So...are there any numbers on this? Would the Raspberry Pi still be around if it weren't for the non-educational communities? --- not speculating, just genuinely curious
I think we'd be around, but our charitable work would be at a much smaller scale. Our estimate is that sales to hobbyists and (especially) industry make up well over half of the total.
I believe there were five on the station (two of ours, three we only recently found out about) until a couple of weeks ago. Now I think we're down to three (our two plus one other). Ours are used to run the Astro Pi program in partnership with ESA:
Actually we can now play pretty much all the 8-bit 1080p30 content we can find, and a fair proportion of 10-bit content too. The GPU-accelerated HEVC implementation was a beast to get working, but it runs nicely now. Checkout Milhouse's Kodi 18 nightlies here:
This is the point. There are several useful architectures (and early iterations of the ARM and x86 architectures) which have good tooling support and are out of patent coverage.
Quite eye-opening. I generally use the Pi-branded Samsung card, which is middle of the road on this test. Perhaps I should treat myself to an EVO+ for Christmas.
That's about right. If you're making a small (and by small I mean 50k unit) run it's likely to be worth buying a system on module rather than paying someone to do the fiddly HDI PCB design, finding someone who can assemble PoP reliably and buying your application processor, RAM and Flash out of distribution at high margin.
If I were building this rig, I would have used the $40 Model A+camera bundle for a cost-per-node of ~$50 including a USB Ethernet adapter and an SD card per node and a decent PSU shared between four nodes.
A bigger issue looking at the videos is the need to equalize the AGC setup (easy) and color temperature correction (harder) across the modules. Perhaps shoot RAW and then fix it with post-processing? This is where the CHDK alternative, with it's better optics and lower sensor variability, really wins out. Plus you'll have Christmas gifts for all your friends and family once you take the rig apart:)
I hope the Foundation folks say "Thank you, much appreciated", and let the kids decide.
That was pretty much what I spent the day saying. Atmosphere among the educators in the room when Conrad announced it this morning was pretty electric. If people don't like the fact that it's only free as in beer, there's always Sage.
Actually, we discard the AxPROT AXI signals as they leave the ARM complex, so it's not possible to distinguish between trusted and untrusted transactions at the memory controller. BCM2835 is actually one of the few ARM APs *not* to use TrustZone technology.
Replying on the (probably optimistic) assumption that you're actually interested.
The Raspberry Pi Foundation is a somewhat unusual charity, in that it derives the bulk of its funding from trading activities rather than through "shake the tin" fundraising. In this respect we resemble charities which run high-street retail businesses to supplement their charitable income. Once you reach a certain size, it is considered good practice to separate trading activities into another, generally wholly-owned, business entity, and to have substantially non-overlapping board membership between the charity and the trading entity. I resigned in order to reduce the overlap to a single person, Jack Lang, who chairs the boards of both entities; we subsequently added Louis Glass to the Foundation board, restoring it to the original complement of six people.
More detail in our public filings at Companies House (http://www.companieshouse.gov.uk/).
I resigned as a Foundation trustee in December, though I continue to run the Foundation on a day-to-day basis. I am also managing director of the Foundation's wholly-owned trading subsidiary, Raspberry Pi (Trading) Ltd, which handles the engineering work associated with the Pi.
- You can now buy a Model B Pi bundled with a fast 8GB SD card for $40 from both our primary distributors. - Most other cheap boards use the Cortex A8 core, which is rather a primitive implementation of the ARMv7 ISA. In particular, while it's great at memcpy() and reasonable at integer operations, it has rather poor floating point performance; for a floating-point performance comparison of ARM11+VFP, Cortex A8 and the (more modern and capable) Cortex A9, see:
Not at present, but we're expending quite a lot of effort on getting hardware-accelerated Webkit running at the moment; Wayland is a key enabler for this.
It's not a lie at all - way to play the man rather than the ball.
Most of our users care more about local performance than they do about network transparency, so this is where we're investing our (limited) resources. People who care about network transparency can continue to use X either to or from the Pi; I don't think anyone is seriously suggesting that X is going to be replaced by Wayland in all use cases, merely that Wayland meets a need for high-performance, high-quality, low-power (and lower software complexity) local composition.
Fascinating points, except that once you've offloaded top-level composition to hardware you've claimed 90% of the benefit that you would have gained from full X hardware acceleration; even on the Pi it makes sense to use the software fallback path for all in-window rendering. I did bother to look into this a bit before opening my checkbook.
Slashdot Mirror
So...are there any numbers on this? Would the Raspberry Pi still be around if it weren't for the non-educational communities? --- not speculating, just genuinely curious
I think we'd be around, but our charitable work would be at a much smaller scale. Our estimate is that sales to hobbyists and (especially) industry make up well over half of the total.
I believe there were five on the station (two of ours, three we only recently found out about) until a couple of weeks ago. Now I think we're down to three (our two plus one other). Ours are used to run the Astro Pi program in partnership with ESA:
https://astro-pi.org/
Actually we can now play pretty much all the 8-bit 1080p30 content we can find, and a fair proportion of 10-bit content too. The GPU-accelerated HEVC implementation was a beast to get working, but it runs nicely now. Checkout Milhouse's Kodi 18 nightlies here:
https://forum.kodi.tv/showthre...
