Topic for #qi-hardware is now Copyleft hardware - http://qi-hardware.com | hardware hackers join here to discuss Ben NanoNote, atben / atusb 802.15.4 wireless, and other community driven hw projects | public logging at http://en.qi-hardware.com/irclogs
guanucoluis1 has quit [Ping timeout: 245 seconds]
wolfspra1l has quit [Quit: leaving]
wolfspraul has joined #qi-hardware
woakas has quit [Ping timeout: 246 seconds]
xiangfu has joined #qi-hardware
guanucoluis has joined #qi-hardware
emeb has quit [Quit: Leaving.]
guanucoluis has quit [Ping timeout: 245 seconds]
nikescar has quit [Read error: Connection reset by peer]
nikescar has joined #qi-hardware
woakas has joined #qi-hardware
rejon has joined #qi-hardware
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined #qi-hardware
rejon has quit [Ping timeout: 244 seconds]
jekhor has joined #qi-hardware
kristoffer has joined #qi-hardware
Ayla has quit [Quit: dodo]
xiangfu has quit [Ping timeout: 244 seconds]
wolfspraul has quit [Ping timeout: 255 seconds]
wolfspraul has joined #qi-hardware
LunaVorax has joined #qi-hardware
jekhor has quit [Ping timeout: 256 seconds]
fire_ has quit [Ping timeout: 244 seconds]
LunaVorax has quit [Ping timeout: 260 seconds]
fire_ has joined #qi-hardware
paul_boddie has joined #qi-hardware
<paul_boddie>
Can't say I totally agree with rjeffries: a lot of people are buying Raspberry Pi as a plaything, and some of them are buying them up for business purposes ("cheap home media centre!1!"), lending credibility to the argument that it has largely been a "bottoms on seats" exercise by Broadcom.
<paul_boddie>
Also there may be a lot of people who just get upset about how it isn't plug-and-play and vow never to have anything to do with "difficult" Linux again. But I have to admit that the search engine results for various things are filling up with Raspberry Pi users. I would say that it's a bit like the days of AOL's rapid expansion, but that could be interpreted as being elitist.
kristoffer has quit [Quit: This computer has gone to sleep]
LunaVorax has quit [Ping timeout: 248 seconds]
kuribas has joined #qi-hardware
xiangfu has quit [Ping timeout: 245 seconds]
GNUtoo-desktop has joined #qi-hardware
xiangfu has joined #qi-hardware
fire_ is now known as nerd
cladamw has joined #qi-hardware
jluis|work has quit [Remote host closed the connection]
rejon has joined #qi-hardware
rejon has quit [Ping timeout: 240 seconds]
heberth has joined #qi-hardware
erikkugel has joined #qi-hardware
kristoffer has joined #qi-hardware
kuribas has quit [Quit: ERC Version 5.3 (IRC client for Emacs)]
woakas has quit [Ping timeout: 246 seconds]
cladamw has quit [Quit: Ex-Chat]
kristianpaul has quit [Quit: leaving]
kristianpaul has joined #qi-hardware
kristianpaul has quit [Changing host]
kristianpaul has joined #qi-hardware
xiangfu has quit [Quit: Leaving]
heberth has quit [Quit: leaving]
rejon has joined #qi-hardware
qwebirc82032 has joined #qi-hardware
qwebirc82032 is now known as rjeffries
<rjeffries>
haters gonna hate, as the saying goes. about Raspberry Pi: I do not see it as a threat on any level to the open hardware movement. That is not their objective. But I fail to see it as a coup for BRCM. The SOC in these little 'puters is not a big deal.
<whitequark>
oh, it pretty much is, broadcom found a very wise way to spend their marketing budget.
<whitequark>
which would be at least somewhat fine except for the fact that they make really crappy SoCs, even if we don't talk about openness
<rjeffries>
Allwinner [A10, A13] and Rockbox both have decent SOCs. Both are being used for inexpensive Android computers on a stick. And an Android computer can become a Linux computer not a huge deal.
<rjeffries>
Putting aside "they should have done X, Y or Z differently" the team behind the Raspberry Pi Foundation made a simple but adequate computer that is great for learning to program in Python or C or Scratch or whatever.
