Most active commenters
  • creshal(9)
  • satai(5)
  • keenerd(5)
  • david-given(4)
  • ggreer(4)
  • dandelion_lover(3)
  • nextos(3)
  • throwaway7767(3)

←back to thread

154 points walterbell | 61 comments | | HN request time: 2.884s | source | bottom
1. INTPenis ◴[] No.10736741[source]
Since I'm completely surprised by this project and very attracted to it I thought it was best to google around for some perspective. Found this http://www.pcworld.com/article/2960524/laptop-computers/why-...

Among other things. My first question was, is the hardware open? Couldn't find an answer to that.

Edit: Apparently revision 2 of Purism will possibly have Coreboot.

replies(3): >>10736758 #>>10736798 #>>10736827 #
2. darkr ◴[] No.10736758[source]
I think it's on the roadmap according to this diagram, but not sure how confident I am that they will ever get there:

http://cdn.arstechnica.net/wp-content/uploads/sites/3/2015/1...

3. conradev ◴[] No.10736798[source]
Also worth noting is the Novena, which has similar goals: https://www.crowdsupply.com/sutajio-kosagi/novena
replies(1): >>10736813 #
4. satai ◴[] No.10736813[source]
Novena contains 4x Cortex A9 CPU, thet is much less power and not enough power for more serious work :(
replies(4): >>10736880 #>>10737090 #>>10737138 #>>10738414 #
5. creshal ◴[] No.10736827[source]
The CPU uses proprietary, binary microcode blobs.

The graphics chip needs proprietary, binary firmware blobs.

The ethernet chip needs proprietary, binary firmware blobs.

The BIOS is a proprietary, binary firmware blob.

"Respects your freedom" my ass. The only difference to a whitebox laptop is marketing. Dell's or Lenovo's linux offerings are just as "free".

(And chromebooks with Coreboot are, technically, more free than both.)

replies(4): >>10736975 #>>10737206 #>>10739064 #>>10739904 #
6. random778 ◴[] No.10736880{3}[source]
I would argue that it's really powerful for certain, niche applications. Having an FPGA onboard is awesome. Now if we can get software everyone uses (office apps, browsers, and the big ones, not specialized ones 1% of people use) maybe the Novena will be fine. We should get that anyway so mid/low-range smartphones can replace the PC for most people.
replies(1): >>10736944 #
7. satai ◴[] No.10736944{4}[source]
You are right, it has it's own (sadly small) niche. But it's hardly comparable to Librem.
8. dandelion_lover ◴[] No.10736975[source]
They are not hiding it at all. Their goal is to achieve freedom in each of those components [0].

[0] https://puri.sm/road-to-fsf-ryf-endorsement-and-beyond/

replies(1): >>10737067 #
9. creshal ◴[] No.10737067{3}[source]
So they're selling vague promises. Getting everyone to open source their binary blobs (and the Librem has a lot) is highly optimistic at best.
replies(2): >>10737388 #>>10737977 #
10. david-given ◴[] No.10737090{3}[source]
I just benchmarked my shiny new Asus Chromebook Flip (4-core ARM) against my ludicrously overpowered desktop (12-core Xeon E5-1650). I ran povray --benchmark, so it was a float-heavy number-crunching exercise.

The figures were about 1500 CPU-seconds for the desktop and 3000 CPU-seconds for the Chromebook. Of course, wall-clock time was significantly less for the desktop due to having many more cores, but that's showing that per-core, the high-end Intel was only about twice as fast as the ARM.

I do development work on the Chromebook, via a Debian chroot inside Crouton. And you know what? It's fine. It's probably the fastest laptop I've ever owned. The filesystem's a little slow, but compilation speeds are perfectly adequate.

replies(4): >>10737395 #>>10737615 #>>10737658 #>>10738389 #
11. tinco ◴[] No.10737138{3}[source]
Are you seriously saying a machine that can process over a billion instructions per second, four times in parallel is not powerful enough for serious work? What kind of serious work do you do?
replies(1): >>10737218 #
12. nextos ◴[] No.10737206[source]
Actually, a RockChip based Chromebook like C201 is completely free except for the 3D acceleration. Not even CPU microcodes. And it's dirty cheap.

