[9fans] Raspberry Pi 2 Model B

[9fans] Raspberry Pi 2 Model B

[Shane Morris]

[-- Attachment #1: Type: text/plain, Size: 366 bytes --]

Just saw this come up on my Facebook feed:

http://www.raspberrypi.org/raspberry-pi-2-on-sale/

Noted it *should* be backwards compatible with previous software. Does a
9fan wish to vet the 9pi release against this new hardware? I'd love to
give it a go, but I'm not in a position time wise, being made one of the
directors of two companies just recently...

[-- Attachment #2: Type: text/html, Size: 511 bytes --]

Re: [9fans] Raspberry Pi 2 Model B

[Richard Miller]

> Noted it *should* be backwards compatible with previous software. Does a
> 9fan wish to vet the 9pi release against this new hardware?

Compatible with user level software, yes. But it will need a new kernel,
because it has 4 x cortex-a7 cores compared with 1 x arm11 on the older pi.

Re: [9fans] Raspberry Pi 2 Model B

[Sean Hinchee]

The difference between Armv6 and Armv7 isn't it?

On 2/2/15 7:14 AM, Richard Miller wrote:
>> Noted it *should* be backwards compatible with previous software. Does a
>> 9fan wish to vet the 9pi release against this new hardware?
> Compatible with user level software, yes. But it will need a new kernel,
> because it has 4 x cortex-a7 cores compared with 1 x arm11 on the older pi.
>
>

Re: [9fans] Raspberry Pi 2 Model B

[Skip Tavakkolian]

[-- Attachment #1: Type: text/plain, Size: 597 bytes --]

hardkernel's odroid-c1 is similar and slightly better performance for the
same price; any sense which port might be easier?

http://www.hardkernel.com/main/products/prdt_info.php?g_code=G141578608433


On Mon Feb 02 2015 at 5:15:34 AM Richard Miller <9fans@hamnavoe.com> wrote:

> > Noted it *should* be backwards compatible with previous software. Does a
> > 9fan wish to vet the 9pi release against this new hardware?
>
> Compatible with user level software, yes. But it will need a new kernel,
> because it has 4 x cortex-a7 cores compared with 1 x arm11 on the older pi.
>
>
>

[-- Attachment #2: Type: text/html, Size: 949 bytes --]

Re: [9fans] Raspberry Pi 2 Model B

[Bakul Shah]

On Mon, 02 Feb 2015 20:54:02 GMT Skip Tavakkolian <skip.tavakkolian@gmail.com> wrote:
>
> hardkernel's odroid-c1 is similar and slightly better performance for the
> same price; any sense which port might be easier?

Odroid-c1 is Cortex-A5 while Pi2 is Cortex-A7 so Pi2 is more
performant (but has worse ethernet and horrible usb). Port to
pi2 should be easier as the periphs are the same as in Pi1;
only their IO maps have changed a bit.

> http://www.hardkernel.com/main/products/prdt_info.php?g_code=G141578608433

Re: [9fans] Raspberry Pi 2 Model B

[Joe Bowers]

[-- Attachment #1: Type: text/plain, Size: 748 bytes --]

Does anybody know where to look for documentation on the changes, or
general documentation for the bcm2836?

On Mon, Feb 2, 2015 at 1:24 PM, Bakul Shah <bakul@bitblocks.com> wrote:

> On Mon, 02 Feb 2015 20:54:02 GMT Skip Tavakkolian <
> skip.tavakkolian@gmail.com> wrote:
> >
> > hardkernel's odroid-c1 is similar and slightly better performance for the
> > same price; any sense which port might be easier?
>
> Odroid-c1 is Cortex-A5 while Pi2 is Cortex-A7 so Pi2 is more
> performant (but has worse ethernet and horrible usb). Port to
> pi2 should be easier as the periphs are the same as in Pi1;
> only their IO maps have changed a bit.
>
> >
> http://www.hardkernel.com/main/products/prdt_info.php?g_code=G141578608433
>
>

[-- Attachment #2: Type: text/html, Size: 1251 bytes --]

Re: [9fans] Raspberry Pi 2 Model B

[Steve Simon]

Re: Ether speed

I was just asking in case there was a perfmormance improvment to be had for free.

To be honest, the ether performance is not really a limiting feature of the
pi for me, if fact the cpu and ether are a fair match for each other.

A significantly faster cpu would upset this balance IMHO. This is just gut
feelings from using my Pi quite a bit as a terminal and as an internet radio.

I do wonder if they have got it wrong with the Pi-2 and should have gone with
Gig Ether - Time will tell.

The Odroid-c1 looks nice but so does the Beagle-bone black and the
Hummingboard, there seem to be quite a few these days.