<rjeffries>
It exposes low level i/o (SPI I2C SPIO) in a robust way. It does not cost what the elegant FPGA-based and VERY open Milkylist hardware does, because it has different objectives.
<rjeffries>
Given what a boost RasPi is for Debian and free software, it puzzles me that some invest time dismissing something that is selling in very large quantities. How does that hurt anyone?
<rjeffries>
Reality check: for a device to offer a high level of function AND be cheap, the design will use an SOC. Nobody is forced to purchase these devices. Anyone who perfers a Milkymist or Ben Nanonote will vote with their pocketbook.
<rjeffries>
May 1,000 Debian-based flowers bloom.
<whitequark>
rjeffries: yeah, and not every SoC is as bad as Broadcom ones
<whitequark>
Samsung makes pretty nice ones, eg Exynos 42xx
<rjeffries>
whitequary (Peter) I read your analysis. It was a good piece. But I disagree. ;)
<paul_boddie>
I think it's always worthwhile to indicate where things could have been done better. It's not as if whitequark says that it absolutely has to be an FPGA, just that it would then open up a new area that you obviously aren't going to be able to access using a locked-down SoC.
<whitequark>
rjeffries: keep in mind that Samsung has made a production device (Galaxy S II) which is more open than a educational device (RPi)
<rjeffries>
whietquark teh answer is for qi hardwrae or some other entity to build the coimputer of your dreams.
<rjeffries>
whitequark I think that point is not relevant. RasPi was not intended as an excercise in open hardware. It is simply a to9ol that allows millions of people to learn programming and also experiment with physical computing in a high-level OS rather than using Arduino
<rjeffries>
if memory serves, you and others also think Arduino is a POS. <smile>
<paul_boddie>
And I think the remark about the BBC Micro hits a nerve with some of the Raspberry Pi people. They want it to be the spiritual successor, but when people who were major developers say exactly what whitequark has said about the level of openness in terms of documentation, they have neither a decent response nor any clue about who the person they're lecturing actually is.
<whitequark>
rjeffries: Arduino is significantly overrated in the sense that people are trying to put it everywhere, even where a much more efficient solution exists.
<rjeffries>
I do not think attacking RasPi serves a purpose. It exists, it is selling. There is a vibrant (!!) ecosystem forming aroudn it. All using Debian.
<whitequark>
rjeffries: if you think that my remarks are not relevant, they won't make any "harm" to RPi; otherwise, I don't understand what your comments about "attacking" are about
<rjeffries>
paul_bodie the reason RasPi is getting all the heat is it is selling in mass quantities. Could it have been done better? YES.
<whitequark>
it's not like I'm trying to sabotage production or so.
<viric>
I think it's more the joy of "I have a small computer running", similar to gentoo people "I'm compiling!".
<rjeffries>
Will someone else do somethingt better, more open? Prolly will.
<viric>
than making much of it.
<whitequark>
^ that
<viric>
All that feels like working.
<viric>
feels like doing work. And doing work, feels like doing useful work.
<paul_boddie>
rjeffries: Is your point that the benefits of open hardware don't matter if it's cheap enough? Something like, "Oh that chipset won't work with anything recent, but they only cost $30, so off to the recycling centre they go!"
<whitequark>
there is a huge gap between "it's cool" and "it's actually useful".
<rjeffries>
viric I am not sure there is anything wrong with that. This whole business remionds me of early days when I assembled an IMSAI kit computer in my college days.
<viric>
ah yes, sure it provides learning
<viric>
I see people in toolchain channels that look like discovering multiplatform building
<viric>
thanks to the pi.
<viric>
you're also supposed to write very unuseful programs before you learn to program something useful.
<rjeffries>
whitequark read the damn specs. RasPi is imperfect, has limited RAM and CPU. but it is VERY capable. twice the speed of Ben. 256MB as opposed to 32MB. Has Ethernet. Has great display hardware (that is I agree not open).
<paul_boddie>
Actually, regardless of whether Arduino is overrated or gets shoehorned into silly projects or not, I think it does expose people to stuff that they wouldn't normally do with a personal computer, and that is then the gateway to more hardcore electronics for those who are interested. It's not Arduino's fault that people use it to do weird art projects involving lots of LEDs.
