWER9 notes in news, although misleading since POWER9 may also be provided only to selected customers (google) and general public may not get this for years,

Ok. You answer yourself. Power9 is not yet here, and we don't know when or when or who or how will get it.

(2) AMD Zen is around the corner, (3) they don't like it, but ARMv8 is getting more and more powerful,

ARM is quite against free software. Many prorpietary drivers and often closed bootloaders.
AMD is already designing their CPUs against free software.

(4) generic IBM's driven OpenPOWER movement is not that powerful as they would need.

So we should help them being stronger as long as they allow the CPU owner to control what it does ?

So please Raptor Engineering, provide us with libre version of AMD Zen or Intel Xeon E3 or even Xeon E5 board and I'll be your customer. Do the same for X-Gene future chip or Cavium Thunder and I'll be probably too. POWER8 is unfortunately too little and too late to the party...

They did so while they could. But AMD and Intel decided to actively prevent customers from controlling the hardware they're sold, so they had to leave it.
Intel did it long before AMD, but customers didn't care, coreboot alllowed them, and in the AMD went along and closed their chips. And apparently,
from the Talos crowdfunding, customers still don't care, so they get what they deserve. I think I'm going to build myself an abacus or something...

BTW: I'm curious why it's not possible to do libre board from Intel/AMD offerings when google do that for their Chromebooks? Thanks!

Google doesn't. It may not actively prevent customers from controling their chromebooks, but it keeps all the restrictions the CPU vendors put in.
Many chromebooks have blobs and signed components you can't replace (ME, and so on from Intel). Other chromebooks are liberable, but they still
need proprietary software for wifi or graphics, so there is really no freedom respecting chromebook, even if the restrictions are put in by chip providers
not Google themselves. What Google does is to design equipment with very little storage and often little computing power or RAM so that users are dependent on cloud computing for any nontrivial need. That is orders of magnitude less bad than forcing signature checks on firmware and remote controlled CPUs and so on, but it is still not good.
But the main problem is most chromebooks retain the restrictions imposed by their chip makers.
The closest to a free chromebook is an ASUS C201 (veyron-speedy) without wifi or 3D acceleration (or hardware video decoding, I think). Note it does not have an ethernet port either, only 2 USB 2 ports. So you can try an atheros wifi dongle (*) or ethernet dongle or 3G modem or something if you find free drivers/firmware. There may be others than may be similarly liberable (with Tegra K1?), but I'm not aware of people doing it.

(*) But be aware of kernel versions:

[pal666]

I was comparing Raspi Zero (the DIMM module) with microcontrollers. Given what it is usually used for, a microcontroller is also 100% fine.

rpi zero runs linux, microcontrollers don't

I'm also unsure how "cannot run linux" is an issue. They don't need linux in there, their firmware is a single C++ (with some C++ features missing) program.

microcontrollers don't need anything, programmers do.

[pal666]

FYI atmel/Arduinos can operate ethernet, use wifi

they use separate microcontroller for that, which implements fixed constrained subset of tcp/ip in firmware, which allows making few simple connections, but is not real thing

[Min1123]

BTW: I'm curious why it's not possible to do libre board from Intel/AMD offerings when google do that for their Chromebooks? Thanks!

So, Intel (they don't make AMD Chromebooks, I wish they did) has binary blobs that are required to initialize their security processor that is inside the CPU die. These blobs cannot be made Libre, and it can additionally be quite problematic. Intel's chips will turn off the main CPU if they don't get the blob within half an hour of startup (last I heard). AMD has a similar security co-processor. So the firmware is FOSS, but not Libre because of the blobs.

[kgardas]

Thanks a lot to Min1123 and especially to phoron for corrections. Indeed, I've though that when google does coreboot for their chromebooks then this is w/o Intel blobs completely. Real mistake on my side. From this point of view, then indeed Talos/Raptor Engineering attempt looks really unique.