-Steve

Re: [9fans] Raspberry Pi 2 Model B

[Sean Hinchee]

[-- Attachment #1: Type: text/plain, Size: 1049 bytes --]

Some documentation can be found here:
http://elinux.org/RPi_Documentation#Raspberry_Pi_Processor_Broadcom_System-On-Chip:


On 2/2/15 3:37 PM, Joe Bowers wrote:
> Does anybody know where to look for documentation on the changes, or
> general documentation for the bcm2836?
>
> On Mon, Feb 2, 2015 at 1:24 PM, Bakul Shah <bakul@bitblocks.com
> <mailto:bakul@bitblocks.com>> wrote:
>
>     On Mon, 02 Feb 2015 20:54:02 GMT Skip Tavakkolian
>     <skip.tavakkolian@gmail.com <mailto:skip.tavakkolian@gmail.com>>
>     wrote:
>     >
>     > hardkernel's odroid-c1 is similar and slightly better
>     performance for the
>     > same price; any sense which port might be easier?
>
>     Odroid-c1 is Cortex-A5 while Pi2 is Cortex-A7 so Pi2 is more
>     performant (but has worse ethernet and horrible usb). Port to
>     pi2 should be easier as the periphs are the same as in Pi1;
>     only their IO maps have changed a bit.
>
>     >
>     http://www.hardkernel.com/main/products/prdt_info.php?g_code=G141578608433
>
>


[-- Attachment #2: Type: text/html, Size: 2384 bytes --]

Re: [9fans] Raspberry Pi 2 Model B

[erik quanstrom]

On Mon Feb  2 13:20:08 PST 2015, bakul@bitblocks.com wrote:
> On Mon, 02 Feb 2015 20:54:02 GMT Skip Tavakkolian <skip.tavakkolian@gmail.com> wrote:
> >
> > hardkernel's odroid-c1 is similar and slightly better performance for the
> > same price; any sense which port might be easier?
>
> Odroid-c1 is Cortex-A5 while Pi2 is Cortex-A7 so Pi2 is more
> performant (but has worse ethernet and horrible usb). Port to
> pi2 should be easier as the periphs are the same as in Pi1;
> only their IO maps have changed a bit.

hmm.  the arch is just part of the story.  the odroid is 1500 MHz vs
the pi2's 900 MHz, and that's ddr3 vs ddr2.  so it's not quite all
stacked in one direction.

- erik

Re: [9fans] Raspberry Pi 2 Model B

[Bakul Shah]

On Mon, 02 Feb 2015 19:52:08 PST erik quanstrom <quanstro@quanstro.net> wrote:
> On Mon Feb  2 13:20:08 PST 2015, bakul@bitblocks.com wrote:
> > On Mon, 02 Feb 2015 20:54:02 GMT Skip Tavakkolian <skip.tavakkolian@gmail.c
> om> wrote:
> > >
> > > hardkernel's odroid-c1 is similar and slightly better performance for the
> > > same price; any sense which port might be easier?
> >
> > Odroid-c1 is Cortex-A5 while Pi2 is Cortex-A7 so Pi2 is more
> > performant (but has worse ethernet and horrible usb). Port to
> > pi2 should be easier as the periphs are the same as in Pi1;
> > only their IO maps have changed a bit.
>
> hmm.  the arch is just part of the story.  the odroid is 1500 MHz vs
> the pi2's 900 MHz, and that's ddr3 vs ddr2.  so it's not quite all
> stacked in one direction.

Indeed.

Feature comparison here:
http://www.cnx-software.com/2015/02/02/raspberry-pi-2-odroid-c1-development-boards-comparison/

Will post some benchmarks comparing the two (under linux) once
I get a pi2, hopefully by this Wednesday. [How did they miss
calling it 2pi or tau?]

The way I see it, the RPi Foundation knew they had painted
themselves in a corner with the 2835. Any new design would've
been hugely disruptive.  By just replacing the processor core
and leaving rest of the cruft + GPU exactly as before they
managed to get the 2836 out in a relatively short time and all
the hardware addon will continue to work. Now that they've
escaped, I expect to see more designs.

Re: [9fans] Raspberry Pi 2 Model B

[hiro]

> To be honest, the ether performance is not really a limiting feature of the
> pi for me, if fact the cpu and ether are a fair match for each other.

This might be because the ethernet adapter uses USB which is taxing
the CPU a lot.

Re: [9fans] Raspberry Pi 2 Model B

[Peter Hull]

I struggle a bit with ARM version numbers ... is the Pi2 the same
architecture as Beaglebone Black - so they could share kernel source
(I know peripherals would be different)
Pete

Re: [9fans] Raspberry Pi 2 Model B

[hiro]

> I do wonder if they have got it wrong with the Pi-2 and should have gone
> with
> Gig Ether - Time will tell.

They did not, one of the most common usages of the pi is to decode
h.264 video, which works just fine over the slow ethernet and even via
usb wireless adapters.
Fine means the videos break more often (cant be played due to wrong,
too high encoding) than the network (adapter) in use.

The main advantage of the Pi-2 for the masses will be that they can
finally decode DTS on the pi without having to reencode and remux the
audio on an intel beforehand. Power usage is also just as low as
before (here in europe we have to pay for electricity).

kirkwood had gigabit ethernet way before the rpi even existed and I'm
regularly using multiples of 100Mbit with samba on one, it's super
stable *and* faster which is why I would never pick the rpi for a NAS.
otoh this device doesn't even have video out.

Re: [9fans] Raspberry Pi 2 Model B

[hiro]

btw if you really are that bored:
please port 9front to one of those wifi SD cards.

Re: [9fans] Raspberry Pi 2 Model B

[Costin Chirvasuta]

> kirkwood had gigabit ethernet way before the rpi even existed and I'm
> regularly using multiples of 100Mbit with samba on one, it's super
> stable *and* faster which is why I would never pick the rpi for a NAS.
> otoh this device doesn't even have video out.

What kirkwood device are you using for a NAS?

Re: [9fans] Raspberry Pi 2 Model B

[hiro]

Dockstar and only via USB 2.0
Nowadays i'd buy a pogoplug, they were still available on amazon or
ebay for 10$ last time I looked.
I'm also using the dockstar for routing. The gigabit is stable and
gives enough headroom for my pity 10Mbit uplink, whereas a rpi would
break down bec. of USB hardware (or software?) failures.

