On Wed, 12 Nov 2008 23:57:28 +0000 Matthew Allum <[EMAIL PROTECTED]> babbled:
> Hi; > > On Wed, 2008-11-12 at 01:32 -0800, Carsten Haitzler wrote: > > > > > > Uuhhh - if your saying you take a hit in moving image data from system > > > to texture memory then yes of course you will on certain hardware > > > setups. How much this hit is and how relevant that is to the real world > > > is debatable *but* this is a dumb argument anyway Clutter on GPU will > > > outperform many 100x in terms of rendering operations (and more) > > > compared to software only options. If it did not, GPU's would not exist. > > > > hmmm i would have to say "many 100x" is going to be a major leap. we > > actually have benchmarks. gl vs software. sure.some things are a lot faster > > in gl (5-10x). others.. software beats gl. in fact the averags fps i get on > > a normal desktop (2x core core2 3ghz) vs an nvidia 8600gt - evas. gl vs > > software_x11 - over all tests in expedite (which is a fair few) gl gets an > > average of about 1100fps (from memory) and software is pushing 800fps or > > so... got real numbers. that's from memory - but on my current laptop. > > > > nvidia 8600m (mobile) vs core2 due 2.3ghz mobile: > > 868fps vs 671fps (for all tests) > > > > let me quote some "interesting results". > > [snip] > > I cannot believe you are trying to argue that graphics on a GPU are only > marginally faster than all done on the CPU. Have you ever tried running > a modern first person shooter all in software ? on the other hand i can't believe you will quote totally non-factual numbers without actual benchmarks - which unlike you, i have. comparing using gl for 2d rendering vs cpu. it is a relatively fair test. we were in the context of clutter vs efl which is is primarily the world of 2d ui rendering. we were not talking about FPS's etc. i will also point out that software gl implementations are very far from optimal. very far. no attention is paid beyond just making it correct/work. the intent is to go via hardware. you quote figures that are not true - given the context. clutter does do 3d - and do effects. but it's primary use is 2d with "3d bonus effects" from everything i have seen. it is NOT an FPS engine. neither is EFL. my problem is that you like to quote fallacies on efl's part (eg it saturates your cpu which is false - if you use the gl engine - which means you are doing just what clutter does). you quote speedup numbers that i know are just not true and i have numbers to prove it. they are nowhere near the 100x you quote. if you are going to state that clutter is far superior as it's faster because gl is 100x faster - then that is a fallacy. it'd be good to stick to facts. people read this list and then just take what you say as "the truth" unless someone calls you out on your figures and disputes them. > My point re CPU usage is that by its very nature graphics effects done > with OpenGL will use less CPU as they are doing the work on the GPU not > the CPU. I am sure if you know what you are doing - i.e avoiding heavy > large animations etc - you can avoid starving the CPU with evas. With > Clutter this kind of thing isn't so much an issue. The downside is of > course with Clutter is that your hardware needs accelerated GL and evas > does not have that requirement. > > When I created Clutter I made the assumption that GPUs were becoming > commodity and thus it made sense to build and design Clutter only > targeting OpenGL based hardware accelerated backends. That also meant > not bringing in the 'baggage' of also supporting software rendering in > its design of which I think is a big advantage (mainly due peculiarity's > of OpenGL API). Its obvious you disagree with this approach - and surely > thats a good thing, it means Clutter is not treading on evas's toes. > > Please can we just leave it at that, this is getting silly. Surely you > have better things to do with your time ? > > == Matthew > > > > --------------------------------------------------------------------- > Intel Corporation (UK) Limited > Registered No. 1134945 (England) > Registered Office: Pipers Way, Swindon SN3 1RJ > VAT No: 860 2173 47 > > This e-mail and any attachments may contain confidential material for > the sole use of the intended recipient(s). Any review or distribution > by others is strictly prohibited. If you are not the intended > recipient, please contact the sender and delete all copies. -- ------------- Codito, ergo sum - "I code, therefore I am" -------------- The Rasterman (Carsten Haitzler) [EMAIL PROTECTED] _______________________________________________ Moblin dev Mailing List [email protected] To manage or unsubscribe from this mailing list visit: https://lists.moblin.org/mailman/listinfo/dev or your user account on http://moblin.org once logged in. For more information on the Moblin Developer Mailing lists visit: http://moblin.org/community/mailing-lists