<rjeffries>
If I want to use softwar eor write software, it is ok/fine.
<whitequark>
paul_boddie: oh, it's perfectly fine for _art_, just don't claim you're an engineer if you've managed to connect 256 leds with 8 arduinos or so
<rjeffries>
if one wishes to use Python, it comes ready to go out of the box.
<whitequark>
paul_boddie: similarly, don't try to claim that's the right way to do it, don't write tutorials, etc.
<rjeffries>
guys RaspI was not designed to scratch teh embedded itch. but some will us eit that way. who cares?
<whitequark>
rjeffries: like any computer existing at this moment can run Linux and Python
<whitequark>
I don't see how RPi solves any problem
<paul_boddie>
whitequark: True enough, but I think it has probably helped to get people into electronics and to see what else is available. Just giving people a platform to start experimenting with can be very useful. Most people wouldn't ever bother to start otherwise.
<rjeffries>
so why why why do you dislike this sucessfull cheap system? who gives a damn? peopel who choose to buy it will do so.
<viric>
it allows people to discover non-x86
<viric>
for cheap.
<whitequark>
rjeffries: well, I _could_ see how it is _supposed_ to solve absence of computers in poor countries, through the question of total cost of ownership and building infrastructure remains
<viric>
(as other platforms do. but if it has better marketing, it reaches more people to learn that)
<whitequark>
rjeffries: I don't see how it does any good for education in UK or even Russian schools
<rjeffries>
I do not think it was designed for that
<whitequark>
and a lot of commenters claim it does
<whitequark>
[talking about UK]
<whitequark>
you have a computer with Windows. ok, cool, there's python for that. want to run Linux without hassle? ok, there's bootable USB sticks for that
<rjeffries>
ANDF... I think we aree already seeing a wave of innovation where various peopel are designing realatively inexpensive ARM systems. I like that a lot. Cubbieboard looks pretty nice.
<whitequark>
Debian can do that for like ages.
<paul_boddie>
I think we can sum up the objections as being that the Raspberry Pi is being sold as a definitive solution for lots of things, but when you look more closely, it fails to address most of those things properly, perhaps with the exception of its price. And even then, the price per unit is not going to be what you will end up having to pay for a complete solution.
<whitequark>
^ that
<whitequark>
rjeffries: cubieboard suffers from exactly same problems
<whitequark>
allwinner is only marginally better than broadcom
<wpwrak>
what's missing with the arduino is a smooth way for developers to emancipate themselves from the arduino. it seems that many don't quite realize that just adding the chip to their circuit is only a relatively insignificant additional effort and that it gives them a lot more flexibility (besides lower cost).
jekhor has quit [Read error: Operation timed out]
<rjeffries>
whitequark so the fact that a university such as Cambridge can use a $35 compuer in a class on deigning an OS is bad somehow? LOL
<whitequark>
rjeffries: you CANNOT teach designing an OS on an RPi.
<whitequark>
that's the whole point of my article!
<whitequark>
I've seen that course.
<whitequark>
two thirds of it is hand-waving and "yeah GPU does that"
<wpwrak>
rpi .. dunno. a fad, i guess. like the sinclair zx81. you got it, you played with it, you learned some things, you grew bored with it, you moved on.
<rjeffries>
whitequark I hope you can raise money and design the opne hardware alternative to Raspberry Pi. qi hardwrae was started for jsut such a dream
<whitequark>
rjeffries: there are more open alternatives to Pi right now. Beagleboard can be perfectly used to teach Operating Systems course
<rjeffries>
whietquark your statement baffles me TOTALLY. wtf ar eyou talking about? LOL
<rjeffries>
so? so that's an extra $5o per student.
<whitequark>
then my comments apply.
<wpwrak>
what's more worrying is if the rpi is used to replace platforms better suited to the purpose. that risk exists in education. platforms are often long-lived there, so a bulk purchase of rpis can block better things for several years.
<whitequark>
rjeffries: skip one $600 textbook purchase, suit a class with proper boards.
<paul_boddie>
wpwrak: For those of us who didn't do electronics, the Arduino is probably mostly about working your way towards developing that circuit before cutting the cord. That said, most of my Arduino hacking has been based on interfacing with various embedded controllers, and for that the Arduino is rather educational and brings back the microcomputer era mindset that such work entails.