On 2/3/15, Costin Chirvasuta <costinc@gmail.com> wrote:
>> kirkwood had gigabit ethernet way before the rpi even existed and I'm
>> regularly using multiples of 100Mbit with samba on one, it's super
>> stable *and* faster which is why I would never pick the rpi for a NAS.
>> otoh this device doesn't even have video out.
>
> What kirkwood device are you using for a NAS?
>
>

Re: [9fans] Raspberry Pi 2 Model B

[Dante]

Hi all,

Does Plan9 have support for multi-core processors?
Is explicit support needed at all (like in SMP)?

D

Re: [9fans] Raspberry Pi 2 Model B

[Shane Morris]

[-- Attachment #1: Type: text/plain, Size: 259 bytes --]

Doesn't Plan 9 run on the dual core ARM Tegra2?

On Thu, Feb 5, 2015 at 7:17 PM, Dante <subscriptions@posteo.eu> wrote:

> Hi all,
>
> Does Plan9 have support for multi-core processors?
> Is explicit support needed at all (like in SMP)?
>
> D
>
>

[-- Attachment #2: Type: text/html, Size: 628 bytes --]

Re: [9fans] Raspberry Pi 2 Model B

[David du Colombier]

As far I know, only the pc, teg2 and k10 kernels support multiprocessing.

--
David du Colombier

Re: [9fans] Raspberry Pi 2 Model B

[cinap_lenrek]

and pc64 and zynq :-)

--
cinap

Re: [9fans] Raspberry Pi 2 Model B

[Siarhei Zirukin]

He didn't ask about 9front.

On 2/5/15, cinap_lenrek@felloff.net <cinap_lenrek@felloff.net> wrote:
> and pc64 and zynq :-)
>
> --
> cinap
>
>

Re: [9fans] Raspberry Pi 2 Model B

[Steve Simon]

I am fairly sure there was multicore support in the MIPS
kernels for the big challange machines they had at the labs.

-Steve

Re: [9fans] Raspberry Pi 2 Model B

[Steven Stallion]

[-- Attachment #1: Type: text/plain, Size: 167 bytes --]

On Thu, Feb 5, 2015 at 8:46 AM, Siarhei Zirukin <ftrvxmtrx@gmail.com> wrote:

> He didn't ask about 9front.
>

What's 9front?

(Apologies, couldn't resist...)

[-- Attachment #2: Type: text/html, Size: 506 bytes --]

Re: [9fans] Raspberry Pi 2 Model B

[Kurt H Maier]

Quoting Steven Stallion <sstallion@gmail.com>:

> What's 9front?

Medium- to low-quality mushroom cloud computing project designed to
vandalize treasured artwork.

Every interaction I've had with the dev team has been awful.

khm

Re: [9fans] Raspberry Pi 2 Model B

[David du Colombier]

> I am fairly sure there was multicore support in the MIPS
> kernels for the big challange machines they had at the labs.

I was only listing the kernels still present in Plan 9.
The vt5 and ch kernels (and probably others) had
multiprocessing support as well.

--
David du Colombier

Re: [9fans] Raspberry Pi 2 Model B

[Dante]

Don't be evil :-)

On 05.02.2015 16:40, Steven Stallion wrote:
> On Thu, Feb 5, 2015 at 8:46 AM, Siarhei Zirukin <ftrvxmtrx@gmail.com>
> wrote:
>
>> He didn't ask about 9front.
>
> What's 9front?
>
> (Apologies, couldn't resist...)

Re: [9fans] Raspberry Pi 2 Model B

[cinap_lenrek]

labs loaddevdescr() reads 255+18 bytes device descriptor
instead of 18 bytes for some reason. try changeing the
following line in /sys/src/cmd/usb/lib/dev.c in loaddevdescr():

	uchar buf[Ddevlen+255];

to:

	uchar buf[Ddevlen];

and see if you still get the short descriptor warning.

[REDACTED] did this change in [REDACTED] fixing problem with some
usb ethernet device.

--
cinap

Re: [9fans] Raspberry Pi 2 Model B

[Charles Forsyth]

[-- Attachment #1: Type: text/plain, Size: 1019 bytes --]

Basically, Plan 9 (kernel and applications) was designed and written for
multiprocessors,
and the kernel is written with pre-emptive concurrency in mind (rather
than, say, retrofitting
it all). That extends to the drivers and most platform-specific kernel code
(except where someone slipped up,
which is rare). Even on things that are currently uniprocessor, the
discipline is to write the
mutual exclusion code as required. Cache control is less well-developed,
since most platforms so far have offered some adequate form of coherency,
but
explicit cache flushing and invalidation is missing from some x86 drivers,
because
the architecture did the work, so that DOS would run. In practice, most
embedded platforms
have had custom SoC devices, or different devices from x86 at any rate,
so the cache flushing was included when a new driver
was written (once we understood the problem). Unfortunately, it was done
using different primitives, or at least primitive names, for different
architectures.

[-- Attachment #2: Type: text/html, Size: 1154 bytes --]

Re: [9fans] Raspberry Pi 2 Model B

[erik quanstrom]

On Thu Feb  5 09:28:09 PST 2015, cinap_lenrek@felloff.net wrote:
> labs loaddevdescr() reads 255+18 bytes device descriptor
> instead of 18 bytes for some reason. try changeing the
> following line in /sys/src/cmd/usb/lib/dev.c in loaddevdescr():
>
> 	uchar buf[Ddevlen+255];
>
> to:
>
> 	uchar buf[Ddevlen];
>
> and see if you still get the short descriptor warning.
>
> [REDACTED] did this change in [REDACTED] fixing problem with some
> usb ethernet device.

that sounds about right.  but there are many other gaffes like this.