I wonder why Purism didn't simply commission such a machine with the right 3D chip instead of going with a non-free and expensive option.

I would also love similar initiatives in the mobile space, but I reckon it is more challenging. Neo900 and Pyra are kind of cool though. And I'm hoping Jolla open sources Sailfish OS later this month or early new year.

replies(4): >>10737228 #>>10737541 #>>10740311 #>>10745038 #
13. satai ◴[] No.10737218{4}[source]
It includes several open firefox tabs and running an IDE and a compiler.

Anyway Cortex A9 is barely enough for web surfing if you visit some more demanding pages.

14. creshal ◴[] No.10737228{3}[source]
> I wonder why Purism didn't simply commission such a machine with the right 3D chip instead of going with a non-free and expensive option.

Because they can sell the "expensive" option (which, for the OEM itself, isn't even too expensive) at a much higher premium.

> I would also love similar initiatives in the mobile space, but I reckon it is more challenging.

In the mobile space it would be an even bigger exercise in futility: There is no, and will never be, a baseband chip with a free firmware. The FCC made that pretty clear back in the OpenMoko days – use our NSA-approved proprietary blob or you'll never sell in the developed world.

replies(4): >>10737623 #>>10737654 #>>10737849 #>>10738650 #
15. dandelion_lover ◴[] No.10737388{4}[source]
They are by far the best except for the Ministry of Freedom [0], which sell FSF-certified, but comparably slow laptops. And they admit that as I mentioned above.

Only after enough users is involved in freedom seeking, it can be possible to demand large companies to provide something we need. In my opinion, Purism do a lot in this direction.

[0] http://minifree.org/

replies(1): >>10737531 #
16. ggreer ◴[] No.10737395{4}[source]
I don't think your numbers are correct. My Core i7-4770K (4 cores) runs povray --benchmark faster than your 12 core Xeon:

    Render Time:
      Photon Time:      0 hours  0 minutes  1 seconds (1.256 seconds)
                  using 11 thread(s) with 1.456 CPU-seconds total
      Radiosity Time:   No radiosity
      Trace Time:       0 hours  2 minutes 18 seconds (138.426 seconds)
                  using 8 thread(s) with 1099.214 CPU-seconds total
That's POV-Ray 3.7 with no architectural optimizations. I just apt-get installed it.

I would be very surprised if the Chromebook's 1.8GHz Cortex-A17 was only 3x slower. Googling around, I see people mentioning numbers like 10,000 CPU-seconds.

replies(1): >>10740464 #
17. madez ◴[] No.10737531{5}[source]
What things did they do beside marketing? What was achieved?
replies(1): >>10740819 #
18. revanx_ ◴[] No.10737541{3}[source]
Would love to know where you got this information from as in Chromebook C201 has no hardware blobs except for 3D acceleration.
replies(1): >>10737850 #
19. satai ◴[] No.10737615{4}[source]
The issue is Cortex A9 not ARM. Flip has Cortex A12 or Cortex A17.
20. rtpg ◴[] No.10737623{4}[source]
what's the story on OpenMoko? I have a hard time seeing the FCC directly saying something like that, and google isn't showing anything...
replies(3): >>10737905 #>>10739482 #>>10739568 #
21. Gregordinary ◴[] No.10737654{4}[source]
The FreeCalypso project is working on free baseband firmware with an older TI Chipset. If I recall from their mailing list, which is fairly active, they have voice calls working now. It will of course not be a smartphone, and it'll be GSM only.

https://www.freecalypso.org/

replies(1): >>10737680 #
22. satai ◴[] No.10737658{4}[source]
(I am not ARM hater, the oposite is true. I admire the work Apple did when designing A9 -not the Cortex one - and I hope I can buy a lightweight laptop with Cortex A72 or Cryo next year. But this doesn't make Novena effort any way usable for heavy web usage or more demanding tasks or comparable with i5 notebooks for work..)
23. creshal ◴[] No.10737680{5}[source]
How do they plan to get it FCC certified? Without FCC certification, it may not be legally used outside shielded testing environments.
replies(1): >>10737807 #
24. Gregordinary ◴[] No.10737807{6}[source]
I'm actually not sure, recently joined the mailing list and have been passively monitoring.

