Thanks for clarifying what I should have said.
Amazing thread, at least were talking about it now. Thats progress
Not really… we’ve been talking about it for 10 years. There were even bounties for it in the past (I think the radeon_hd drive might have been kicked off with some of that perhaps don’t quote me on that though).
I was just rambeling to a friend of mine about hardware aceleration, and what he said made a lot of sense… It will be nearly IMPOSSIBLE to get hardware acceleration happening for all available cards…
The first step sould be to limit the cards that will support it… (pretty sure thats how they did it over at MorphOS)
There has already been a lot of work done in the past on this, and (even though i cant get it to work on Haiku) nvidia and ati did have a good driver for at least 2d accell on BeOS.
Maybe this is where we should start discussing which chipset would be most favourable?
You can discuss this as much as you want. The first chipset to get 3D acceleration will be whatever the developer who gets 3D acceleration to work decided to use. Probably because it’s the one they have in their own computer.
Remember Haiku is not a corporate-driven project with a clear roadmap and plans and all that. It’s people with generally a common vision, but all working towards it in their own ways and means.
So, unless the thing is “I’m planning to start hacking on 3D acceleration, which of the cards I own would be easiest to get working or are they all desperate cases and should I buy a new one?”, there is no point in discussing which card would theoretically be easier.
In general there is only one manufacturer and thus driver possible to arrive first, and that’s the open radeon driver one. nVidia is basically closed source compared to this one. Anything else but the open source driver is practically impossible since it requires reverse engineering.
That said, don’t get your hopes high. Radeon driver is build on Linux Kernel/DRI/DRM so that’s not a flick-a-switch task. But compared to anything else most probably faster since it’s actually documented (somewhat).
Intel and Radeon HD are probably equally viable (I think the Radeon documentation is a bit better though perhaps). Intel is also porting thier recent GPUs to also use gallium instead of a classic mesa style driver I think they call the new driver “Iris” but I don’t think it is specific to the Iris models of the GPUs.
I’m partial to radeon but… most developers probably have Intel laptops currently at least though AMD is catching back up in that space and has some really nice features in some laptops (Freesync etc…).
The Iris driver is a user-space change only, as far as I’m aware. The kernel-mode components are exactly the same and are not changing much, if at all. So their efforts there are nice but ultimately don’t make much difference to us for porting drivers.
True, it does at least make the all userspace codepaths all go through gallium though… so that’s probably a win at some level maintenance wise.
What is the prefered bounty site these days for Haiku?
When you find out, please shout. I’m happy to put some cash towards someone working on this. It’s the last hurdle for Haiku.
There is no bounty site currently. Our current devs all have full time jobs they do not wih to quit for the uncertain life of software bounty hunters. If, for example, you want me to work full time on Haiku for long period if times, you have to secure funding for this to be a long term job (years, not months), and at a decent pay rate (I may be ok to accept less than what I earn currently, but still).
There are personal choices. In my case, I know working as a contractor was not so great with all the paperwork and other minor annoyances I had to deal with (like, different healthcare service). This is specific to my country of residence, but eventually I’d say working on my free time with no contract is more productive and motivating because it allows me to try various things, start many projects, and explore and discover. I’m wasting only my own time, not someone else’s money. My year of work on WebKit also put me in the uncomfortable situation that I get blamed for all the bugs there, yet no one else took a look at the code. I value working as a team a lot (with code reviews, etc), and having one person working full-time on something is a risk of something like that happening (bypass reviews because they slow you down, then no one becomes knowledgeable about the code and when you’re in trouble, no one can help you).
Throwing money at people is not an universal solution. It’s an enabler in some cases, but not always.
I am thinking more tactically than that. When I was completing my CS degree in the UK I worked for an IT consultancy building C++ and VB6 apps and being paid £10 per hour, ish. I think the same scenario would work out here, if the requirements and the scope were limited.
Sounds pretty much exactly GSoC and Outreachy?
I suppose, but as a white straight male, I’d have been excluded from participating in Outreachy
We can see how we go with GSoS and Outreachy, but I strongly suggest we need an alternative strategy.
Haiku inc accepts proposals from people who would want to work for Haiku. If you’re interested, get to talk to them. It’s not that hard, I already did it (first time in 2010, after a succesful GSoC but without much more experience besides that).
We also used to have bounties for specific things. Pros and Cons of course to that… but as it stands there isnt a lot driving advancement in Haiku other than just the developers own motivation.
I’d like to see the return of bounties for specific tasks eg, Web+ adblocking, user scripts, multiple audio cards, USB audio, 3d driver work split across several bounties probably best to make it 1 fully working driver , the smaller bounties for the next cards implemented that way we don’t pay out for unfinished work etc… I’m more likely to donate for specific bounties than open ended support of the project.
Bounties for arch support, like support for sbus/upa/afx on sparc etc…
There’s a big difference between contract work as you say which is untenable for developers with day jobs, but bounties on the other hand make more sense for small jobs that can get done in pieces here and there… to make Haiku better as well as reward and say thanks for all the care that is taken in development.
Maybe you are targetting me with these sparc things?
My position is clear: I don’t need the money, I have a well paying full time job, and I do Haiku things for fun in the evening and weekends. If you set a bounty, why would I care?
Funding would be great only if it allows me to stop having a job besides Haiku, so I can spend more time on Haiku. Otherwise, the outcome is null. If I happen to fill a bounty by accident (because I was working on the stuff anyway), I would then give the funds back to Haiku inc. Because I have absolutely no need for them.
Of course, this is only my personal situation, so if you can find someone else who thinks earning money on their work is helpful, let that happen. But it won’t be me.
Not targeting at all… it just happens to be my area of interest so, that’s what I mentioned the same would go for any other arch, or any other feature for that matter that anyone wanted. I certainly don’t think I can pressure you into doing things by throwing money at you haha.