- erik

Re: [9fans] Raspberry Pi 2 Model B

[Shane Morris]

[-- Attachment #1: Type: text/plain, Size: 736 bytes --]

Did someone say there was a Zynq kernel?

On Fri, Feb 6, 2015 at 6:15 AM, erik quanstrom <quanstro@quanstro.net>
wrote:

> On Thu Feb  5 09:28:09 PST 2015, cinap_lenrek@felloff.net wrote:
> > labs loaddevdescr() reads 255+18 bytes device descriptor
> > instead of 18 bytes for some reason. try changeing the
> > following line in /sys/src/cmd/usb/lib/dev.c in loaddevdescr():
> >
> >       uchar buf[Ddevlen+255];
> >
> > to:
> >
> >       uchar buf[Ddevlen];
> >
> > and see if you still get the short descriptor warning.
> >
> > [REDACTED] did this change in [REDACTED] fixing problem with some
> > usb ethernet device.
>
> that sounds about right.  but there are many other gaffes like this.
>
> - erik
>
>

[-- Attachment #2: Type: text/html, Size: 1270 bytes --]

Re: [9fans] Raspberry Pi 2 Model B

[erik quanstrom]

On Thu Feb  5 11:20:53 PST 2015, 9fans@hamnavoe.com wrote:
> > try changeing the
> > following line in /sys/src/cmd/usb/lib/dev.c in loaddevdescr():
> >
> > 	uchar buf[Ddevlen+255];
> >
> > to:
> >
> > 	uchar buf[Ddevlen];
>
> See /n/sources/patch/maybe/usb-short-desc

i should have remembered my own patch ...

- erik

Re: [9fans] Raspberry Pi 2 Model B

[Richard Miller]

> try changeing the
> following line in /sys/src/cmd/usb/lib/dev.c in loaddevdescr():
>
> 	uchar buf[Ddevlen+255];
>
> to:
>
> 	uchar buf[Ddevlen];

See /n/sources/patch/maybe/usb-short-desc

Re: [9fans] Raspberry Pi 2 Model B

[cinap_lenrek]

yeah, maybe. i dont know if this is the problem. just what i'd try
first. :)

--
cinap

Re: [9fans] Raspberry Pi 2 Model B

[cinap_lenrek]

aiju build a computer arround the xilinx zynq-7000 (dualcore arm
cortex A9 with fpga):

http://aiju.de/electronics/aijuboard/

and wrote kernel and bootloader here:

http://code.google.com/p/plan9front/source/browse/sys/src/9/zynq/
http://code.google.com/p/plan9front/source/browse/sys/src/boot/zynq/

the kernel supports multiprocessing (with all the caches enabled),
the fpga, gigabit ethernet and displayport. boots over tftp.

http://img.stanleylieber.com/?tags=aijuboard

things todo: usb, sata, sdcard.

--
cinap

Re: [9fans] Raspberry Pi 2 Model B

[Shane Morris]

[-- Attachment #1: Type: text/plain, Size: 936 bytes --]

Champion.

I have a Parallella, with a 7010, and a Epiphany-III. Was wondering if I'd
pull the finger out and port Plan 9, or even pay for it. Seems I might not
have to. There is also a SnowLEO SDR unit that uses a 7010 and a LimeMicro
chip (think BladeRF).

Thank you! Great news!

On Fri, Feb 6, 2015 at 6:42 AM, <cinap_lenrek@felloff.net> wrote:

> aiju build a computer arround the xilinx zynq-7000 (dualcore arm
> cortex A9 with fpga):
>
> http://aiju.de/electronics/aijuboard/
>
> and wrote kernel and bootloader here:
>
> http://code.google.com/p/plan9front/source/browse/sys/src/9/zynq/
> http://code.google.com/p/plan9front/source/browse/sys/src/boot/zynq/
>
> the kernel supports multiprocessing (with all the caches enabled),
> the fpga, gigabit ethernet and displayport. boots over tftp.
>
> http://img.stanleylieber.com/?tags=aijuboard
>
> things todo: usb, sata, sdcard.
>
> --
> cinap
>
>

[-- Attachment #2: Type: text/html, Size: 1660 bytes --]

Re: [9fans] Raspberry Pi 2 Model B

[Quintile]

ooo! I think we are about to start using zinq's in a new project... 

I have gotta try it 😄

-Steve





> On 5 Feb 2015, at 19:42, cinap_lenrek@felloff.net wrote:
> 
> aiju build a computer arround the xilinx zynq-7000 (dualcore arm
> cortex A9 with fpga):
> 
> http://aiju.de/electronics/aijuboard/
> 
> and wrote kernel and bootloader here:
> 
> http://code.google.com/p/plan9front/source/browse/sys/src/9/zynq/
> http://code.google.com/p/plan9front/source/browse/sys/src/boot/zynq/
> 
> the kernel supports multiprocessing (with all the caches enabled),
> the fpga, gigabit ethernet and displayport. boots over tftp.
> 
> http://img.stanleylieber.com/?tags=aijuboard
> 
> things todo: usb, sata, sdcard.
> 
> --
> cinap

Re: [9fans] Raspberry Pi 2 Model B

[Shane Morris]

[-- Attachment #1: Type: text/plain, Size: 896 bytes --]

Hells yes!