Would it make a difference if the chipset being used was already used for a cellphone that was FCC certified? If I put DD-WRT on my router, do I need to re-apply for FCC certification? (Wondering)

replies(1): >>10737897 #
25. nextos ◴[] No.10737849{4}[source]
But see how the Neo900 guys have worked this out. They isolate the baseband chip, and handle it as a threat, which is a very good model I think.
replies(1): >>10737907 #
26. nextos ◴[] No.10737850{4}[source]
http://www.libreboot.org/docs/hcl/c201.html

Well, also the wifi requires a blob, but one can use small usb adapters sanctioned by FSF, and blobfree.

27. creshal ◴[] No.10737897{7}[source]
I'm not sure whether the certification guidelines for wifi and cell devices are the same.

For Wifi it's surprisingly strict:

• Every antenna+transmitter configuration has to be certified separately (that's why Lenovo and other laptop vendors have Wifi card whitelists and refuse booting with uncertified chips installed).

• The software that directly drives the hardware must be certified to conform to the transmission power limits etc.

For DD-WRT and others neither is a problem, because the hardware combination has been certified by the router vendor, and DD-WRT uses the wifi chip vendor's firmware blob to drive the hardware, which is certified by the vendor.

replies(2): >>10738577 #>>10738715 #
28. ansible ◴[] No.10737905{5}[source]
That was an attempt at a Linux-based phone before Android. It is very old (close to 10 years?).
replies(1): >>10738208 #
29. creshal ◴[] No.10737907{5}[source]
It's a good model to limit the damage a hijacked baseband chip can do, yes. But it is still not "free".
30. chrsw ◴[] No.10737977{4}[source]
You're right, Purism doesn't offer anything no one else offers at the moment. It looks like they're trying to grow a customer base then use that base as leverage when it comes time to negotiate with OEMs over features, open docs, open firmware, etc.

Since they're not at all transparent on the details about how they will actually achieve true Freedom on modern hardware, and since modern hardware IP is deeply entangled in patent and licensing issues, it's reasonable to be high skeptical of what's going on here.

Then again it can all be a scam which would render anything I just said irrelevant anyway.

31. rtpg ◴[] No.10738208{6}[source]
Was more wondering about the accusation that the FCC shot down the effort. Google seems to show that it launched something, can't find any trace of controversy
replies(1): >>10738470 #
32. keenerd ◴[] No.10738389{4}[source]
> via a Debian chroot inside Crouton

Try a native install on the Flip, it is nice: http://kmkeen.com/c100p-tweaks/

replies(1): >>10740120 #
33. davexunit ◴[] No.10738414{3}[source]
We use Novenas as build slaves in GNU Guix's build farm and they work well. Is that not serious work?
34. creshal ◴[] No.10738470{7}[source]
The project itself didn't fail because it – that was just due to Android being more attractive by the time it was working –, but they never managed to opensource the baseband firmware for that reason.
35. lmns ◴[] No.10738577{8}[source]
>DD-WRT uses the wifi chip vendor's firmware blob to drive the hardware, which is certified by the vendor.

At least for many Atheros-based chipsets they use ath9k instead of the vendor blobs.

36. throwaway7767 ◴[] No.10738650{4}[source]
If someone were to develop an open-source replacement firmware for a baseband chip, the hardware project could use that chip but ship with the manufacturers firmware. It would then be on the users to reflash if desired. I doubt the FCC can do anything about that, people are already doing this with the TI Calypso replacement firmware.
replies(2): >>10738753 #>>10739407 #
37. throwaway7767 ◴[] No.10738715{8}[source]
> Every antenna+transmitter configuration has to be certified separately (that's why Lenovo and other laptop vendors have Wifi card whitelists and refuse booting with uncertified chips installed).

Are you sure about that? The fact that not every vendor has such a lock suggests to me that there is no legal requirement for it.

38. creshal ◴[] No.10738753{5}[source]
Oh, they cannot prevent you from installing the firmware… but if anyone catches you using it in the wild, you're in deep REDACTED.