/me checks driveway hopefully for Ferrari
Any chance we could get our hands on a couple of your kiln-killed units? Be interesting to see what failed.
This is an interesting point. I clearly need to think about this some more.
This is the point. There are several useful architectures (and early iterations of the ARM and x86 architectures) which have good tooling support and are out of patent coverage.
What he said.
The droids you're looking for:
http://www.midwesternmac.com/b...
Quite eye-opening. I generally use the Pi-branded Samsung card, which is middle of the road on this test. Perhaps I should treat myself to an EVO+ for Christmas.
It's a quad-core Cortex A7, so very firmly ARMv7.
No it isn't. It's a quad-core Cortex A7.
Yes - we bring out both PWM outputs to the GPIO connector now.
It's almost like it has twice as much RAM and onboard Flash, isn't it?
That's about right. If you're making a small (and by small I mean 50k unit) run it's likely to be worth buying a system on module rather than paying someone to do the fiddly HDI PCB design, finding someone who can assemble PoP reliably and buying your application processor, RAM and Flash out of distribution at high margin.
They have edited out a single black and white photograph from the original cut. Compare the first thirty seconds from the two versions:
http://www.bbc.co.uk/news/uk-2...
http://www.youtube.com/watch?v...
I wonder who asserted a copyright claim on a grainy seventy year old photograph to get this taken offline?
I grabbed a screenshot before it disappeared:
http://www.gastronomydomine.co...
Would be interesting to know why it was pulled.
If I were building this rig, I would have used the $40 Model A+camera bundle for a cost-per-node of ~$50 including a USB Ethernet adapter and an SD card per node and a decent PSU shared between four nodes.
A bigger issue looking at the videos is the need to equalize the AGC setup (easy) and color temperature correction (harder) across the modules. Perhaps shoot RAW and then fix it with post-processing? This is where the CHDK alternative, with it's better optics and lower sensor variability, really wins out. Plus you'll have Christmas gifts for all your friends and family once you take the rig apart :)
I hope the Foundation folks say "Thank you, much appreciated", and let the kids decide.
That was pretty much what I spent the day saying. Atmosphere among the educators in the room when Conrad announced it this morning was pretty electric. If people don't like the fact that it's only free as in beer, there's always Sage.
Actually, we discard the AxPROT AXI signals as they leave the ARM complex, so it's not possible to distinguish between trusted and untrusted transactions at the memory controller. BCM2835 is actually one of the few ARM APs *not* to use TrustZone technology.
Replying on the (probably optimistic) assumption that you're actually interested.
The Raspberry Pi Foundation is a somewhat unusual charity, in that it derives the bulk of its funding from trading activities rather than through "shake the tin" fundraising. In this respect we resemble charities which run high-street retail businesses to supplement their charitable income. Once you reach a certain size, it is considered good practice to separate trading activities into another, generally wholly-owned, business entity, and to have substantially non-overlapping board membership between the charity and the trading entity. I resigned in order to reduce the overlap to a single person, Jack Lang, who chairs the boards of both entities; we subsequently added Louis Glass to the Foundation board, restoring it to the original complement of six people.
More detail in our public filings at Companies House (http://www.companieshouse.gov.uk/).
I resigned as a Foundation trustee in December, though I continue to run the Foundation on a day-to-day basis. I am also managing director of the Foundation's wholly-owned trading subsidiary, Raspberry Pi (Trading) Ltd, which handles the engineering work associated with the Pi.
A couple of points to bear in mind:
- You can now buy a Model B Pi bundled with a fast 8GB SD card for $40 from both our primary distributors.
- Most other cheap boards use the Cortex A8 core, which is rather a primitive implementation of the ARMv7 ISA. In particular, while it's great at memcpy() and reasonable at integer operations, it has rather poor floating point performance; for a floating-point performance comparison of ARM11+VFP, Cortex A8 and the (more modern and capable) Cortex A9, see:
https://www.riscosopen.org/forum/forums/5/topics/466?page=7
In other news, the FreshPaper guys are amazing. Definitely the stars of the show here in Copenhagen.
Not at present, but we're expending quite a lot of effort on getting hardware-accelerated Webkit running at the moment; Wayland is a key enabler for this.
It's not a lie at all - way to play the man rather than the ball.
Most of our users care more about local performance than they do about network transparency, so this is where we're investing our (limited) resources. People who care about network transparency can continue to use X either to or from the Pi; I don't think anyone is seriously suggesting that X is going to be replaced by Wayland in all use cases, merely that Wayland meets a need for high-performance, high-quality, low-power (and lower software complexity) local composition.
Fascinating points, except that once you've offloaded top-level composition to hardware you've claimed 90% of the benefit that you would have gained from full X hardware acceleration; even on the Pi it makes sense to use the software fallback path for all in-window rendering. I did bother to look into this a bit before opening my checkbook.