On Fri, Feb 6, 2015 at 8:17 AM, Quintile <steve@quintile.net> wrote:

> ooo! I think we are about to start using zinq's in a new project...
>
> I have gotta try it 😄
>
> -Steve
>
>
>
>
>
> > On 5 Feb 2015, at 19:42, cinap_lenrek@felloff.net wrote:
> >
> > aiju build a computer arround the xilinx zynq-7000 (dualcore arm
> > cortex A9 with fpga):
> >
> > http://aiju.de/electronics/aijuboard/
> >
> > and wrote kernel and bootloader here:
> >
> > http://code.google.com/p/plan9front/source/browse/sys/src/9/zynq/
> > http://code.google.com/p/plan9front/source/browse/sys/src/boot/zynq/
> >
> > the kernel supports multiprocessing (with all the caches enabled),
> > the fpga, gigabit ethernet and displayport. boots over tftp.
> >
> > http://img.stanleylieber.com/?tags=aijuboard
> >
> > things todo: usb, sata, sdcard.
> >
> > --
> > cinap
>
>

[-- Attachment #2: Type: text/html, Size: 1810 bytes --]

Re: [9fans] Raspberry Pi 2 Model B

[erik quanstrom]

On Thu Feb  5 11:37:56 PST 2015, cinap_lenrek@felloff.net wrote:
> aiju build a computer arround the xilinx zynq-7000 (dualcore arm
> cortex A9 with fpga):
>
> http://aiju.de/electronics/aijuboard/
>
> and wrote kernel and bootloader here:
>
> http://code.google.com/p/plan9front/source/browse/sys/src/9/zynq/
> http://code.google.com/p/plan9front/source/browse/sys/src/boot/zynq/
>
> the kernel supports multiprocessing (with all the caches enabled),
> the fpga, gigabit ethernet and displayport. boots over tftp.
>
> http://img.stanleylieber.com/?tags=aijuboard
>
> things todo: usb, sata, sdcard.

does anyone have a picture of this board?

- erik

Re: [9fans] Raspberry Pi 2 Model B

[lucio]

> does anyone have a picture of this board?

That sounds like the challenge Olimex <olimex.com> used to rise to,
back when they started as a prototyping and production board
manufacturers.  I don't see that that is their focus still, but they
surely have not lost their skills.

It would be nice to have a community driven project in this case.
Maybe we can at least investigate?

Lucio.

Re: [9fans] Raspberry Pi 2 Model B

[Kurt H Maier]

Quoting erik quanstrom <quanstro@quanstro.net>:

>
> does anyone have a picture of this board?
>

If you're asking for technical purposes, there's a render at
http://9front.org/img/aijuboard.png but I'm not sure that's the
final cut.

If you're asking for physical proof of its existence, one of the
guys who owns one is going to have to feel the need to prove
something.

khm

Re: [9fans] Raspberry Pi 2 Model B

[erik quanstrom]

On Thu Feb  5 20:38:05 PST 2015, khm@sciops.net wrote:
> Quoting erik quanstrom <quanstro@quanstro.net>:
>
> >
> > does anyone have a picture of this board?
> >
>
> If you're asking for technical purposes, there's a render at
> http://9front.org/img/aijuboard.png but I'm not sure that's the
> final cut.
>
> If you're asking for physical proof of its existence, one of the
> guys who owns one is going to have to feel the need to prove
> something.

i was not thinking along those lines.  :-(  i just wanted to see a picture of
the board.  the render is nice, but the toy budget is not so large.

- erik

Re: [9fans] Raspberry Pi 2 Model B

[Kurt H Maier]

Quoting erik quanstrom <quanstro@quanstro.net>:

>
> i was not thinking along those lines.  :-(  i just wanted to see a picture of
> the board.  the render is nice, but the toy budget is not so large.
>

I did spend a bit of time trowling through IRC logs; I could have sworn
aiju had posted a photo of a populated (and running) board. I wasn't able
to dig up the link.  I did find a couple screenshots of 9front running on
it -- and of course we wouldn't have the Zynq support code without it! --
but it was a lot of fun hearing aiju talk about different speedbumps he
hit while working on the design.  And of course it's a really impressive
feat, but I'm also firmly in the "can't afford this" camp, sadly.

khm

Re: [9fans] Raspberry Pi 2 Model B

[erik quanstrom]

On Thu Feb  5 21:07:18 PST 2015, khm@sciops.net wrote:
> Quoting erik quanstrom <quanstro@quanstro.net>:
>
> >
> > i was not thinking along those lines.  :-(  i just wanted to see a picture of
> > the board.  the render is nice, but the toy budget is not so large.
> >
>
> I did spend a bit of time trowling through IRC logs; I could have sworn
> aiju had posted a photo of a populated (and running) board. I wasn't able
> to dig up the link.  I did find a couple screenshots of 9front running on
> it -- and of course we wouldn't have the Zynq support code without it! --
> but it was a lot of fun hearing aiju talk about different speedbumps he
> hit while working on the design.  And of course it's a really impressive
> feat, but I'm also firmly in the "can't afford this" camp, sadly.

yes, quite impressive.  i did find one screenshot.  i'm not sure if i could
fit something like this in the toy budget or not.  i'm really curious how
the keyboard and mouse work with no usb.

- erik

Re: [9fans] Raspberry Pi 2 Model B

[Steve Simon]

See /n/sources/patch/maybe/usb-short-desc

That works a treat,
thanks Cinap, Richard, and Erik.