E.g., If you're worried about the police monitoring your communications, giving them a perfectly legal reason to detain you is likely not your preferred course of action.

replies(2): >>10738951 #>>10739082 #
39. throwaway7767 ◴[] No.10738951{6}[source]
Sure, but assuming there isn't a serious bug in the baseband causing other spectrum users grief, how likely is it that someone will check your phone's baseband for tampering?

If you're a person of interest, the police can come up with a better reason to detain you than this.

40. nickpsecurity ◴[] No.10739064[source]
Open a few side windows, add one to basement, drill a tunnel to basement, balsawood for backdoor, ladder to hole in roof, and... now with brand new locks on front door. Shit, security and freedom have never been better. Sign me up!
41. jessaustin ◴[] No.10739082{6}[source]
In this case we're not really worried about police monitoring. The police aren't magical, yet. If they're trying to surveil without the target's knowledge, and the first attempt fails, they'll try something else. If no electronic surveillance works, they'll find another way to investigate, or they'll prioritize other investigations. They really have no way to catch random people with unauthorized firmware, so long as that firmware generally follows FCC guidelines.

However it may be that other, less Constitutionally-constrained parties would have the ability to dragnet for nonstandard firmware to highlight people for more intense scrutiny. The police could use a parallel construction based on that. Then they could say that unauthorized firmware on a seized phone establishes some sort of criminal intent.

42. hackuser ◴[] No.10739407{5}[source]
Here are a few possibilities I've come across:

* OsmocomBB (http://bb.osmocom.org/trac/)

* An old HN discussion: https://news.ycombinator.com/item?id=7064187

* OKL4, a hypervisor, is used widely in basebands. AFAICT It was developed by Open Kernel Labs and was open. It seems to have been acquired by General Dynamics and I don't know it's current status (does anyone know more about it?) (https://gdmissionsystems.com/cyber/products/trusted-computin...)

43. hackuser ◴[] No.10739482{5}[source]
Some follow-on and related projects:

* GTA04 by OpenPhoenux (http://projects.goldelico.com/p/gta04-main/)

* Neo900 (http://neo900.org/)

* QTMoko (http://qtmoko.sourceforge.net/)

* SHR (http://shr-project.org/)

44. wawi ◴[] No.10739568{5}[source]
The OpenMoko project sort of fizzog'ed, but you can get a good kick out of the OpenPandora (and new: OpenPyra) projects, which the GTA04/neo guys have helped along a bit, I think ..

http://openpandora.org/

45. yuhong ◴[] No.10739904[source]
And the microcode is built into the CPU so skipping microcode updates would be useless.
46. ggreer ◴[] No.10740120{5}[source]
From that page:

> The C100P Flip is the best netbook I have seen in eight years.

That's quite the praise! This thing must be really good.

> The default ALSA config was completely silent. Enabling Right Speaker Mixer Right/Left DAC fixed that. Supposedly there is a risk of burning out the amp if you thoughtlessly enable every option.

Uhh... wow. Well, OK.

> After suspend/resume, wifi will not work if the btsdio module was automatically loaded.

The best netbook in eight years, yet it can't even suspend/resume properly? What?

> The best video output mode is X11 video output. Despite what everyone says about being slow, this is the only driver that doesn't have major desync problems.

> Stellarium would run at a buttery smooth 60 FPS for a few minutes and then everything would die.

> Chromium will not run on this hardware.

> Screen rotation through xrandr doesn't work at all...

Then under "Things to Fix":

> - HMDI output. Very wonky, usually crashes X11 after a few minutes.

> - USB ethernet. The cdc_ether module will load but nothing happens.

> - Webcam. Crazy bucket of fail here. Maybe 25% of the time fswebcam can grab a single frame. Good luck with video.

> - Multitouch on the panel. No idea how to get that working.

To me, —even with the fixes and workarounds you describe— that device sounds like a nightmare to use.

replies(2): >>10740513 #>>10740651 #
47. drudru11 ◴[] No.10740311{3}[source]
They needed x86 virtualization. The cheaper ARM systems don't support that (yet).
48. david-given ◴[] No.10740464{5}[source]
I just reran it. I don't have access to the 12x Xeon now, but:

8x i7-3770K: 1123 cpu-seconds (wall-clock: 144 seconds)

4x Cortex A17: 3196 cpu-seconds (wall-clock: 963 seconds)

I just wish it had more cores, but I suspect that Rockchip are raking in money from these things, and I expect we'll get more cores next year.

