They have created an open source "daughter board" for this purpose.
As it stands, System76 has mediocre branding and design in my personal opinion. Not great, but also not bad. I am not excited about the wooden trim approach integrated in the PC case. But, these things are subjective. If they nail ID, I believe they have a really bright future. When I say, hire a top notch designer - I mean, someone who can change the world type; not someone that they can just hire by posting a job ad. But, it has to start from the owners of the company. If they think Design as a cost center in marketing, forget about it.
They also copied a page from Apple’s eco playbook and will plant a tree for every machine sold which is nice to see. A solid move given the company’s size and reach. I wish more IT companies would use that as a marketing move, it has good signaling value and helps drive change.
I’d like them to do even more with the wooden look and feel, but their cases are still not far removed from a typical desktop. I think it would be great to see them do something really small, in the Mac Mini style, but with that warm wood look. That would look great on any desk.
I’d also like to see someone Hackintosh those machines just for fun.
I disagree with the idea of creating "fresh" and "trendy" design. In all cases, except say Fashion, it creates substandard design to cater to originality & staying trendy. Timelessness cannot be realized if your design philosophy is to try to stay "fresh".
What is fundamentally wrong with metal & clean design? The answer shouldn't come from the fact that "Oh shit, someone is already doing that, let's do something different. May be wood?". The answer should come from fundamental understanding of materials, their use, durability, haptics & touch, engineering needs (bend radius, coefficient of thermal conductivity, flextural ridigity, etc.) and DFM aspects. Everything else is decoration, not design.
As Paul Rand said, "Don't try to be original. Just try to be good."
And although this isn't strictly the point you were making, wood is an excellent material to house electronics - it absorbs moisture and heat well, is extremely durable, it is a renewable resource, and while it goes in and out of fashion through the decades, its consistently attractive when used judiciously. This desktop case probably just has a sliver of wood panel which exempts it from these benefits, but I hope in the future we will be able to use natural materials more effectively, perhaps even to make the computer hardware itself.
In the case of System76, they had to bend wood vaneer (poor bend radius) which probably added additional expense to the manufacturing process.
You can actually cut the stuff with a normal pair of paper scissors. It's pretty wild stuff.
Nothing is "fundamentally" wrong with it. In fact we have gotten to a point that laptop chips can function just fine for many tasks in suboptimal chassis because they are fairly efficient, let alone desktops. The reasons to choose a different design as a customer at least are personal. In my opinion the minimalist industrial design is overused and interacts poorly with my brain. In the absence of any design elements on it I find the products bland and with little character. I have a hard time remembering another glass/metal phone or another bland-painted building with aluminum stairs. Apple knew that years ago and imo it's probably why they stuck with the iphone 6 design for so long, so the market floods with it and thus it becomes memorable. The iphone 10 has more glass and the notch, so in a way it's scaling back the minimalism, closer to the iphone 4/5 era when the phone had a few more geometric elements on it.
In my opinion what is missing from every modern design, be it the applesque industrial or google's kitsch "throw all the colors that don't mix together, remove all textures" material design are little things that make the product memorable. Ideally these come from quirks that add usability or from the materials used. In a different era with a more limited materials available choosing the right thing for the job meant more diverse products. A macbook today is not more memorable than a razer blade or some other copycat. The same goes for chinese phones and iphones. I've confused an iphone with a cheap android replica so many times. Industrial design works when you are the only player doing it. But apple is not.
Product design is a relatively new thing and it is going through the same phase that architecture did with modernism and brutalism( and post modernism probably too). But this endless streamlining and simplifying of devices is not as positive as designers want us to believe. At some point they'll run out of stuff to simplify that people will buy and will try to differentiate by going to the opposite direction or just stick with brand recognition. So while I don't think a wooden pc case is the end all be all of design I'm glad s76 is doing something different. And this kind of change always comes from the "alternative" underdog.
And has anyone achieved timelessness? Ever? I cannot immediately think of any industrial design which does not feel either "current" or "dated".
