RIP Talos

>>58421023
another successful operation.

>appelbaum
>libreboot
>talos

>>58421023
Who did they expect to be able to afford it? And why is libre hardware so ludicrously expensive anyway?

Kek, the anon on here a few weeks ago planning his Talos build will be crushed.

I fucking called that shit though, tried to explain to him it would likely never happen and even if it did it was a colossal waste of money.

>b-but anon muh freedoms

>>58421023
I don't really get it. Why not design an ARM based machine? That way they would not have to spend many resources to design a custom CPU and instead just piece toegether some hard IP they bought off from FPGA or ASIC vendors.

>>58422398
The miracle that is mass production does not apply to small scale operations. Also they aren't a large corporation that can eat a few mill in production and R&D costs. Also a lot of other things

>>58423085
ARM architecture isn't open source.

>>58423085
>Why not design an ARM based machine?
>>58423248 and ARM isn't good for high-performance stuff.

>>58423248
Neither is Power

They should try again with something hipsters can understand and throw money at like RISC-V

>>58423293
>RISC-V
RISC-V isn't ready for any use besides microcontrollers yet.

>>58423293
>Neither is Power
It can run fully blob-free though. Are there any ARM-chips that aren't SoCs?

File: 161205-qualcommcentriq11668_sized-100697769-large[1].jpg (90KB, 700x467px) Image search: [Google]

90KB, 700x467px

>>58423316
A few maybe I'm not really sure

Qualcomms Centriq CPUs are going to be LGA, have PCI-e, and use UEFI

>$3,700 for just the motherboard
>$7,100 if you want CPU too

I think I know why it failed.

>>58423309
Someone will have to make it ready :^)

I bet it was the Fucking elves again

>>58422398
Freedom isn't free.

>try to sell unsupported hw for 20 times as much as normal PCs
>whonder why nobody can afford one

What about loongson? Are there any desktops with that?

>>58425061
Well, apparently, I can't afford freedom.

>>58421023
dont worry our favorit he/she developer/sjw leah rowe hailing from libreboot got your back http://phoronix.com/scan.php?page=news_item&px=No-Talos-Now-AMD-Libre-Effort

>>58421023
guess it's really not surprising, who actually thought that shit was going to get out the door? it was extremely expensive, obsolete from day one, very under-performant for a "workstation" and targeted a niche market that barely existed, entities that have a vested interest in covering all possible bases with their valuable information have been working on their own technologies and no scene poorfag or wannabe "security expert" that thinks the NSA gives a fuck about some peon's loli stash could've justified the expense when you could just buy a used pre-2006 system and gotten the same result

besides, if you're going to buy into the idea that mainstream chips are equipped with backdoors that are exploited regularly and effectively for every trivial reason and anyone who believes something to the contrary is just a paid shill that's lying to you, why then would you take POWER marketing as gospel just because it has "open" on it, especially when they're built by a massive American company that has heavy involvement with government agencies?

when you really think about it, it's surprising it even made it to the drawing board

>>58425094
it's not really entirely their fault, low volume hardware is always expensive and there's nothing that really can be done about it

>>58423309
I'm really hoping RISC-V pulls through and becomes something wonderful, they kind of shot themselves in the foot by using a google search operator in the name though.

>>58423440

who was making the board?

>>58425939
What google search operator? Hyphen?

>>58425181
>Well, apparently, I can't afford freedom.

I don't think a truer statement has ever been made about the current state of American, or Global, society.

>>58425749
"Performant" is not a word

>>58425220
Cool, thanks for the link.

> The Libreboot D16 desktop with the D16 motherboard, one Opteron 6272, 16GB of RAM, and the NVIDIA GTX 660 Ti / 670 is marked up at £3,450.00 (~$4188 USD) while the D16 server with 1 CPU and 16GB of RAM is £3,450.00 or up to £5,600.00 ($6800 USD) if wanting two of the Interlagos CPUs and 128GB of RAM.

> If wanting to build the system yourself and flash Libreboot on the same motherboard, you can likely build the system for around $1k USD or less: $400~500 for the KGPE-D16 at the aforementioned links, the CPUs for $16+, a couple hundred or less depending upon how much DDR3 RDIMMs you want, etc.

So as a complete system flashed with libreboot it costs almost as much as the Talos system would have, but it also has the benefit of existing.

>>58427600
it seems to be a bit of a subject of debate that is also ultimately meaningless to the actual substance of my post

>>58423309
There are 1,000 core risc-v chips.

File: orly.jpg (21KB, 413x384px) Image search: [Google]

21KB, 413x384px

>>58428379

>>58428426
Epiphany-V. 1024 cores.

>>58428379
>>58428771
doesn't mean much in the context of desktop usage though since that chip would be absolute fucking ass outside of its niche

>>58429371
That post said ris-v was useless beyond microcontrollers. The fact that a kilocore design is in production disproves that. The same company also makes risc-v microservers.