The 12x Xeon was my work machine. I might have to have a word with them about it.

49. david-given ◴[] No.10740513{6}[source]
As a counter, ChromeOS will boot from cold in about five seconds. It's got a web browser, file browser, video and music player (with super-slick video quality --- no tearing!), the WiFi just works, there's built in support for mapping the caps key to ctrl, it maintains and updates itself, all the fiddly audio and touchscreen and gestures all just work.

And then if I want to use a real OS, I switch to my fullscreen Debian installation running awesome and all my xterms and it all just seamlessly interoperates. Except I don't need NetworkManager or PulseAudio or any of that nonsense because ChromeOS does it all for me. The brightness buttons work! The audio volume buttons work! Suspend and resume works! Everything works! There's even two-way clipboard support! Which works!

I've been using Linux for years, and I think this is the first time I have ever had a Linux-based laptop where all this stuff wasn't a total PITA.

replies(2): >>10740717 #>>10743276 #
50. keenerd ◴[] No.10740651{6}[source]
Well it is the best netbook I've seen in eight years. With the exception of the meager number of ports and the glossy screen, the hardware is wonderful.

Everything else comes down to software. Some of the difficulties were because I prefer to do things a difficult way. Eg, Arch is going to require more configuration and sound Just Works if you use Pulse. Having ALSA correctly autoconfigure everything except a single boolean flag is pretty good in my book. The warning about burning out hardware is just me passing on the advice from an engineer who actually works on these boards.

Suspend/resume is in fact flawless. It has never failed to come out of suspend, unlike quite a few thinkpads I've used. However linux does the wrong thing here with the bluetooth module. Trivial to fix, and it happens on other hardware too. Not the fault of the Flip.

Chromium doesn't run on Arch Linux ARM at all. Any hardware, regardless. So you can't hold that against the Flip. And I don't use Chromium, so this is a non-factor.

Almost everything else is the usual crap you have to put up with using garbage closed source video drivers and a kernel hacked out of ChromeOS. But I spend the majority of my time in the terminal, so the graphics are nothing I am concerned about. Similarly, I don't care about the webcam or multitouch, or HDMI because I don't own an HDMI monitor. I feel they are irrelevant details hardly worth mention, but they sound like dealbreakers to you.

Literally all I do is typing, reading, listening to music. I wanted fanless, all-day battery life, that weighs under a kilogram and is under $300. The Flip delivers and overall it has been a better experience than running Linux on a thinkpad. There are very few netbooks that could meet those requirements, and none that do it as elegantly as the Flip does. Therefor it is the best in my regard. If you disagree, it is up to you to name something superior according to the given criteria.

replies(1): >>10742628 #
51. keenerd ◴[] No.10740717{7}[source]
Counter counter :-)

Linux boots in 6 seconds. Caps is mapped to Mod4/super, you already have a GIANT control key. Linux does not need NM or Pulse either. And the operating system is not in the middle of being axed by Google, so while updates are not automatic I don't have to worry about them stopping.

But the two biggest points: Linux gets you an extra five hours of battery life, and lets you configure charge limits so that the non-replacable battery won't rot away in three years.

52. dandelion_lover ◴[] No.10740819{6}[source]
I am not sure whether you consider this marketing, but for me important things are (0) demonstrating the interest of customers in freedom, (1) explaining what's wrong with "ordinary" laptops to the public; increasing awareness in the media [0], and (2) working closely with QubesOS to make this system work on their laptops. Hardware switches are also very good.

And yes, it might be a scam. But might be not.

[0] (1) is being already done by FSF, but to me it looks like it's not enough.

53. ggreer ◴[] No.10742628{7}[source]
Our use-cases are similar, but if you value your time at anything reasonable, your budget should be much higher. Let me explain.

I spend about 10 hours a day on my main computer. If I upgrade every 18 months, even a $1500 machine only costs 28 cents per hour. Set aside frustration with drivers or software bugs; expensive laptops win for purely economic reasons. If a $1500 laptop makes me even 10% more productive, it's worth buying. This is the case for practically anyone who works in tech.