Imho, it's just that the case does not feel nice. While reading the specs, you have the impression that they have good engineers, that they carefully select components and design the air flow... but the whole case is still not there. As if someone made a draft and stopped there, carrying a "good-enough" tag.
I agree with the GP: They would benefit from an outstanding designer, and that comes with CEO blessing and will.
In fact, not sound as bad. Is necessary to get more details of the hardware to check compatibility, but could be a great option
Not that there is anything wrong with it, but there has been so little innovation in PC design for so many years. Everything is still dominated by LEDs and sharp edges and some desire to appear "techy"
It's like we're still living in an era where computers are a novelty, not just another thing in the house
Coolermaster served that role for years but seem to have vanished.
Recently I ended up putting nearly £2K worth of equipment into one of these.
It's quiet, neat and doesn't have a glow in the dark multi colored LED anything anywhere.
Eventually I decided on Fractal Design's Define series, they have variants with and without the windows. Now I have a Define RS 5 Black (without the window!) and it is a great quiet case with many details, such as keeping dust out with meshes and vacuumable mesh filters.
The System76 case looks nice, my initial impression was that it looked almost like a quality speaker with a kind of walnut finish. On further examination, it did not look so great, unfortunately. I'd like to see that tree part replaced with a dark metal-bluegrayish matte metal or something.
(another good one was Sharkoon REX8 or its predecessor T6 Value but both aren't apparently produced anymore, bah)
As for why most cabinets look so similar outside of that, I’d say it is probably at least in part thanks to such designs being the best we have in terms of airflow and ease of access to components within.
Going into an organic artsy & crafty direction vs. high-end industrial design like Apple is a good choice if you support the DIY spirit (e.g. Linux & open source), it's more authentic in that context, and you won't be compared to Apple or perceived as imitation of them.
It's also a good alternative to the cyber futurist gamer look that dominates the PC industry.
I don't know, it's looking better now than some generic boxes you see from big name companies.
Just to quote two examples:
But I like the lighter wood. It feels almost midcentury modern.
More importantly, someone's made a tower computer that doesn't look like something from a 10-year-old's comic book collection, or every Wintel box made since 1993.
But the value of the compute power for the buck was top notch. Also, their support was always stellar. Ill certainly buy form them going forward, just wish the gazelles had more physically sound chassis.
So, i would just want to get my hands on a thelio to test out. If System76 is listening, how bout getting a base model to me to review here on HN. :-)
i am looking at this laptop:
it's obvious these guys took their design cues from apple's laptops, but they've paid about a tenth as much attention to the details as apple would. unnecessary bumps and blemishes, odd discolorations here and there, couldn't be bothered to do anything about that hideous rectangle at the top of the screen where the camera is located, etc.
i am contemplating a career move (no more ios development) that would mean i am free to leave apple's ecosystem. if some other company could make laptops as beautiful as apple's, without duplicating the mistakes apple has been making over the last few years, it would really grease the skids for me.
At the top of that list I would probably put the Dell XPS 13 , and it can be bought with Linux. HP (Envy/Spectre) and Lenovo (x1 carbon) have laptops that look to be very high quality, but I don't have personal experience with them and they don't ship Linux.
For one, ID is not a mutually exclusive endeavor. If a company has A, B, C and D. Adding E, in this case - ID, only adds to the value of the company. Good ID never diminishes it.
I am comparing System76 with 70's IBM, Apple and Braun in my original comment. If IBM didn't pay attention to design - sure, they'd be successful probably. But ID has only added to their goodwill and brand name. Unfortunately, it is no longer true with IBM today. :-(
Executives that pay attention to ID and its value create a direct unspoken visual link with their customers. Customers notice the smallest details, perfection in every aspect of the product and they realize & get the impression about company's dedication towards other non-ID related things. Stripe does this really well. It is even more critical for a physical product that you touch and feel. ID is sort of like a short-cut of communicating the abstract concept of quality to customers. It works brilliantly as a marketing tool as well as a tool to deliver actual measurable benefits to the customer (ease of use, clarity in operation, aesthetic appeal, etc.)