<rjeffries>
teh only solution is to convince US authorities that Raspberry Pi Foundation is a therat to national security. then send in the drones. Problem solved. LOL
<rjeffries>
RasPi is a decent environment for interfacing with snesors using SPI and I2C. aso is ASrduino, in a lower level way.
<rjeffries>
Raspberry Pi si not a problem. It is an opportunity. RNOBODY and I mean not a single soul is required to purchase a RasPi.
<wpwrak>
paul_boddie: oh, i think the arduino is a great intermediate step. but it seems that a lot of people stop there and consider it an unavoidable building block even though their projects suggest they could do much better. it's a mental barrier - people think there is some "magic" in the arduino they couldn't tackle themselves.
<whitequark>
rjeffries: can you say which exact problem RPi solves?
<rjeffries>
in any case, I have said more than enough. Move along pople, nothing to see here. Nevermidn that Ben NN used an SOC that was not open. ;)
<wpwrak>
rjeffries: in the free market, everyone competes against everyone. so of course the rpi is a threat in that sense :)
<whitequark>
rjeffries: but one can make a MIPS softcore without risking being sued
<whitequark>
and ARM won't allow you to do that even if you have an ability.
<rjeffries>
whitequark RasPi offers a LOT of computing power at a low price. It has internet connectivity, unlike Ben NN. it is a great platofrm for individuals to learn stuff they will not learn using a PC
<whitequark>
rjeffries: like python and debian?
<rjeffries>
If ARM is teh enemy, then the enemy won this war a long time ago
<paul_boddie>
wpwrak: Agreed. It's too easy to avoid building a circuit and instead continue with the software developer mindset and the luxuries of being able to fire signals on various pins from a program instead of using a circuit that would do the job more efficiently. It's a prototyping tool, I guess. Then again, you could criticise people for using FPGAs and not using discrete logic components... ;-)
<whitequark>
paul_boddie: discrete logic is directly equivalent to an FPGA, arduino isn't equivalent to discrete logic
<rjeffries>
whitequark I think interfacing with I2C SPI and GPIO on a CHEAP platform that runs Debian is uiseful and a lot of people will learn.
<whitequark>
paul_boddie: it's serial, not parallel, to say it concisely
<whitequark>
rjeffries: it's quite hard to use I2C, SPI and GPIOs on RPis without frying it immediately
<whitequark>
I've explained why
<rjeffries>
whitequark I understand the issues. and thye best point you made is that it would have been inexpensive to make the i/i interfacing safer. Point well taken.
<whitequark>
and, well, I'll let you guess how long a (school-owned) rpi would live if kids will plug and unplug peripherals in it daily
<paul_boddie>
whitequark: I was joking, mostly, but there's also a point about what the "right way" to do things is. One can tell someone off for flashing their LEDs using a program, but if they're doing a one-off art exhibit or something, are we being too harsh?
<wpwrak>
paul_boddie: that's not what i mean. i mean replacing the arduino with an avr. same architecture. same capabilities (if you choose the chip accordingly), but a lot more flexibility when if comes to integration.
<whitequark>
rjeffries: arduino is really really hard to kill
<rjeffries>
I think there will be a B RISK business in the Gertboard.
<rjeffries>
but you also displie Arduino. LOL
<rjeffries>
dislike
<rjeffries>
Arduino is also sucessfull and also uses poroprietary chip. AVR is not open.
<whitequark>
rjeffries: incorrect
<whitequark>
1) AVR has completely open specifications
<whitequark>
2) there are AVR softcores
<whitequark>
the precise chips which are in Arduino aren't open-source, but I don't see this as a real problem.
<rjeffries>
before I go on mute, consider: what is to prevent Raspberry Pi Foundation or someone else from designing a next system that meets Peter's objections?
<paul_boddie>
wpwrak: Yes, I've seen a number of Arduino alternatives and looked at the actual AVR parts, and Arduino is quite clearly a compromise that doesn't work optimally for various use-cases. It's good as a springboard for some things, but it isn't everything to everyone. I think that's what upsets people about the Raspberry Pi.