-Steve

Re: [9fans] Raspberry Pi 2 Model B

[cinap_lenrek]

http://felloff.net/usr/cinap_lenrek/aijuboard/

--
cinap

Re: [9fans] Raspberry Pi 2 Model B

[Richard Miller]

> http://felloff.net/usr/cinap_lenrek/aijuboard/

http://i1.kym-cdn.com/photos/images/original/000/014/416/X2R4UMUE536Z3WIALA7SKNVKP4RXX22Y.jpeg

Re: [9fans] Raspberry Pi 2 Model B

[erik quanstrom]

On Fri Feb  6 03:56:17 PST 2015, cinap_lenrek@felloff.net wrote:
> http://felloff.net/usr/cinap_lenrek/aijuboard/

nice.  thank you!

- erik

Re: [9fans] Raspberry Pi 2 Model B

[erik quanstrom]

On Fri Feb  6 03:51:08 PST 2015, steve@quintile.net wrote:
> See /n/sources/patch/maybe/usb-short-desc
> 
> That works a treat,
> thanks Cinap, Richard, and Erik.

great.  the 9atom version of usb improves on the original by
adjusting a few constants to the standard.  but the main fix is
several places where a devices could return a lot of data, asking
only for enough return data to get the true size, then asking for
the exact amount of data the device claims.

i've also reworked the pc scancode interface which was originally
used with a copy of the scancode data for every arch into one
that's shared by all using /dev/kbin, which now includes the
actual pc keyboard.  many lines of redundant code bit the dust.

the reason things are still scan code based and not usb code based
is that /dev/kbmap uses scan code translation directly.  it would
be nice to have the keyboard map say

CapsLock -> LeftCtlr
Shift+Function12 -> '☺

instead of dealing with scan codes, but it may be more trouble
than it's worth.

(the 9atom usb/kb should work fine with older kernels.)

- erik

Re: [9fans] Raspberry Pi 2 Model B

[Richard Miller]

To return to the original subject ...

/n/sources/contrib/miller/9pi2 is a Plan 9 kernel which runs on the
Raspberry Pi 2 (one core only, so far).  I'll put updated source
in contrib/miller/9/bcm later today.  mk CONF'='pi2 for the new
model, CONF'='pi for the original model.

	Plan 9 from Bell Labs
	firmware: rev 1422642103
	cpu0: 900MHz ARM Cortex-A7 r0p5
	fp: 32 registers,  simd
	fp: arm arch VFPv3+ with common VFP subarch v2; rev 5
	eMMC external clock 250 Mhz
	#u/usb/ep1.0: dwcotg: port 0X0 irq 9
	992M memory: 200M kernel data, 792M user, 3762M swap
	usb/hub... usb/ether...
	etherusb smsc: b827eb4f2fbd
	usb/kb... usb/kb... root is from (local, tcp)[local]: tcp

Even with one core activated, the rpi2 is noticeably quicker than
the rpi.  Decoding a 1600x1200 jpeg with 'jpg -t' (from ramfs)
takes about 5.8s on rpi, 3.2s on rpi2.

Note that the publicity says 900Mhz, but the firmware boots at
600Mhz, and relies on dynamic clock and voltage management in
linux to adjust the speed.  To get a fixed 900Mhz speed, I put
this in config.txt:

	kernel=9pi2
	gpu_mem=16
	disable_overscan=1
	arm_freq=900
	force_turbo=1

Disclaimer: if you put silly numbers in arm_freq, bad stuff
might happen.  Supposedly the firmware detects this and sets
an irreversible bit somewhere that voids your warranty.

Re: [9fans] Raspberry Pi 2 Model B

[Joe Bowers]

[-- Attachment #1: Type: text/plain, Size: 1659 bytes --]

FANTASTIC! Thank you so much! That's so cool!

On Fri, Feb 6, 2015 at 10:13 AM, Richard Miller <9fans@hamnavoe.com> wrote:

> To return to the original subject ...
>
> /n/sources/contrib/miller/9pi2 is a Plan 9 kernel which runs on the
> Raspberry Pi 2 (one core only, so far).  I'll put updated source
> in contrib/miller/9/bcm later today.  mk CONF'='pi2 for the new
> model, CONF'='pi for the original model.
>
>         Plan 9 from Bell Labs
>         firmware: rev 1422642103
>         cpu0: 900MHz ARM Cortex-A7 r0p5
>         fp: 32 registers,  simd
>         fp: arm arch VFPv3+ with common VFP subarch v2; rev 5
>         eMMC external clock 250 Mhz
>         #u/usb/ep1.0: dwcotg: port 0X0 irq 9
>         992M memory: 200M kernel data, 792M user, 3762M swap
>         usb/hub... usb/ether...
>         etherusb smsc: b827eb4f2fbd
>         usb/kb... usb/kb... root is from (local, tcp)[local]: tcp
>
> Even with one core activated, the rpi2 is noticeably quicker than
> the rpi.  Decoding a 1600x1200 jpeg with 'jpg -t' (from ramfs)
> takes about 5.8s on rpi, 3.2s on rpi2.
>
> Note that the publicity says 900Mhz, but the firmware boots at
> 600Mhz, and relies on dynamic clock and voltage management in
> linux to adjust the speed.  To get a fixed 900Mhz speed, I put
> this in config.txt:
>
>         kernel=9pi2
>         gpu_mem=16
>         disable_overscan=1
>         arm_freq=900
>         force_turbo=1
>
> Disclaimer: if you put silly numbers in arm_freq, bad stuff
> might happen.  Supposedly the firmware detects this and sets
> an irreversible bit somewhere that voids your warranty.
>
>
>

[-- Attachment #2: Type: text/html, Size: 2133 bytes --]

Re: [9fans] Raspberry Pi 2 Model B

[Dante]

Thanks a lot, Richard! That's awesome!!!

