A birthday present from Broadcom

http://www.raspberrypi.org/archives/6299

by 1ris • 12 years ago

625 161 12 years ago

raspberrypi.org

leigh_t• 12 years ago

This is awesome news. I hope other vendors follow suit.

LukeB_UK• 12 years ago

This is great news. Hopefully this means that ports of android and chromiumOS will be able to utilize this.

fnordfnordfnord3• 12 years ago

"Earlier today, Broadcom announced the release of full documentation for the VideoCore IV graphics core, and a complete source release of the graphics stack under a 3-clause BSD license. The source release targets the BCM21553 cellphone chip, but it should be reasonably straightforward to port this to the BCM2835, allowing access to the graphics core without using the blob. As an incentive to do this work, we will pay a bounty of $10,000 to the first person to demonstrate to us satisfactorily that they can successfully run Quake III at a playable framerate on Raspberry Pi using these drivers. "

At last!

marincounty• 12 years ago

Just $10,000? Nice gesture, but they can afford to dig a little deeper. Their card running Quake III at reasonable frame rate on a Pi is a big deal, and these cheap skates know it. (I hope the person who figures it out, doesn't give them the code; just show them and collect $10,000 (A years rent in a bad part of Oakland).

dredge• 12 years ago

(a) They are a charity.

(b) They demonstrated the Pi running Quake 3 at a reasonable frame rate in 2011. There's a picture of it in the blog post.

(c) The competition rules say "To submit an Entry, and Entrant shall email a link to a public GitHub repository containing the full source code for the Entry..." so just showing the result wouldn't be eligible to win. That and they already know what it looks like, because of (b).

Narishma• 12 years ago

Q3 is already running on the Pi using the proprietary drivers. This competition is for porting these new open sourced drivers to the Pi and using them to run Q3.

malkia• 12 years ago

It could be that $10,000 is the amount some mid-to-high level manager can probably write off.

Also $10,000 is a significant amount of money.

patrickyeon• 12 years ago

That's really a surprise. Broadcom is cemented in my mind as one of the hardest companies to get even basic datasheets out of.

I know the saying is "don't look a gift horse in the mouth", but could it be that their graphics cores are not competitive in pure technical capabilities, so they've decided to compete on open-ness? I'm not really involved in that area of hardware/software.

ebenupton15• 12 years ago

Nope (though note, I'm biased for several reasons). In terms of performance per unit area and performance per Watt, VideoCore IV is about as good as you're going to get. There are chips with higher performance GPUs out there, but they get there by brute force: throwing area (and therefore cost) and/or power at the problem.

stingraycharles2• 12 years ago

This is confirmed by the recent move of AMD towards extremely hot GPUs. You already see that in the most recent 290x, but they're talking about making hardware that is comfortable running at >100C.

I'm not sure how relevant that is with regards to performance per watt, but I do know that heat is one of the most expensive (by-)products of electricity.

AceJohnny23• 12 years ago

Two years ago (or whenever the BCM2835 itself came out on the market), sure, but surely things have improved since? We've at least jumped a technology node, and hasn't nVidia improved?

xamlhacker• 12 years ago

Waiting for the VideoCore V :) I noticed that the Broadcom LTE offerings such as the M320 do not use VideoCore and instead use PowerVR 5XT. Is it because they were originally Renesas designs? Hoping to see more VideoCore action in the future :)

akiselev7• 12 years ago

Looking at the long term future of the market, Broadcom's move isn't surprising at all. Broadcom (and Qualcomm for that matter) is feeling a lot of pressure from Intel and NVidia, both of whom have a lot of new stuff that are scaring competitors (which is fucking great for us). They're scared not from just a technological or capital/market perspective, but also from a community exposure one. They're both firing on all cylinders right now improving their technology and trying to get as many engineers to use it as possible. While firmware engineers tend to be older, there is a constant cycle and the industry is starting to see the importance of open source, especially as die complexity grows and more and more testing is required for each silicon design.

Intel: 2.5W TDP quad core Atoms with Intel's x86 instruction pipeline and memory management unit have been released as well as a low cost SDR that should be able to dynamically switch between unlicensed, GSM, CDMA, and LTE spectrum. Intel's graphics work is also quickly catching up and will soon start trickling into their low power embedded chips. Intel has also made clear moves with Galileo to open up their technology to the hobbyist engineers.