<whitequark>
rjeffries: the fact that it's basically owned by Broadcom
<rjeffries>
BRING IT ON. Make it happen, run it up the flagpole. I will salute. ;)
<wpwrak>
maybe one should make an fpgauino :) navre shouldn't need a very potent FPGA ...
<whitequark>
wpwrak: might even fit in a big CPLD
<paul_boddie>
Did any of you ever get to look at a Balloon Board?
<wpwrak>
even easier then
<whitequark>
rjeffries: there _are_ better platforms. OK, for 2x the cost (BeagleBone). I don't see this as a problem, sorry.
<rjeffries>
wpwrak I like that idea. I sthat what wolfspraul i sup to these days? it see,s his interest in Milkymist and Ben have flatlined
<rjeffries>
whietquark as you know well, Beaglebone does not have a display system. But nice try with apples/oranges. LOL
<wpwrak>
i don't think he has any specific hardware in mind at the moment. (except for a very crude experimental board)
<wpwrak>
so this is more something along the lines of basic research. first figure out how these quarks work, then build supercomputers ;-)
<paul_boddie>
I actually think a better title for whitequark's post might have been "Why the Raspberry Pi isn't the BBC Micro".
<rjeffries>
Ballonboard looks interetsing. do we have any idea about price?
<whitequark>
paul_boddie: kinda
<rjeffries>
paul_brodie as Brits might say "Spot on."
<paul_boddie>
I'm not even sure the Balloon Board is available any more. I spent some time despamming those people's Wiki as a sort of random act of kindness, but it all seemed very slow-moving and behind closed doors.
<wpwrak>
paul_boddie: balloon sounds like a worthy arduino replacement :)
<paul_boddie>
Some of the Debian Embedded people are involved in it.
* paul_boddie
goes to check their Web site.
<whitequark>
judging from the chip form factor, it's really really ancient
<whitequark>
oh wow, XScale
<wpwrak>
they write they're working on something more modern
<whitequark>
I don't think these chips are still manufactured :)
<paul_boddie>
As I said, not much going on these days.
<wpwrak>
whitequark: that may be why they are / have been trying to modernize :)
<whitequark>
I like the idea of integrating an FPGA to such a board
<paul_boddie>
I should pester some of those people and get a status report.
<paul_boddie>
whitequark: Yes, it's the same companies paying for the work.
<wpwrak>
rjeffries: (iendian) now you can complain about the ben being expensive ;-)
<paul_boddie>
There are actually a few companies who were doing ARM-related stuff in the 1990s, and some of them have managed to stay alive doing things like development boards and, I imagine, projects for clients.
<whitequark>
wpwrak: I see their point about compatibility, but I don't think this would actually prevent anyone from causing fragmentation
<whitequark>
if a company has enough traction to pull that off, they'll just ignore the license.
<wpwrak>
yeah, it's another case of punishing the honest
* kristianpaul
had been playing with solar balloons latelly
viric has quit [Read error: Connection reset by peer]
<whitequark>
wpwrak: I think that in such cases, it's enough to forbid people building incompatible variants to call their stuff with the original name
viric has joined #qi-hardware
<whitequark>
i.e. like Sun requiring certification to allow you to call your implementation of some language Java
<wpwrak>
yes, trademarks can work that way
dandon has quit [Ping timeout: 256 seconds]
<whitequark>
requires you to actually define the certification process, through, but if it's intended to be a _standard_, you'd need that anyway
<whitequark>
wpwrak: I don't see any of these claims as incorrect :)
<whitequark>
I'd say it is quite precise actually
<whitequark>
esp. given the fact that they do publish the design and so
Ayla has joined #qi-hardware
<whitequark>
and not using those as reasons to justify the closedness
<wpwrak>
if you have already most of the design plus a bit of experience, it shouldn't be nearly as intimidating. of course, component pricing (low volume) may be what would kill DIY economically.
<paul_boddie>
"The nous to identify partners who will support the project from beginning to end" - a lot of people are finding this out the hard way on their "shopping trips" to China. Take the Vivaldi tablet as an example.