Kind Regards,
Dante

On 06.02.2015 19:13, Richard Miller wrote:
> To return to the original subject ...
>
> /n/sources/contrib/miller/9pi2 is a Plan 9 kernel which runs on the
> Raspberry Pi 2 (one core only, so far).  I'll put updated source
> in contrib/miller/9/bcm later today.  mk CONF'='pi2 for the new
> model, CONF'='pi for the original model.
>
> 	Plan 9 from Bell Labs
> 	firmware: rev 1422642103
> 	cpu0: 900MHz ARM Cortex-A7 r0p5
> 	fp: 32 registers,  simd
> 	fp: arm arch VFPv3+ with common VFP subarch v2; rev 5
> 	eMMC external clock 250 Mhz
> 	#u/usb/ep1.0: dwcotg: port 0X0 irq 9
> 	992M memory: 200M kernel data, 792M user, 3762M swap
> 	usb/hub... usb/ether...
> 	etherusb smsc: b827eb4f2fbd
> 	usb/kb... usb/kb... root is from (local, tcp)[local]: tcp
>
> Even with one core activated, the rpi2 is noticeably quicker than
> the rpi.  Decoding a 1600x1200 jpeg with 'jpg -t' (from ramfs)
> takes about 5.8s on rpi, 3.2s on rpi2.
>
> Note that the publicity says 900Mhz, but the firmware boots at
> 600Mhz, and relies on dynamic clock and voltage management in
> linux to adjust the speed.  To get a fixed 900Mhz speed, I put
> this in config.txt:
>
> 	kernel=9pi2
> 	gpu_mem=16
> 	disable_overscan=1
> 	arm_freq=900
> 	force_turbo=1
>
> Disclaimer: if you put silly numbers in arm_freq, bad stuff
> might happen.  Supposedly the firmware detects this and sets
> an irreversible bit somewhere that voids your warranty.

Re: [9fans] Raspberry Pi 2 Model B

[Richard Miller]

I said:

> Disclaimer: if you put silly numbers in arm_freq, bad stuff
> might happen.  Supposedly the firmware detects this and sets
> an irreversible bit somewhere that voids your warranty.

There are claims in the raspberry pi forums that the 'void
warranty' bit will be set if you configure force_turbo=1,
whatever the frequency.  So if you care about this, you
may be stuck at 600Mhz unless someone cares enough to
reverse engineer the linux speed management code.

Re: [9fans] Raspberry Pi 2 Model B

[Bakul Shah]

On Sat, 07 Feb 2015 16:50:24 GMT Richard Miller <9fans@hamnavoe.com> wrote:
> I said:
>
> > Disclaimer: if you put silly numbers in arm_freq, bad stuff
> > might happen.  Supposedly the firmware detects this and sets
> > an irreversible bit somewhere that voids your warranty.
>
> There are claims in the raspberry pi forums that the 'void
> warranty' bit will be set if you configure force_turbo=1,
> whatever the frequency.  So if you care about this, you
> may be stuck at 600Mhz unless someone cares enough to
> reverse engineer the linux speed management code.

Hot off the press (from the RPi forum):

  mpnico wrote:
    I thought that force_turbo with no over_voltage doesn't void
    the warranty. Is this not correct for the RPi2 ?


  You are correct, but Pi2 runs at a higher voltage by default
  which meant that force_turbo was incorrectly setting the
  "warranty" bit. That was unintended.

  So, what we've agreed to do is to no longer treat bit 24 of
  the board revision as the warranty warranty bit on Pi 2.  The
  latest rpi-update will now set bit 25 when a real warranty
  condition arises (e.g. force_turbo=1 *and* over_voltage > 0)

  If bit 24 is currently set, then don't worry, it will be
  ignored. Bit 25 is the new Pi 2 warranty bit.

  Pi1 will remain as it did with bit 24 being the warranty bit.

  dom (Raspberry Pi Engineer & Forum Moderator)

Re: [9fans] Raspberry Pi 2 Model B

[hiro]

Do the RPI2s break very fast or why is the warranty such an issue?

Re: [9fans] Raspberry Pi 2 Model B

[Bakul Shah]

On Sun, 08 Feb 2015 02:10:25 +0100 hiro <23hiro@gmail.com> wrote:
> Do the RPI2s break very fast or why is the warranty such an issue?

A chip's rated clock rate is typically much lower that the max
freq it at which it can run stably -- and there is fair bit of
variation in this max freq.  A rare few 2836s can even run at
double the default frequency.  But as a side effect of
speeding it up and/or increasing internal voltage it will also
run much hotter & if you don't use a heat sink, it is likely
to fail much sooner.  They catch such use by setting an
irreversible bit inside the chip. RPF's warranty is valid only
if you left these parameters at their default value or changed
them as per their instructions (used raspi-config under linux
or use the setting n config.txt as per what raspi-config
does).

Re: [9fans] Raspberry Pi 2 Model B

[Kurt H Maier]

Quoting hiro <23hiro@gmail.com>:

> Do the RPI2s break very fast or why is the warranty such an issue?