It turns out, if you are manufacturing original products for mass production, you are doing ID. So yes, there is no path to success manufacturing original products without doing ID. I wasn't implying one needn't have ID any more than your comment implies one needn't have engineering.
What I was referring to your comment about "hire a top-notch industrial designer" to produce distinctive designs of impeccable taste.
Yes, products with exceptional ID will sell in major ways.
It turns out, that products that are exceptional in a lot of dimensions sell in major ways. Products that are exceptionally inexpensive sell in major ways. Products that are exceptionally marketed sell in major ways. Products that are exceptionally well engineered sell in major ways.
...and if everyone is employing exceptional ID, then doing so yourself doesn't actually make your product exceptional.
And no, customers don't "notice the smallest details, perfection in every aspect of the product". There are some products that certain customers fetishize and absolutely do notice this stuff, but the vast majority of products people interact with on a daily basis barely register.
I agree that ID is a short cut for communicating quality to customers. Sometimes you shouldn't take the shortcut. Sometimes it's better to take the long way around.
Great ID can work brilliantly as a marketing tool as well as a tool to deliver actual measurable benefits to the consumer. Great engineering can do the same. Having a unique value proposition (which I'd argue is System76's case) can do the same.
There's more than one kind of market success and more than one path to market success.
Don't you feel that there are a lot of products out there that are not designed well? They're just slapped together, molded shut and ready to mass market? ID would just be a one time NRE cost if they don't use expensive materials.
If you get a chance, do catch the "Dieter Rams" documentary by Gary Hustwit. I thought it was amazing and you'd I think enjoy it.
I do, but I feel the same way about engineering. Curse of high expectations.
Don't get me wrong, I think it's great they made a cool looking case and threw an AVR in there for controlling the fans, and if I wanted to buy a new desktop, I'd seriously consider one of these...
...but open source hardware it is not... and the claim about it being manufactured in Colorado seems a bit specious. Maybe they're milling the cases in Colorado and installing the parts? The real parts that do the computer part seem to be the same as the real parts in any PC.
They tell you exactly what they mean in that paragraph:
> US-sourced wood and aluminum are formed, finished, etched, and built by artisans in our Denver, Colorado factory. Premium components from around the world are then assembled to your final specifications. And to offset environmental impact, every Thelio sold plants a tree with the National Forest Foundation.
So yes, they make the case in the US and assemble the parts in the US.
They mentioned multiple times that their intentions are to make the hardware as open as possible with each generation. They never claimed that this hardware is 100% open. My guess is that it has to do with how many parts they can order from mobo manufacturers so they slowly release more and more pieces.
>and the claim about it being manufactured in Colorado seems a bit specious. Maybe they're milling the cases in Colorado and installing the parts? The real parts that do the computer part seem to be the same as the real parts in any PC.
That's correct. But coolermaster or corsair cases aren't necessarily made in the USA while these are. Basically what they are saying is that everything they do themselves, including the value-adding steps is done in the USA.
I so wish they had put this bang in the middle of their home page.
I think open hardware is the future, and those who are able to get there first are going to see increasing gains as time goes on. Of course my engineer friends always talk about how you can never verify down to the chip level, but I think there are ways it can be done, the market just hasn't been pressured to do it... yet.
Now if we can just start by getting Amd PSP and Intel ME removed! The first CPU manufacturer to do this is going to gain huge marketshare. (hint hint AMD!)
Even if a battle is uphill it doesnt mean its not worth the fight.
Danny Casalaro, PROMIS, and the octopus, and all that... the point being this is nothing new, just newly pervasive.
Board members and shareholders might (some day -- hopefully) care, too.
This is impossible. The ME/PSP are instrumental in the boot process; modern multi-core CPUs are complicated to initialise and these chips generally need substantial initialisation before the main cores are ready to execute instructions. This role is not something put in there gratuitously out of malice to stop people removing it.
The problem is not with having the ME/PSP as an auxillary core per se, but that it only executes closed-source vendor provided blobs (nor is documentation available to allow an alternative to be made), and more critically that it only executes vendor-signed code, precluding owner control.