NVIDIA: The Tegra K1 is out to NVIDIA's partners (supposedly it beats VideoCore and even low end integrated desktop graphics) and they are currently being worked into tablets. NVIDIA also has recently released the i500, their own software defined radio to compete with Intel, and both companies have the resources to end the lock-in that Qualcomm and Broadcom have had in their respective baseband markets. If you're already buying radios from them, it makes sense to get your processor from them too and Broadcom may be trying to hedge against that.

NVIDIA (and maybe Intel) aren't known to be pioneers of open source but they're better than most other Silicon companies and everyone else is really starting to feel the pressure. The problem of supporting many different device configurations is an old one on desktop systems and that problem is only magnified 10x for embedded engineers because of the opacity. With Intel and NVIDIAs expertise on the desktop as they're running full steam ahead into embedded (and kicking ass along the way), they're scary to the old guard.

zanny2• 12 years ago

NVIDIA (and maybe Intel)

Intel has a huge workforce dedicated to open source. They fund huge chunks of kernel, Wayland, X and Mesa development, and I'm sure they have engineers on projects throughout the foss stack. Yeah, their hardware is pretty systemically evil patent drivel black box nonsense, and their chipsets are some of the most obscure bullshit ever, but everything past the firmware they do a good job on.

Nvidia on the other hand consistently blows whole stack. Their foss Tegra drivers are always, at most, 2d accelerated, and all their recent contributions to Nouveau (all two of them) are just to make SoCs bootable on novueau so you can install their blob driver.

Meanwhile, technologies like CUDA, GSYNC, PHYSX, etc are all proprietary platform locked messes that hurt open tech in general. So I'd definitely rank them way below Intel on a comparison of openness.

pjc50• 12 years ago

This is almost certainly driven entirely by Eben Upton, who is the link between Broadcom and Raspberry Pi. He's probably been bundling up the huge list of demands for OSS video drivers and dumping them on the desk of the right person, while wearing down the objectors. It's something that Raspberry Pi have wanted for ages.

Edit: ah, Eben has turned up here himself :)

noonespecial• 12 years ago

This has been my biggest disappointment about the raspi since day one. If true, this is huge news indeed.

13throwaway• 12 years ago

They donate to WebKit? I thought that was fully funded by apple.

skykooler• 12 years ago

Google funds WebKit as well.

1ris• 12 years ago

No, they did fund large parts of it, but forked quite a while ago.

ebenupton15• 12 years ago

Yes. See here:

http://www.raspberrypi.org/archives/5535

a lot of this work goes upstream.

mjg59• 12 years ago

There's a significant WebKit consultancy industry.

davexunit2• 12 years ago

This is a step in the right direction. Does this mean that the RaspberryPi can run blob-free now or with one less blob? I don't know what else is needed for the SoC to operate.

mhurron• 12 years ago

From the article -

"This isn’t the end of the road for us: there are still significant parts of the multimedia hardware on BCM2835 which are only accessible via the blob."

davexunit2• 12 years ago

Ah, thanks. Poor reading comprehension on my part.

ebenupton15• 12 years ago

Not quite yet. We're hoping to be able to provide a minimum-functionality blobless world fairly soon.

MrZongle2• 12 years ago

I think this is another example of how Raspberry Pi Foundation's decision not to move forward with a gazillion different boards, but rather to stick with A and B and let the software evolve, was a good one.

I'm very excited to see what the next year brings.

polshaw2• 12 years ago

Are you able to give any indication what might be included (or not) in such minimum functionality?

pbhjpbhj• 12 years ago

Will running without relying on blobs mean that processing can be further optimised or is it more of a moral victory than a technical progression.

I'm assuming doing away with the shim means that there is some gain but in practical terms is it likely to be quite minimal? Will there be routines accessible that weren't accessible via the shim?

ebenupton15• 12 years ago

More a moral victory from my PoV. Remember, code that runs in the blob is running on another processor, so not costing you ARM cycles. That's why we've set the Quake fps target much lower than what you can hit with the blob.

josh-wrale2• 12 years ago

I wonder if anything GPGPU will become of this... I'm thinking it's probably not possible. OpenCL?

ZeroGravitas• 12 years ago

One of the bullet points on the broadcom blog post is: "Write general-purpose code leveraging the GPU compute capability on VideoCore devices"

josh-wrale2• 12 years ago

Keen eye (and mind).. Thanks.

justin66• 12 years ago

It's great news. It's also not quite enough to get rid of the entire binary blob, as the article notes near the bottom. (I have the feeling some didn't read that far.)

rcarmo• 12 years ago

Well, let's see if someone can get Android to use the GPU now.