Apparently you can crash one with a light bulb:

http://www.raspberrypi.org/forums/viewtopic.php?f=28&t=99042


khm

Re: [9fans] Raspberry Pi 2 Model B

[erik quanstrom]

On Sat Feb  7 18:18:29 PST 2015, bakul@bitblocks.com wrote:
> On Sun, 08 Feb 2015 02:10:25 +0100 hiro <23hiro@gmail.com> wrote:
> > Do the RPI2s break very fast or why is the warranty such an issue?
>
> A chip's rated clock rate is typically much lower that the max
> freq it at which it can run stably -- and there is fair bit of
> variation in this max freq.  A rare few 2836s can even run at
> double the default frequency.  But as a side effect of
> speeding it up and/or increasing internal voltage it will also
> run much hotter & if you don't use a heat sink, it is likely
> to fail much sooner.  They catch such use by setting an
> irreversible bit inside the chip. RPF's warranty is valid only
> if you left these parameters at their default value or changed
> them as per their instructions (used raspi-config under linux
> or use the setting n config.txt as per what raspi-config
> does).

i believe the gp understood this issue, but guessed that there
was little chance of breaking this particular chip with this particular
clock speed, so it doesn't much matter if the warranty is void.

On Sat Feb  7 18:32:31 PST 2015, khm@sciops.net wrote:

> Apparently you can crash one with a light bulb:
> http://www.raspberrypi.org/forums/viewtopic.php?f=28&t=99042

crash, not broke.  and the light sensitive chip was located at iirc
u16.

- erik

Re: [9fans] Raspberry Pi 2 Model B

[Brian L. Stuart]

> Apparently you can crash one with a light bulb:

I once read that a similar thing happened when the IBM 701 was first
unveiled to the press.  IBM had put the CRT-based storage devices
behind smoked plexiglass, and one could see the memory visually.
Naturally, the photographers took flash pictures of the machine which
caused the memory to fail, crashing the machine.  Wastson's reaction
was to decree that the plexiglass be replaced with a steel plate.

BLS

Re: [9fans] Raspberry Pi 2 Model B

[Bakul Shah]

On Sat, 07 Feb 2015 18:48:28 PST erik quanstrom <quanstro@quanstro.net> wrote:
> On Sat Feb  7 18:18:29 PST 2015, bakul@bitblocks.com wrote:
> > On Sun, 08 Feb 2015 02:10:25 +0100 hiro <23hiro@gmail.com> wrote:
> > > Do the RPI2s break very fast or why is the warranty such an issue?
> >
> > A chip's rated clock rate is typically much lower that the max
> > freq it at which it can run stably -- and there is fair bit of
> > variation in this max freq.  A rare few 2836s can even run at
> > double the default frequency.  But as a side effect of
> > speeding it up and/or increasing internal voltage it will also
> > run much hotter & if you don't use a heat sink, it is likely
> > to fail much sooner.  They catch such use by setting an
> > irreversible bit inside the chip. RPF's warranty is valid only
> > if you left these parameters at their default value or changed
> > them as per their instructions (used raspi-config under linux
> > or use the setting n config.txt as per what raspi-config
> > does).
>
> i believe the gp understood this issue, but guessed that there
> was little chance of breaking this particular chip with this particular
> clock speed, so it doesn't much matter if the warranty is void.

Who's gp?

> On Sat Feb  7 18:32:31 PST 2015, khm@sciops.net wrote:
>
> > Apparently you can crash one with a light bulb:
> > http://www.raspberrypi.org/forums/viewtopic.php?f=28&t=99042
>
> crash, not broke.  and the light sensitive chip was located at iirc
> u16.

Freeze, not crash. All semiconductors are photosensitive. In
highschool I made my first photosensor by cutting off part of
a transistor metal can and removing all the white goop in it!

Re: [9fans] Raspberry Pi 2 Model B

[Richard Miller]

> Do the RPI2s break very fast or why is the warranty such an issue?

Not an issue for me - the pi is cheap enough that I consider it
essentially disposable, so I don't think twice about tinkering.  But
since I was advising people to use turbo mode it seemed courteous
to mention the consequences.

So far I've worn out the SD card slot on one two-year-old pi.
Otherwise they seem pretty robust.

Re: [9fans] Raspberry Pi 2 Model B

[Richard Miller]

>   You are correct, but Pi2 runs at a higher voltage by default
>   which meant that force_turbo was incorrectly setting the
>   "warranty" bit. That was unintended.

The mailbox property interface to the gpu has a tag for VCMSG_SET_TURBO.
I wonder if doing it that way would bypass the warranty check?

Re: [9fans] Raspberry Pi 2 Model B

[Kurt H Maier]

Quoting Richard Miller <9fans@hamnavoe.com>:

>> http://felloff.net/usr/cinap_lenrek/aijuboard/
>
> http://i1.kym-cdn.com/photos/images/original/000/014/416/X2R4UMUE536Z3WIALA7SKNVKP4RXX22Y.jpeg

GOOD NEWS

https://www.indiegogo.com/projects/aijuboard


khm

Re: [9fans] Raspberry Pi 2 Model B

[Lyndon Nerenberg]

[-- Attachment #1: Type: text/plain, Size: 133 bytes --]


On Apr 20, 2015, at 7:23 AM, Kurt H Maier <khm@sciops.net> wrote:

> GOOD NEWS

Better news!  It made it over the $5K hump.


[-- Attachment #2: Message signed with OpenPGP using GPGMail --]
[-- Type: application/pgp-signature, Size: 817 bytes --]