Actually, the way it's written it seems that folks at Avionics Designs did not have access to any documentation and it's pretty much a reverse-engineered effort anyway.
I am no expert in this field but I distinctly remember back when I had a Tegra 2 (LG Optimus 2X) phone that the major obstacle of porting ICS to it was that there was no documentation and no open source code. With this barrier gone I hope Nvidia will be able to pick up some steam again looking as to how they pissed most of the cellphone manufacturers according to this article http://semiaccurate.com/2013/02/18/nvidias-telegraphs-tegras... (again, don't know how true this is as I am no expert)
This could also be an interesting development for the Ouya game console, maybe game devs will be able to get more performance out of the Tegra 3 chip with it's sourcecode out in the open.
Yeah, got a Moto Atrix 4G lying around running Gingerbread when it could perfectly run ICS/JB. Not only that I have had the most instability with Nvidia Android devices. Nvidia really needs to learn a lesson and up their mobile game.
Edit: From another HN thread - Nvidia seems to be almost at the bottom as far as mobile GPU market share is concerned - http://stats.unity3d.com/mobile/gpu.html - which is definitely a cause for concern given the volumes they need.
Even if you ignore that data - all of the popular Android devices have been non-Nvidia except for the Nexus 7. (GS3, Note2, various Asian/European devices etc.) Even for next gen Nexus 7 there are rumors about Google/Asus dumping Tegra and going with Qualcomm.
Isn't this the case with a lot of Android devices, regardless of the SoC vendor or device manufacturer? There are tons of devices out there that are stuck on Android GB or older, not because of hardware limitations but because of software.
Android is not a very stable platform, there's a big code release coming from Google once or twice a year and it's a lot of work to make drivers and other software play nice with the new version. Device manufacturers and SoC vendors have no financial motivation on keeping the software up to date with previous generation devices, the priorities are on shipping new devices.
This is the biggest problem I see in the Android ecosystem. Smartphones and tablets reach the end of their life prematurely when you can't use the latest software even if the hardware has capabilities to run it. When I buy a high end device, I expect it to serve me for 5 years or so and stay up to date with the software.
While I would love to have a mobile device that gets updates for 5 years, I don't believe such a device exists (even Community ROM Projects for Android have troubles keeping up on performance after a couple of years [like the Nexus One]). Nexus devices and iPhones are near the max and that's around ~2 years.
I'm just guessing your comments lacked some context and you meant something slightly different than what was stated?
Is this nVidia's driver? Or some random code that can draw a solid color triangle on a tegra? Where's the shader compiler backend and the rest of libGLESv2?
Yes, grate is similar to my effort for Vivante GPUs https://github.com/laanwj/etna_viv . I'm not sure how they confused this with official NVidia drivers, as it's built by a third party. NVidia open sourcing drivers would be huge news indeed.
FOSS linux drivers (or even just driver specs) on Tegra 3 open up those devices to run so much more stuff, and allow communities to start working on full, open linux stacks for Tegra/ARM devices. Yay!
From what I've read it sounds like there's Nvida's official closed-source driver and a separate third-party open-source driver that's been approved by Nvidia in some way (I guess that means they won't sue over it).
nVidia chips are intended for high-end tablets but fall short. They are essentially trying to compete with "good enough" GPU/CPUs that cost less than $10 wholesale, but a Tegra 3 is about $20 wholesale.
Actually, the way it's written it seems that folks at Avionics Designs did not have access to any documentation and it's pretty much a reverse-engineered effort anyway.