The ME/PSP are also used to implement DRM, so Intel and AMD are not only not unlikely to ever enable owner control of the ME/PSP, it is likely they are contractually prohibited from doing so. I've previously explained this: https://www.devever.net/~hl/intelme
Then they need to put something in that does that instead of always-on, networked backdoor you cant disable. There's a huge gap between what you described and what they're doing. They also used to build complex, custom CPU's without it. So, we know it's unnecessary from their prior products.
I'd rather use the space for an extra core, a low-latency GC, TCP/IP offlad, etc. A lot of stuff people or companies might pay for who didnt want or use ME. The enterprises that want ME can buy chips with it.
Well, I don't think that is the general case for software, that people will manually assamble a bootstrap assembler from machine code (and how would you verify your editor? Maybe write to floppy and verify the bit pattern via electron microscope?) - and build up from there.
Proving an entiresystem secure would be very hard.
But to have parts that you can say "if I trust this, then it follows that I can trust all of these things that follow" is very useful.
And so it is with hardware, too.
Not to mention how useful it is to have a well documented, open platform (even if it's not libre), like eg the commodore Amiga 2000 that came with full schematics of the motherboard, along with measuring points, so it was feasible to find and replace faulty components, write drivers and build third party add-ons.
How are they not doomed to go out of business?
People, especially the professionals, want the complete experience, from hardware to support and everything in between. Can you provide the level of support they provide while being cheaper than them?
> With zero overhead I can sell everything they make cheaper.
Chinese brands do this all the time with Apple. They copy the Apple iPhone designs wholesale including the UI themes, make it cheaper, and yet Apple makes more money than all of them combined.
Some people want the "real deal". In fashion, that's because some consumers are superficial and vain and want to signal their wealth by getting the official stuff instead of using their wealth for something worthwhile. But for a case like System76, some consumers might care about funding the company that is doing the innovation that they appreciate…
At any rate, your question applies to all open-source software — how is it not doomed to be unfunded or go out of business? It's much easier to distribute software than to compete with a hardware company!
I'd rather buy from the company pushing the envelope rather than some obvious company just trying to make a quick buck.
The Pursa i3 is a fully open source 3d printer with countless clones available but still the creators company is the most popular source to get it from.
It takes a lot of knowledge + time + connections to make the whole products worth it.
And remember that they will always have the advantage that the knowledge of hardware and firmware is in house, while you rely on their updates
I’m hoping they do a nice laptop. HN will need to put their money where there mouth is if System76 comes up with a thick laptop with replaceable battery, upgradeable RAM, lots of non USB-C ports and a keyboard with some travel. A niche System76 could really fit into.
Found the compromise, the battery is tiny.
With the circa-2009 Acer plastic look? Big plastic air fins, feet that stick out the bottom to stop it sliding into t bad properly, huge (huge) bezel around the screen?
The Oryx looks much more modern - are you sure you aren't talking about that? https://system76.com/laptops/oryx
Every laptop either looks like this, or a MacBook clone.
I'll buy one anyway, simply because it's supposed to be made in America. But I can't find anywhere on the web site where they're made.
The main problem I've seen eith that sort of thing is that there still aren’t many 5k monitors on the market and I’m not sure the LG one will work with Linux over USB-C. (Would love to be wrong on that if anyone knows better).
The wood finish? Really? A desktop that looks like a trash can, and now this?
I really hope they continue their open source aspirations with their laptop line as it well be a compelling alternative to the typically macbook dev machine.
There are plenty of people that are developers or Linux users (e.g. data science) that don't know anything about hardware nor do they care to know anything about it they just want something that works, sure they can spend time "researching" it but why bother? famously even Linus said he doesn't install his own linux distro not that he couldn't but even he would actually need to do some research if he would want to build one completely from scratch.
So their target market is essentially the person who would buy a Mac Pro or a dell workstation and if they have good support for theirs both in terms of FOSS software and post purchase hardware support they have a market to sell too.