<wpwrak>
s/already/already have
<qi-bot>
wpwrak meant: "if you have already have most of the design plus a bit of experience, it shouldn't be nearly as intimidating. of course, component pricing (low volume) may be what would kill DIY economically."
<wpwrak>
argh. it was correct before
<whitequark>
paul_boddie: what's with that tablet?
<wpwrak>
paul_boddie: china is a great trap for the naive ;-)
<paul_boddie>
They need to find a supplier who won't change the parts list to save 5 cents and not tell anybody, even if it means another iteration of driver development.
<paul_boddie>
I think this more or less illustrates why open hardware is important, but as long as people can get cheap stuff that almost works, they will keep pointing out the cheap stuff and ridiculing anyone wanting to do things differently.
LunaVorax has joined #qi-hardware
<whitequark>
cheap stuff with lots of strings attached
<rjeffries>
what the world needs is a $100 Linux tablet that is as open as practical. May not meet Peter's criteria, but in-built display would be a Good Thing. BT keyboard is fine. needs wi8fi. whoops, there we go, not totally open hardwrae.
<whitequark>
rjeffries: ever tried programming on a tablet?
<whitequark>
rjeffries: I believe that just won't work
<rjeffries>
whitequark a tablet with a Bluetooth keyboard? No, but how ios that different vs a lappie
<paul_boddie>
rjeffries: You can get cheap tablets now. Don't you read the arm-netbook list? When not discussing their EOMA-68 card project, which is most of the time these days because it's in super-secret factory mode, it's a parade of cheap tablet suggestions and "deal" Web site ordering experiences. And Q&A about people's recent purchases not working properly.
<rjeffries>
i will check out arm-netbook list thanks!
wpwrak has quit [Ping timeout: 252 seconds]
<whitequark>
rjeffries: 1) ARM license agreement explicitly forbids open-sourcing the design
<whitequark>
2) massively multicore designs are HARD, that was already determined in quite a powerful way by companies bigger than some random Kickstarter-funded one. I would happily be proven incorrect
<rjeffries>
that is an opportunity for Intel then, isn't it?
<whitequark>
3) Kickstarter? IC? $99? No way.
<whitequark>
rjeffries: yeah, there were some experiments in the industry
<rjeffries>
greenarrays is an interesting exception. but thou shalt speak FORTH
<whitequark>
greenarrays is completely different business
<whitequark>
it's not a von Neumann architecture
<rjeffries>
underastood. I do not expect that kickstarter project to be a success but the idea is interesting
<whitequark>
i.e. you cannot make gcc emit GA code
<whitequark>
interesting device, but practically useless
<whitequark>
rjeffries: ahhh, they don't design their own chip, they use Epiphany coprocessors
<rjeffries>
oh I was talking about open netbook designs that ARM prohibits. Intel does not have such a restriction to my knowledge
<rjeffries>
those guys have a laser focus on floating point performance is my understanding
<whitequark>
rjeffries: AFAIK ARM does not prohibit open netbook design
<rjeffries>
somebody above just said they do
<whitequark>
rjeffries: as per Parallela, this is a project I would happily back.
<whitequark>
the part about revolutionizing the industry with parallel programming is utter bullshit
<whitequark>
(because "serial" programming has its own big advantages)
<whitequark>
but the 64-core chip is just like a GPU
<rjeffries>
whitequark as in marketing BS agree.
<whitequark>
well, you cannot remake a generic algorithm to run in a parallel fashion in like 95% of cases
<whitequark>
i.e. any memory-bound one won't work this way. Quicksort to name an example.
<rjeffries>
there was an interesting hint yesterday soimewhare that people are making progress with opening up some nclosed aspects of that BRCM SOC. not sure if this is a dream or what. But given the marketing success of RasPi, I could see BRCM seeing more benefit than risk. that would be A Good Thging. Besides, the RasPOi is an older SOC. who cares?
<whitequark>
rjeffries: remember the part I said about someone remaking an ARM core and getting sued?
<whitequark>
the core in question was out of production
<whitequark>
so, lawyers care.
<rjeffries>
that core must still be be able to be fabb'd
<whitequark>
and Broadcom _already_ sees huge benefit: it has forced a lot of people to use their proprietary platform
<whitequark>
there's no reason for them to open anything. RPi is already produced.
