Originally Posted by dogsleg
I agree with you, especially since there is no even 2d acceleration to r9 290. It seems to me that there are more important things than a few mb.
I totally agree. I wanted to write (and was sure I did, I must be getting old) that I'd love to just switch to ~1gb images since 1gb usb sticks are damn cheap.
Originally Posted by dogsleg
I'm not sure how it's related to security, but I'm not an expert. It should allow for better optimization, but I doubt it will be anything to call home.
though it makes me wonder why all that seemingly shared code is not in a shared / dynamically linked library in the first place...
Originally Posted by Spittie
as i am sure there are many things that can be spared with much greater effect insted, please dont take it for granted that everyone has access to those kkinds of media. While its true on the one hand that we here can afford it there are people that cannot. (probably 40% of worlds population)
and once downloaded 1MB might not be the the dealbreaker. but the downloadtraffic that might come up will multiply on every user and every update downloaded from the server. So considering you dont have a very good broadband connection, and many areas in germany still dont have, and wont have for some time due to economical calculations, minimizing space and traffic should have a high prioroty for distros.
therefore either you build your system on your own with only the necessary stuff or you merge as much code as you can that is otherwise redundant.
And i must say that especially Ubuntu who have said their goal was to get a free OS to everyone has failed there a little. Especially with all that spyware stuff (speaking of space and ressources issues)
Can anyone please for instance tell me what zeitgeist is for? and why it is so damn huge?
i'm not an informaticien but as a customer of amd (graphic cards and processors) i give money to the company who is paying him for his work.I have nothing against Marek,and i think he is making good work.
Originally Posted by staalmannen
Working on broken DPM cards or other things would be better,maybe you are happy with the drivers for your card,but everyone didn't have a fully working card with catalyst or open source drivers,and i don't think this will make them fully working faster.
You have your opinion and i have mine,and i feel free to express it,even if it don't suit you.
I see the "link-time-optimization" stuff, but what about performance optimization if all the drivers get glued together ? It's already difficult to have a decent perf on even old cards, because driver optimization is hard to get even with 3 different drivers, so I don't like the idea of merging all together, with the fear of getting the worst performance possible regardless the card.
I have used r200 driver as mega last year and compared those, so i think there is nothing to fear or to be happy about - basically nothing changed there performance wise
if he is able to squeeze 18 mb of code into 7 mb mega driver that probably means that the drivers have a lot common code is shared among them.
a different approach could have been to package the common code into a base module. alsa is doing a lot of that. just do
lsmod | grep snd
to see what im talking about.
Im sure Marek did consider that route and I wonder why he finally opted for the mega driver option. The LTO benefits seem marginal to me unless proven otherwise.
And what is the size for both, but with existinig mesa mega drivers: THE SAME 5.3MB .