I know people that spend 5-10K on a Dell workstation just to have "official" Linux support and decent onsite service, if these would sell outside of the US they would likely to switch.
I could have built it myself. But I didn't for three reasons:
1) I have other work to do. Spending a day tinkering with components instead of writing papers, getting grants, or working on code is not a good use of my time.
2) Ordering a machine is easier in terms of working through university purchasing than ordering parts.
3) If something's broken, I can call someone.
All of these have value.
Clearing a day of meetings and writing so I can deal with cable management? Meh.
It looks like System76 has delivered precisely that series... and at a reasonable price. Thank you guys!
That being said I really like the machine.
I'm a little concerned that the custom I/O daughterboard might make them only compatible with Linux or the PopOS (custom Ubuntu) that System76 offers.
Seems pretty insane to charge prices like this ($2,500 for a modest i7, 32G ram, and 256GB SSD) and not make it expandable enough to handle the more challenging uses.
But my current default build would be much cheaper and support up to 64GB ram, see https://www.pc-kombo.com/share/mORbPdR4. And that's also in a small case.
Personally I find this industrial design to be a refreshing change from other PCs but I'm sure the wood grain will be divisive.
> Chia looked down at her sandbenders. Turned off the red switch. "Coral," she said. "These are turquoise. The ones that look like ivory are the inside of a kind of nut. Renewable."
> "The rest is silver?"
> "Aluminum," Chia said. "They melt old cans they dig up on the beach cast it in sand molds. These panels are micarta. That's linen with this resin in it."
For the twenty years of Idoru they could have sown some discrete references to Gibson :).
It’s been a while since I read it but I seem to remember Chia was able upgrade her system. It would be nice to see a System 76 offer logic board upgrades for the case too.
I mean you can get motherboards from anyone but if you get them from System 76 you know it’s got good Linux support.
Why do they keep marketing Ubuntu as a different OS? The is confusing to people who don't realize that it's literally just a customized Debian.
Why do they keep marketing Debian as a different OS? It is literally just Linux and some GNU system components cobbled together.
If I suggest running Linux to some of my friends, they'll immediately disregard it as something they wouldn't be able to use, possibly because they know I run it and associate it with being a techie. I suspect I would get a very different reaction if I showed them something like elementary OS without even mentioning the word Linux.
That is not true for Debian. Some packages (like the .deb file for Chrome or VSCode) should work on any distribution that supports dpkg, but there are many packages that break compatibility between Ubuntu and Debian. Thus Ubuntu isn't just a customized Debian; it's a customization that creates a distinct operating system.
For many people in that audience "Linux Distro where the latest CUDA just works" is a huge selling point. That is unfortunately not guaranteed to be true of Ubuntu, necessarily, although things have gotten better.
I just switched to it recently and am very happy, it feels like it was perfectly designed for how I use the computer.
It seems to work very well on my laptop. I would say it is the first real innovation in distros in years.
I built myself a Threadripper 2990WX based system a few months ago, and the hardest part of the built was ECC support. In particular:
1) confirming ECC support on the mobo was not disabled by the mobo vendor
2) finding "reasonable" speed ECC UDIMMs (ended up w/2666) in a high enough density to reach at least 64GB.
I would have been happy to pay a premium to a company like system 76 in order to save days of research into different parts.
I ended up getting Crucial CT16G4WFD8266 dimms. Crucial was painful to deal with, and kept cancelling my order because they "could not verify key information". I finally just ordered the mempory from provantage.com, for slightly less. The underlying micron part number from dmidecode is: 18ASF2G72AZ-2G6D1
If I was doing this today, I might wait for 2933 or even 3200 memory..
Configuration total $76,783
Talos is a fully open hardware platform too.
Talos has fully open source firmware, down to the very first POWER instructions that are executed from efuses. That's pretty damn open.
From what I can tell it's a bog standard system, a thin piece of wood (or is it a veneer?), and a custom fan controller on a small daughter card.
The integration is pretty nice, presumably all the driver related integration "just works", although I generally get that with dell desktops or newegg systems built from parts.