And yes, I'm not expecting miracles -- even though Broadcom never released their 4.x port, what they demoed on the-teaser-video-that-was-never-followed-up-on seemed "fast enough" for a number of purposes.

mik3y• 12 years ago

IIRC one of the bigger obstacles to RPi Android support was that the driver libs were linked against glibc. There will be other bumps, but just getting these linked against bionic will be a big help.

I'd love to see Android support, if only for Android's recovery system and read-only system images.

jxf3• 12 years ago

If you want to go straight to the horse's mouth and just start hacking (this is Hacker News, after all!), the Broadcom downloads are here:

* BCM21553 VideoCore IV graphics driver source: http://www.broadcom.com/docs/support/videocore/Brcm_Android_...

* BCM21553 VideoCore IV documentation: http://www.broadcom.com/docs/support/videocore/VideoCoreIV-A...

ansimionescu• 12 years ago

Good ol' gzip download, because fuck source control.

edit: FWIW, a friend who works there told me flat-out to just avoid them (I'm job-hunting)

CamperBob23• 12 years ago

Look gift horses in the mouth much?

Why "avoid" them?

petermonsson2• 12 years ago

It is open source. You can put it on Github if you want to.

akiselev7• 12 years ago

This is almost a universal problem with firmware and chip engineers, who are generally older and more experienced, and thus apparently can't be taught new tricks. There are also issues with bureaucracies in each and every semiconductor company (which are on average decades old) so firmware is usually hot potatoed around by management and ends up lighting the customers' projects on fire almost every damn time. I don't have enough fingers to count how many times I've encountered problems in a project and wasted weeks treading water, eventually to find out that the code examples weren't tested, a parameter that closely effects very sensitive silicon isn't documented, or even that I can't so much as get the documentation for a licensed USB peripheral that is non EHCI compliant (who knew such a thing even existed anymore).

Everyone from Broadcom to Qualcomm (and many more) are guilty of this and it is a very rare treat to find a chip that is well documented with firmware and code examples that work and give you a full picture of what you can do and where the pitfalls are.

gtlk• 12 years ago

Don't worry, they're going to have to release it again. This zip contains many files with copyright (c) Broadcom (and other companies) missing the 3-clause BSD license.

On an unrelated note, x86 looks like RISC compared to this chip assembly language :)

   vmul  -,H(1,0),3 CLRA ACC
   vadd  HX(3,0),H(0,0),2 ACC
   
   vasr  H(0++,0),HX(2++,0),2 REP 2
   ;vmov  H(0++,0),0 REP 2
   vst  H(0++,0),(r0+=r3) REP 2 
   add   r0,16
   addcmpblt r5,1,r1,loopacross

jotux2• 12 years ago

Why would a large company release the commit history of source code when it could open them up to all kinds of lawsuits? It seems perfectly reasonable for them to release it as a zip.

hermanhermitage2• 12 years ago

For those interested, as far as I understand it this release doesn't include VPU or QPU assembler tools.

We released a QPU assembler at https://github.com/hermanhermitage/videocoreiv-qpu/blob/mast... a couple of weeks ago. And a reference javascript VPU assembler is due very soon (mirroring the path taken with the QPU assembler).

Because our own work was independent and pre-dates this information by a couple of years (in the case of the VPU), we have some alignment work to do in terms of mnemonics.

This release adds a lot of knowledge to the table - extending our understanding beyond the core instruction sets (which were already well reverse engineered) into a deeper understanding of the various hardware registers.

ebenupton15• 12 years ago

Figured you'd get a kick out of this release. Should be enough for people to figure out how Andrew's FFT code works.

srcmap• 12 years ago

I love raspberrypi!

I created the srcmap index of raspberrypi_userland src tree here:

http://www.srcmap.org/s/sl.htm/p=raspberrypi_userland#c=P&d=...

Hope you guys find it useful.

Danieru2• 12 years ago

This is incredible. Thank you Broadcom!

monk_the_dog• 12 years ago

I'm building an environment for computer vision, and I recently made a client that can run on the rpi (two minute video if you're interested: http://www.youtube.com/watch?v=qSigkvbpb_Q). I'm not exactly sure what possibilities this opens up, but I'm definitely going to dig into it and find out.

tejbirwason• 12 years ago

Proud to have worked on this during my 8-month internship at Broadcom. It remains one of the toughest and rewarding projects to tackle within the company. I am pleasantly surprised to see Broadcom open up like this and release the full documentation. A step in the right direction for a grossly under-rated company.

samolang• 12 years ago

I've never heard of an 8-month internship. What sort of program is that? Like a semester+summer?

cbhl2• 12 years ago

Probably a co-op program with a University. the University of Waterloo often does 4-month and/or 8-month (depending on the program); the University of Toronto does 12-month, in partnership with companies that are looking to woo students before they've even finished their undergrad.