My current laptop is a base model 12" MacBook. It is, without a doubt, my favorite computer ever. It's small. It's light. The screen is gorgeous. It resumes before I'm done opening the lid. It has an amazing trackpad, excellent battery life, and a wonderful keyboard. With current technology, it could scarcely be improved upon. I wish there was a combination of hardware and software that could compete with it, but so far, nothing I've seen has come close.

replies(3): >>10742693 #>>10746819 #>>10746836 #
54. keenerd ◴[] No.10742693{8}[source]
Nah, for $1500 I could have five of these machines. Or two laptops and a powerful headless build box. You can do a lot better than a mere 10% gain with that sort of budget. You are throwing good money at diminishing returns.
replies(1): >>10742831 #
55. ggreer ◴[] No.10742831{9}[source]
I think you misinterpreted my point. I didn't say a nicer laptop made me 10% more productive. The real number is likely much higher than that. I gave 10% as an example where it would still be worthwhile to pay $1500 for. Considering the hourly wage of a programmer, 28 cents is a rounding error. Yes the returns are diminishing, but they're still totally worth it.

With regards to being able to buy more equipment: I do have a powerful rack-mount server in addition to my MacBook. But being able to buy five ordinary laptops for the price of an amazing one? That's not very relevant. I can only type on one keyboard at a time. If anything, more laptops would slow me down. I'd have to keep all of their software up-to-date, sync data between them, ensure their batteries were charged, etc.

Imagine making the same argument in other domains. For the price of one plane ticket, I could buy five Greyhound bus tickets to the same destination. For the price of one quality memory-foam mattress, I could buy five innerspring mattresses. For the price of one Aeron chair, I could buy five AmazonBasics office chairs. So what? I don't want more. I want better.

replies(1): >>10744251 #
56. creshal ◴[] No.10743276{7}[source]
> The brightness buttons work! The audio volume buttons work! Suspend and resume works! Everything works!

I can't remember when I last had a Thinkpad where any of that didn't work out of the box with Linux.

replies(1): >>10744619 #
57. keenerd ◴[] No.10744251{10}[source]
If you are paying for Applecare, you are paying for "more not better". A second laptop is zero (amortized) overhead: every night your real laptop rsyncs back to the spare. Battery floats at 70% charge, nothing to worry about there either. Now when you spill coffee all over your laptop, you have a hot spare ready to go. If you are accident prone you can do this "for free" up to four times. Imagine the productivity gain not having to visit a store and deal with the Geniuses.

It is irresponsible to always buy the "best" when you should be looking for the best value per dollar. Your own bank account is zero-sum, be more effective with it.

58. david-given ◴[] No.10744619{8}[source]
I have to admit that I've never used a Thinkpad. I know they have excellent Linux support (lots of people where I work use them because of this).

But on every single other laptop I have ever used Linux on, there has always been something that hasn't worked properly, whether it's audio not waking up properly after a suspend, or occasionally suspend not actually working and I discover a red hot laptop in my bag spinning at 100% CPU, or some such issue.

Having a machine which I don't have to fiddle with to make work is a totally new experience for me.

59. madez ◴[] No.10745038{3}[source]
The Chromebook C201 is not offered in my country, i.e. Germany. Any ideas where to get it from?
60. purismcomputer ◴[] No.10746819{8}[source]
Thank you for reminding everyone why quality matters. (We're flattered to be compared to beautifully designed things like Aeron chairs and the Mac.)

We think that once you actually see and touch the Librem, hands on, you will find the hardware is even better than a MacBook in many ways.

The Librem is definitely much faster, and the screen quality is amazing. The feel is very sturdy.

The usability, look/feel of the amazing Mac OSX is harder to exceed but we are working on the UI and ease of use in our Linux-kernel-based PureOS. It's all a work in progress.

And one more thing -- we are structured more like an Open Source project than a traditional corporation, so we are able to iterate very very rapidly.

The Librem is really created with the feedback of the backers and the community. This is what makes it at the core, very different than any other computer.

61. purismcomputer ◴[] No.10746836{8}[source]
But the real argument to pay more for a Librem should be:

"What is my data worth?"

What is your privacy worth?

What is it worth if even one time you have your identity stolen or tax return hacked, or your company's reputation is ruined by a data breach?