<paul_boddie>
I've been waiting to see what might happen once someone reverse-engineers the proprietary instruction set.
<rjeffries>
forced? no it has not forced anyone. Poeple liked the price and the idea and spent $35. BFD
<whitequark>
paul_boddie: not much
<whitequark>
paul_boddie: you also need to write a complete OpenGL stack, or a half of OpenGL stack if you can reuse Mesa
<whitequark>
(or its EGL sister project)
<whitequark>
this is like 80% of work, you can join #lima and ask about that :)
<paul_boddie>
whitequark: From the lawyers and the Raspberry Pi Foundation, I mean. As in, "We promised to play nice and not try and look inside and now you're going to ruin it for everyone!"
<whitequark>
paul_boddie: there isn't any legal agreement between Foundation and its customers which would protect the latter
<paul_boddie>
But what I mean is that "Why can't you tell us how it works?" is an FAQ. It's all "we can't tell you" and "please respect the IP of our partner", and so any reverse-engineering is likely to provide much finger-pointing and wailing.
<whitequark>
so much no one seems to care about it
<paul_boddie>
I guess it won't matter until someone actually provides their own firmware, if it ever gets that far.
<whitequark>
paul_boddie: that's almost impossible
<whitequark>
well, stranger things have happened, but I don't think that it is realistic. Not in a reasonable timeframe, i.e. 2 years or so.
<paul_boddie>
Yes, I guess they'd have to provide something completely new, and that would take quite some effort. But I wouldn't put it past someone to modify the blob that gets shipped now. There are a lot of tinkerers who don't have the awareness of stuff like copyright who might make their own "mods" and that might get some reaction.
<whitequark>
paul_boddie: I guess the blob is tivoized
<paul_boddie>
I wouldn't know. Maybe rjeffries knows as he has a lot to say about the Raspberry Pi. ;-)
<whitequark>
so it's kinda simpler. Also, you'd need both very good IDA skills and knowledge of internals of the GPU, and so far most people who have those skills aren't very fond of Pi.
<whitequark>
*most people I've spoke with
<paul_boddie>
Some people like a challenge, though.
<whitequark>
well, we've seen an engineer of one company doing embedded GPUs to reverse-engineer a GPU of a rival company in his unpaid time
<whitequark>
I guess he was extremely bored
nerd has quit [Quit: WeeChat 0.3.8]
<rjeffries>
my gut says that BRCM may look the other way. If their mainstream revenu bearing SOCs are sufficently differnt than what RasPi uses, then they could be (in a certain sense) "good guys" and let teh people working on reverse engineering (it IS hapopening as we speak) go forward. Zero harm commercially.
<whitequark>
rjeffries: I doubt they will actively harm these efforts
<rjeffries>
let's assume BRCM makes a net profit of $1 aper SOC they sell to Farnell/E14 (who manufacture RasPi)
<whitequark>
RPi is not about profits from BRCM
<whitequark>
it's about marketing
<whitequark>
they might even manufacture them at a loss
<rjeffries>
if soemoen opnes the old GPU at some level, and suddenly academics have a cool tool to play with, heh, lots of PR goodness.
<whitequark>
unfortunately one does not just "open an old GPU"
<whitequark>
there isn't any RE project of an embedded GPU which can actually provide production results, i.e. something for academics to play with
<whitequark>
some of which have been around for years
<whitequark>
I hope that Lima will be the first to become so. Still, it may not happen in near future for a multitude of reasons.
<rjeffries>
what is Lima? I am not informed
<whitequark>
reverse-engineering of Mali-200/400 GPU by ARM
<rjeffries>
ok got it Lima is a free software graphics driver for the ARM Mali GPUs
LunaVorax has quit [Remote host closed the connection]
<whitequark>
yup. Mali slowly but certainly becomes a standard for mobile GPUs
<rjeffries>
by the way, something much less than a replacment GPU driver could be of interest. if one can use some GPU respources that would be ok/fine
<whitequark>
you need a GPU driver for that :)
Ayla has quit [Quit: leaving]
<GNUtoo-desktop>
rjeffries, really? I tough it was only test cases?
<rjeffries>
GNUtoo-desktop I am way way beyond my depth here. I do not know.