(Usually it involves the school designing the academic program and shifting when courses are offered (including putting courses in the spring/summer) to accommodate the work term.)

The grandparent appears to go to the University of British Columbia, judging from their profile.

Elv132• 12 years ago

My university (Montreal ETS) also do it. They have 3 (real/full) semesters per years. Students usually take 1 of them per year to work as interns. You can technically work as intern for like 5 years and still be considered enrolled into the engineering program. Students will either prefer 8 months internship for well paid jobs or jobs that cannot be done in a 4 months development cycle(like a video game). In the end, it does increase the total time spent to graduate, but students tend to have much lower debts (if not profit) at the end.

These kind of deals seem to be quite widespread in Canada (just like the above commenter show). I know these private sector <-> teaching organization deals are not well considered in some part of the world, but, in the end, someone who just graduated, but accumulate years of full time working experience have an easier time landing a job at the end, at least for STEM jobs.

bliti• 12 years ago

Well done, mate. This is some great work that you did.

rdb• 12 years ago

I remember how satisfying it was to get that demo to work back in 2011 (thanks to the help of Dave Emett). Feels like an incredibly long time ago now! I wonder what framerate people will be able to achieve...

rwg• 12 years ago

But we’re incredibly proud that VideoCore IV is the first publicly documented mobile graphics core

Wasn't Intel first by virtue of using their in-house "HD Graphics" hardware in Silvermont (Bay Trail-M and -T)? Regardless, this is a huge thing in the ARM world, and I hope other vendors will begin to open up documentation.

ebenupton15• 12 years ago

Good point. Still hard for me to remember that mobile != ARM these days.

kllrnohj• 12 years ago

Bay Trail-M means "mobile" as in "laptop", not "mobile" as in "smartphone". Intel's smartphone stuff all still using PowerVR, including the upcoming Merrifield which will be using the PowerVR G64x0.

zurn• 12 years ago

In common with every other ARM-based SoC, using the VideoCore IV 3d graphics core on the Pi requires a block of closed-source binary driver

The Lima driver for ARM Mali is open source and runs Quake 3!

wolfgke2• 12 years ago

The Lima driver for ARM Mali is open source and runs Quake 3!

But the Lima driver was created by reverse engineering instead of using publicly available documentation.

makomk2• 12 years ago

Any open source replacement for the Raspberry Pi GPU driver is going to rely pretty heavily on reverse engineering too. The code they've released runs on the VPU, a custom-designed scalar processor within the VideoCore IV that Broadcom still haven't released architecture documentation or a toolchain for. Without a whole bunch of low-level reverse engineering (much of which was already in progress) it won't even be possible to compile the stuff Broadcom have released. Without working VPU code you can't even boot the Pi, let alone make use of graphics.

Argorak3• 12 years ago

Celebrate the open source release of the drivers by giving out a price for making a non-free software run on it? Also one that the intended audience of the hardware is not allowed to play? Seems a bit odd.

(The Quake 3 engine is open source and free nowadays, but the assets and maps are not, so you cannot run "Quake 3" without using the demo. So "Quake 3" is non-free.)

angersock• 12 years ago

Erm, I'm pretty sure you can use a shareware version (!) of the Q3 assets. Or any of a number of other things.

The idTech2 and idTech3 engines are standbys of the open sores community--it makes sense to use them for testing graphics drivers. You are far more likely to encounter a bug in your driver than in those engines.

Argorak3• 12 years ago

Even if free as beer, the demo version is non-free, in an open source sense.

Sure, but you could use any of the many projects built on the open Q3 engine that are actually free as in speech. (And maybe child-friendly)

Also, to be clear: I find it odd, but still a fun endeavor.

CamperBob23• 12 years ago

Who, other than certain parents with opinions they can't scientifically justify, says anyone isn't "allowed to play" Quake 3?

Argorak3• 12 years ago

In Germany, giving Quake 3 into the hands of minors is an offence, for starters.