>>58429636
dunno though, in a way I'd say it's kind of true to say that since these designs all seem to rely on ultimately really simple microcontroller-tier cores working together that aren't gonna lend themselves well to a lot of applications, especially desktop use as I said which is almost always the lens through which discussions about alternate architectures on /g/ look at things

I mean you're still not wrong but maybe they aren't entirely wrong either

>>58429899
The backstory of the leap from the 64-core Ephiphany IV to the 1024 core EV is DARPA wanted something suited to low-power, scaleable machine learning and smart image/video processing.

The intereraccessible SRAM is an interesting design, and EV is 64 bit. They have no provision for a GPU or any of the myriad of ASICs found in say, an i7. If a new CPU ISA is to prove itself to the world, it first needs to prove that it can be used as a CPU. The desktop is often the last place innovation in this field is made, simply because it is predicated on the underlying (mixed computation and useful throughput here) technologies proving themselves.

tl;dr CPU has to prove it can be a good CPU in high-budget environment, and will get to lower-budget later.

>>58430103
right, I get that and it seems pretty cool, guess what I'm saying is it's just still very niche and specialized and I don't think the current implementations have proven themselves as far as what they'd end up meaning for you and I

>They have no provision for a GPU or any of the myriad of ASICs found in say, an i7.
this is kind of what makes me hesitant about RISC-V, though I haven't followed it as much, it seems a little too simple to me to make me hopeful for more consumer-y/general use prospects

>>58430268
Once the CPU is established in the purely compute space, people will pair it with GPU's and tack on ASICs. My bet is the first one to make it big would be samsung, given their financial state, I'm sure they'd love to be unencumbered by ARM royalties and contractual mazes to make stuff like Exynos (and their embedded stuff, yes).

>>58430480
could see that happening, but I don't know if it would be enough to topple ARM when the sheer profits these vendors make probably far outweighs the upfront licensing costs and royalties as well as the presence of a far more established developer base that knows the architecture well

not that the typical mobile developer would give a shit about compiler optimization and documentation, though

>>58423293
But that's wrong.

>>58425114
There's a mini ITX motherboard and an AIO, I've only searched for the mini ITX mobo but haven't found it for sale anywhere

>>58427523
yes, by typing -word you exclude that word/letter from the search
so you're basically searching for "risc"

>>58433485
It's only an issue for people who search "RISC -V" though. If there isn't a space preceding the hyphen it's not recognized as a search operator.

>>58422398
Because the NSA doesn't subsidize companies manufacturing it.

>>58421023
It was 18,000 dollars. Talos was doomed before they even began.

>>58434027
>Eighteen Thousand United States Dollars
wew

>>58434027
Since when? Last I checked it'd be around $4k for the CPU and motherboard

>>58434003
>I have no idea what the economy of scale is

>>58437445
no, 4k for the CPU, and another 4-5k for the motherboard.

It was never 4k total.

Are there any computers in production that can run properly without blobs, including firmware?

>>58438384
adus c201, though you need to use usb wifi ath9k_htc, not onboard.

>>58429636
>That post said ris-v was useless beyond microcontrollers
No, I just said that there hasn't been anything besides microcontrollers based on RISC-V yet.

File: 1482996910927.png (39KB, 321x322px) Image search: [Google]

39KB, 321x322px

>>58437974
christ almighty

>>58423316
I think they're all SoCs, but some can run blob free. I think the Asus Chromebook C201 is as powerful as you can get at the present though so it probably needs more work.

>>58423085
They didn't develop a custom CPU. OpenPOWER is an IBM CPU.

>>58442270
They're still paying to make a CPU that isn't otherwise in production.

>>58423823
Thalmor a shit

>>58437974
By the nine

>>58442304
it's just a regular production chip that IBM uses in their own products

>>58442425
So was this not actually making anything new and instead just bulk ordering processors and motherboards from IBM then?

>>58442523
they build the board, the CPU was IBM designed and manufactured fuck knows where, OpenPOWER isn't a distinct line of chips, it's just the specification

>>58442304
Any cpu running a IBM POWER8 compatible isa would have worked with the mainboard, there are such chips used in servers on the market already. The cost was developing an ATX compatible board with the crazy specs and features they envisioned, especially as they presumed their market would be so small.

I still don't understand their beef with risc-V, does someone who actually knows want to confirm that, as they stated, in its current development stage it offers little better than a low end arm chip performance wise?

>>58442560

Its fucking hard to design high frequency chips which is why so few do it.

I don't know much about RISC-V but if its marginally better in performance to an ARM design then then its not fit for workstation use. Power 8 is designed and field tested for HPC so why bother with ARM and RISC-V? Besides, every single so called "competitive" ARM server chip in the past or in the pipeline has been a disaster (either late and or way under performing). I am not saying that and ARM or even RISC-V designs couldn't scale but who is going to front the tremendous resources need for it to be competitive in HPC markets?