They don't mention much about the actual quality though. How efficient is the power supply? How many fans does it have? Any sound insulation inside? Maybe a aluminum/plastic/aluminum sandwich for sound dampening?
Why is the rear fan so far from the rear of the case?
Why is the motherboard horizontal which makes the case very wide, especially for a machine with a max of 32GB ram.
For $2,500 for a modest i7 system with 32GB ram, gtx-1070, and a 256GB SSD I was expecting more than a custom fan controller and limited ram.
You get that this is open hardware right? You can go to https://github.com/system76/thelio/ and see the whole story (there are separate projects for Thelio Io and the specifics for each model, but you get the idea). They even have a BOM: https://github.com/system76/thelio/blob/master/Thelio%20Comm...
Here, for example, is all the detail you could hope for on the PSU for Thelio Main: https://github.com/system76/thelio/tree/master/Thelio%20Main...
The Arctic case fans are not the very best but very good for their price, reasonable choice as well. I use them myself, in my very quiet PC.
I don't see sound insulation in the BOM, I might miss it. But the regular customer hardware used there I can judge seems very reasonable picked. Not the usual bad cost optimization you see in regular pre-builts.
If went out and bought the PSU yourself, you wouldn't have any more information. I think that's a reasonable standard to hope for.
And yes, PSU vendors aren't advertising their products as open hardware. I'm aware of this. So people shouldn't be making out that this PSU is open hardware. It isn't. Whether or not people are fine with that is up to them. Personally PSUs are on the bottom of my list for things to care about the openness of.
The fan appears to point inwards, which seems to make more sense for BTX Dells with the fan at the bottom than this with the fan on top.
The site mentions the power supplies are 80plus certified or something with (iirc) 90% efficiency or so.
Quality of the case wise it does look like quite some effort was put into it, given the top appears to be one piece and it certainly looks harder to manufacture than a standard PC case.
With that said, as much as it is nice, I think it's just more of a standard PC with a nice case.
I don't care much about walnut or maple; I just want quiet fans.
In other words, they donate a dollar. Not that I'm against it, I guess. There's just something about the "buy our product for $X and we'll donate $(X/1000) to Y on your behalf" marketing ploy which rubs me the wrong way. I mean I guess it's better than them not doing it, but... it just feels bordering on sleazy.
I'd much rather see "We donate N% of our profits per year to planting trees". It's a subtle difference, but instead of a marketing technique, it feels more like a genuine company principle.
I was testing it today and it seems to work very well on my desktop. I would say it is the first real innovation in distros in years.
The FOSS world should keep the middle finger up until novideo starts releasing actual detailed documentation and collaborating with the nouveau community.
The open hardware part is extremely appealing, and the reason I started paying attention to System76 in the first place.
What's missing for me is form factor. Rather than a desktop tower, I want to rackmount a machine in an AV cabinet and put it in my living room alongside other devices like an ethernet switch, UPS, maybe an AV receiver, set-top box, or other bits of AV/smarthome type gear. Like those home stereo racks people had in the 80s and 90s (tape deck, cd changer, receiver, record player, amplifier). But with actual rack mounting hardware like you get in modern server rooms.
I already shopped around for, and have, a suitable cabinet:
So when I look at their 'servers' page I see machines that belong in some company's server room, away from humans. The small height (1U or 2U) says "compute density" (which is one of the metrics emphasized on their product page) and also means small fans which need to move a lot of air, which means noise.
The servers offered go up to 28 or even 31.5 inches deep. The shallowest is still 22 inches deep. My aforementioned cabinet has 16.5 inches of depth from the front rails to the back rails. So while the rackmount hardware is the same standard 19 inch width with vertical hole spacing measured in U's, there is a distinct difference between what is built for a server rack and what is built for an AV-style cabinet in someone's home.
A chassis with 4U or more of height would allow larger, slower 120mm fans suitable for a quiet living room. I want something like that on sliding rails so it can be easily pulled out like a drawer for component changes.