[kgardas]

Someone from Raptor probably reads this and even reply to all my remarks on their last update to the crowdsupply web: https://www.crowdsupply.com/raptor-c...rd-on-lockdown -- it's certainly worth the read for the insight into the modern computer boot code/firmware architecture...

[Brane215]

kgardas

I've just read their rationale and find it flawed. One can't defend open hardware by bulding things no one can afford to buy.
I also disagree with regard to performance assessment. I'm wirting this on a measly Athlon 5370 ( Kabini), so with nice, prudently tweaked and optimized open system high-end hardware is not that critical.

WRT to Power9, IBM has said it will be offered to third parties and people predicting chinese boards woth Power9 for some time.

This seems as an ideal oportunity. Do open project with Power9 and then let Alibaba and its billin Thieves do the rest.

[phoron]

i want to support this true protect but for me there is no paypal and no BITCOIN and no IBAN payment option.

so you really think you get support without any SANE payment option?

and no expensive-costs- credit cards from crazy bad-banks are no SANE option

Don't tell us, tell crowdsupply. My understanding is that they have been very accomodating and sensible in the past in other projects. I don't know their risk and cost structure and whether it makes sense or not to be equally flexible for Talos, because maybe it's more money or looks less likely to be financed.

Paypal is not great. Bitcoin might be possible. IBAN might not be (in fact crowdsupply looks to me a more local USA business than it seems at first sight, which is not bad per se, I just wish there were other similar companies in more places and they could somehow federate for a project, but things are already hard as are and one must be careful wishing for complexity).

Credit card has the advantage that nobody is charged until the project reaches the milestone. Some other forms of payments require money to go to crowdsupply before the deadline and in case the goal is not reached must be returned. This carries costs, either in bank fees, handling workload, or in risks associated to customer identity to ensure money goes to the right account. I guess general procedure may be too much trouble now but they might adapt in a case by case basis and maybe in the future with more experience change the general procedure if request keep dropping in ?

Just talk to crowdsupply.

kgardas: Indeed, the Talos Update is worth reading, thank you.

[phoron]

kgardas

I've just read their rationale and find it flawed. One can't defend open hardware by bulding things no one can afford to buy.

That's no problem, if noone can afford to buy, then the campaign will fail and the only ones losing their time and money will be Talos/crowdsupply.
But it may be the only way to know. Hence the "referendum on free software" angle.

I also disagree with regard to performance assessment. I'm wirting this on a measly Athlon 5370 ( Kabini), so with nice, prudently tweaked and optimized open system high-end hardware is not that critical.

Yes, I agree. And even sticking to less powerful hardware is a way to make a little pressure in society to not impose ever newer hardware and so give the oligopolists too much power on the people. You can't possibly win always when it comes to chip factories because it requires too much capital, but you can win in stopping society to depend on so powerful hardware and on backing powerful hardware that respects freedom.

Yet, I think it's true some tasks require more powerful hardware, and some part os free software development (or open hardware development) is one.
So it's worth trying, I think.

WRT to Power9, IBM has said it will be offered to third parties and people predicting chinese boards woth Power9 for some time.

This seems as an ideal oportunity. Do open project with Power9 and then let Alibaba and its billin Thieves do the rest.

Go ahead, too especulative for me. It may come too late, if it isn't already. Everybody accepts restrictions each day as inevitable, and
we're late in resisting this mindset.

[Brane215]

No need. Just start a funding project that would pour $$$ directly to those 5 guys/gals in IBM that design boards to do one more for this purpose and open it.
Since you obviously have to pay for work, why not paying directly to the source ?

IBM is in Linux big-time and they do undestand value of openness these days, at least on the fields that suit them. SCO a**holes would have different fate were it not for IBM to stand for interests of public and engage their layers to grind them to a minced meat legal-wise.

Even first PC came out in exactly this way- as an open design, which have been picked up by Taiwanese and Chinese workshops and subsequently eaten everything else on the market.

I'm not sure that IBM doesn't intend to repeat the history, this time intentionally opening the reference design doc for just this purpose.