Lima GPU Driver Faces Changes, Advancements

[phoronix]

Phoronix: Lima GPU Driver Faces Changes, Advancements

There are significant driver changes to report on for the reverse-engineered open-source Lima graphics driver project for providing ARM Mali graphics support...

http://www.phoronix.com/vr.php?view=MTI0NTM

[Krysto]

I think their efforts should concentrate on catching up with the Mali T600 architecture as soon as possible, and forget about Mali 200/400. Mali T600 is what we'll keep seeing not only running under Android devices, but also under Linux machines, so it's much more important to get an open source driver for it. Plus, the other architecture is a dead-end while T600 is just starting and have a life cycle of 5+ years.

[Veerappan]

I think their efforts should concentrate on catching up with the Mali T600 architecture as soon as possible, and forget about Mali 200/400. Mali T600 is what we'll keep seeing not only running under Android devices, but also under Linux machines, so it's much more important to get an open source driver for it. Plus, the other architecture is a dead-end while T600 is just starting and have a life cycle of 5+ years.

As the owner of a 7" Allwinner A10-based Android tablet with the Mali 400 GPU, I'm glad that Luc's working on this project

[cwabbott]

I think their efforts should concentrate on catching up with the Mali T600 architecture as soon as possible, and forget about Mali 200/400. Mali T600 is what we'll keep seeing not only running under Android devices, but also under Linux machines, so it's much more important to get an open source driver for it. Plus, the other architecture is a dead-end while T600 is just starting and have a life cycle of 5+ years.

The main reason we aren't working on the T-600 yet is that we don't have any hardware for it yet; hopefully, though, I'll be able to get my hands on a Samsung chromebook soon enough - should be fun for hacking on. Anyways, we don't have enough people to do both devices at the same time, given that they're likely very different and require a lot of re-learning everything. So abandoning the current driver would mean substantially more work to get something useful. Also, with the Mali 450 MP, ARM has extended the life of the mali 200/400 architecture for at least a couple more years, and consumers will still be using those devices for a few more. So even in this fast-paced market, I suspect these devices will be around for quite some time.

[Adarion]

Nice to read some good news about a project like this. The smaller RISC-CPU-GPUs-area still needs a lot of love. I will put this very very much into my considerations if I ever buy a smartphone or tablet.

[liam]

I think their efforts should concentrate on catching up with the Mali T600 architecture as soon as possible, and forget about Mali 200/400. Mali T600 is what we'll keep seeing not only running under Android devices, but also under Linux machines, so it's much more important to get an open source driver for it. Plus, the other architecture is a dead-end while T600 is just starting and have a life cycle of 5+ years.

I think they should work on whatever equipment they have available. Also, I'd imagine that much of their work can be applied to the T-600 series.

[libv]

Connor was spot on. There is quite a lot of life left in the mali400, and it is amazing how many mali400s are now flooding the market. New SoCs will be produced for the next 2-3 years still, which come with some version of this family. Heck, only now are the chinese producing mali400MP4s with the rk3066. Mali-450 still hasn't shipped anywhere.

Also consider the situation with AMD (where i played a principal role in freeing ATI, but where ATI, with the help of some ""community"" members, got the upper hand in the end): they now have 4 people working full time on it, they have all the documentation that ATI has internally (which, given, isn't much), they have all the ATI engineers to talk to directly. How long does it take them before they even start developing for a new chip? That's 4 people, full time. Connor is in highschool, and i had to lose my job before i could get some more time to work on this, go figure. From where i sit, 8 months have been wasted between FOSDEM and now.

T6xx will happen once we are relatively happy with mali200/400. That is: Connor has a useful compiler out that we can plug into a useful mesa driver. Anything else is stupid timewasting which will never get any results on any front. Finish one job before you start the other!

[libv]

As the owner of a 7" Allwinner A10-based Android tablet with the Mali 400 GPU, I'm glad that Luc's working on this project

Check out http://linux-sunxi.org/Category:A10_Tablets and then read through http://linux-sunxi.org/FirstSteps to get yourself a proper linux on an SDCard. Get onto freenode, the #arm-netbook channel, and ask some good questions, and you should soon have your tablet running a proper linux.

[stqn]

As the owner of a totally useless MK802 (Android sucks, bad resolution, no updates, and it can’t play all webm files) I’m also glad to read there’s progress on the Mali400 driver . I’m more interested in video playback though.

Edit: looks like hw accelerated video decoding is already (partially) implemented? http://linux-sunxi.org/VLC … hm…

[libv]

Edit: looks like hw accelerated video decoding is already (partially) implemented? http://linux-sunxi.org/VLC … hm…

There were also some REing attempts, i think one guy, Iain Bullard?, after i gave him some tips, ended up writing the needed framework for media engine REing, but i think he gave up shortly afterwards. This is too bad, as he seemed a rather promising character, capable of tackling difficult problems.