That would be a great bit of design for a home server, I think. And if System76 or a similar vendor were offering this as a product category, I'd be asking them to take my money :)
You can see it among the benchmarks here.
It's the slowest tested obviously, mainly because of limited memory (so you have to put less on the device and compute less in parallel). But it would still be a huge speed up over CPU, like night and day. And "as fast as a 1080Ti" is really, really great.
I will add, Pop!_OS comes with the latest NVIDIA drivers and lets you apt install the latest CUDA and even Tensorflow without any rigamarole.
As I understand it, AMD is pushing more open standards, although I don't know how it compares in terms of performance. I actually have a tab open right now with the ROCm  docs, since I was looking to learn a bit more about what they were up to. They supports Tensorflow, although I don't think they're first-class citizens like NVIDIA.
>Open hardware licensed to give you rights
>Thelio, Thelio Major, Thelio Massive, and Thelio Io are >OSHWA certified open source >hardware.
The case might be. The mainboard, i.e. the actual computer, sure as hell isn't. Insofar that they're selling "Thelio" as a computer, a reasonable person would interpret this to mean the computer with all parts therein as sold, for which this statement is wholly false. As for the mainboard, my understanding is that it's both designed and manufactured by Gigabyte outside of the US, and the schematics aren't theirs to give.
>Designed and Manufactured in Colorado
>US-sourced wood and aluminum are formed, finished, etched, and built by artisans in our >Denver, Colorado factory. Premium components from around the world are then assembled to >your final specifications.
The case might be. The mainboard certainly isn't. "Premium components from around the world are then assembled to your final specifications." is pretty ambiguous. I wouldn't normally consider a mainboard a "component", it is the damn thing. This is at best highly misleading.
Moreover, according to , the company is "chipping away at the proprietary bits until it's 100% open source." This is, like Purism, stating vague ambitions which cannot plausibly ever be achieved; Intel and AMD platforms will never be blob-free.
Also, I hate to be that guy, but if you are trying to compete in this market, why even spend time doing a Desktop computer. It is a super shrinking market... the only folks that still buy these are gamers (where sadly you just have to live with the proprietary stuff to squeeze every cycle for performance) or something hyper niche that doesn't matter anyway.
Concentrate on one product (probably laptops) and make something outstanding there. There are some of us that want to buy it.
I also have two laptops obviously but my main workhorses are the desktop computers. I'll even to shell into them when I'm on my laptop.
I don't have any data on it but I feel like the desktop market is getting bigger as datasets are growing and software is becoming more distributed. My 2015 macbook sounds like an airplane taking off when I code on it. The fan is at full throttle 100% of the time.
As for gamers of course the majority will use Windows but Win10 is such a shitshow that Linux is going to get progressively more popular, especially now that Steam Play/Proton makes it super easy to run Windows games in Linux.
* Game development engineer who needs to test games at all graphics settings;
* Digital artist who needs to rapidly iterate by creating draft renderings quickly before sending the model/scene to a render farm;
* Roboticist developing robotics algorithms that require lots of local data collected in the field that would be infeasible to upload to AWS;
* Physicist developing computational models which should be tested on smaller instance locally before sending off to a supercomputer.
Besides, if you're planning to use a machine for compute for more than several months, it becomes cost-efficient vs renting an AWS instance.
It manipulates 1's and 0's at a rapid rate and high precision.
"If I need that much compute for a problem, doesn't renting the capacity on AWS make 100x more sense? There's no way I could actually make use of all that hardware"
It's probably not for you then, just like the AWS instance wouldn't make a lot of sense for some guy playing Crysis.
"Perfect with Pop!_OS
Pop!_OS by System76 and Thelio together form the perfect platform to create and discover. Thelio is optimized for maximum performance. Pop!_OS provides tools and development platforms that are always up-to-date and just a single click or command away. "
Forgive me but I'm just not interested in running an obscure custom Linux. I feel the same about Amazon's AWS Linux distro.
Give me one of the following, in order of preference, and we'll talk:
When there's a kernel upgrade it "just works", presumably with DKMS or similar. What pain are you seeing with nvidia?