So, in our case, these nice people: https://en.wikipedia.org/wiki/Federal_Department_for_Media_H...

Like it or not.

georgemcbay• 12 years ago

On the one hand as someone who owns half a dozen Raspberry Pis this is really cool news and I hate to look a gift horse in the mouth.

On the other hand, ARM vendors really have to start opening everything up much more than they have or Intel is going to start eating their lunch in the small embedded Linux arena, and I suspect Intel's mobile resurgence with Bay Trail is at least partly behind this "gift" to us all.

penguindev• 12 years ago

I don't know anything about mobile, but I've been looking at replacing my htpc ION system (edit: not that it really needs it, frankly it works great) and fanless baytrail boards/cases look very interesting. More cost than a pi + pi codec licensing obviously - but not that much, and all open as you pointed out...

userbinator2• 12 years ago

I looked up the documents on it and Bay Trail is surprisingly open for an Intel SoC... although there's probably still plenty of stuff they won't tell you about.

paulofisch• 12 years ago

The main touchy-feely moment for me here is that a notoriously closed, $17-billion company was brought around by a friendly, gentle approach to convincing it to be more open. Group hug for Broadcom.

zanny2• 12 years ago

Same could be said of their (albeit paltry) wifi open driver support since 2011. They at least have one now, for very few chipsets. Doesn't work with any of the half dozen Broadcom NICs I have in my "junk parts" bin, though.

BenSS• 12 years ago

Would be awesome if we soon see OpenCl on the RPi. Now I've got to go find mine!

caycep• 12 years ago

Quick noob question...is the Pi still the best/most flexible option for general hardware hacking? What about ARM dev boards from Samsung, Qualcomm etc?

blacksmith_tb• 12 years ago

'Best' can measured in a few ways, the RPi has a huge community, and it's certainly a bargain (though you will end up buying a bunch of odds and ends to get going). The Beaglebone Black is probably the other major contender, way more GPIO pins, can boot from the onboard memory, is ARMv7, for $10 more.

krasin• 12 years ago

BeagleBone Black has just one USB port, which is often very inconvenient. For example, if you want to run Octoprint accessible via WiFi and control a 3d printer, two USB ports will be needed. Certainly, a USB hub is an option, but it makes the setup more brittle (some USB WiFi dongles won't work over some hubs) and more expensive.

blah324972• 12 years ago

Oh woah - I'm very happy about the BSD license. This means people can more easily potentially spin up business around this platform. I wonder why they chose it...

Does anyone have any insight?

ebenupton15• 12 years ago

BSD-like licenses seem to be the standard for userland graphics libraries. There's a lot of useful stuff in there (for example the shader compiler) that may be of general use elsewhere, and we wanted to make sure people could use it.

The 3-clause BSD is a compromise - we're happy for the stuff to be used but it's nice to get some credit :)

blah324972• 12 years ago

Thank you for the clarification. It seems overly generous =)

Often code is GPL'd with pricey licenses for commercial applications - which always makes me think twice about investing my time into familiarizing with it.

I'm very impressed they aren't trying to make a quick buck here. It sounds like they have a very innovative long term strategy that will pay off many times over.

I think your competition is also fantastic, as it will encourage people to open source their improvements.

Thank you for all your hard work Dr.Upton

aswanson• 12 years ago

Wow. In due time these guys might start actually releasing datasheets to developers who, you know, need to know the register set and API to like, make their chips work. I don't know why the hell these idiots treat their firmware APIs with Manhattan-project style secrecy, but I would highly advise against designing a new product with a broadcom chipset. They are totally asinine.

userbinator2• 12 years ago

Wow. Much open, such surprise.

What next, a full BCM2835 datasheet?

To put in the words of Torvalds: "Hey, this time I'm raising a thumb for Broadcom. Good times."

stbtrax• 12 years ago

I don't think you can even start a new product with a broadcom soc if you're not already a huge player.

tomphoolery• 12 years ago

lol i lost it...

Okay, so sue us: we launched on February 29.
Please don’t sue us.

pja• 12 years ago

This is genuinely great news & a vast improvement over the previous graphics driver!

Hopefully this means we'll see proper DRM / Gallium support for the IV & I'm looking forward to seeing what people can come up with now that they have full access to the GPU.

Congrats to Eben & the team - you've done good.

glennos• 12 years ago

My favourite part of this announcement is:

* Okay, so sue us: we launched on February 29. Please don’t sue us.

A birthday present from Broadcom

Related Stories