Re: OpenGL Blitter for OpenTTD
Posted: 12 Jul 2008 09:53
Tiberius, what kind of file extension is that???
The place to talk about Transport Tycoon
https://www.tt-forums.net/
If you are talking about "opengl-080711_r13691.patch" then it's a file that contains the differences between the official source code of the game and the code that Tiberius is working on.G4 wrote:Tiberius, what kind of file extension is that???
It has not been added to trunk. You can see that your self by looking on the change log or simply downloading last nightly. Also if it would have been added or close to being added I would expect much more activity about it on the forums and/or IRC.GreatBunzinni wrote:So what's the state of this patch? Has it been added to the trunk? Has it died?
It seems a bit strange not do so. Wouldn't openttd benefit from the addition of a opengl blitter?Zuu wrote: It has not been added to trunk. You can see that your self by looking on the change log or simply downloading last nightly. Also if it would have been added or close to being added I would expect much more activity about it on the forums and/or IRC.
The least you could expect if it was added to trunk is a notice in this thread that it has been added.
Would a Toyota Prius benefit from getting it's engine replaced by a Ferrari Formula 1 engine?GreatBunzinni wrote:Wouldn't openttd benefit from the addition of a opengl blitter?
I was merely pointing out that making assumptions that all and any patches are beneficial for OpenTTD is strange if not stupid. Yes, there's an opengl blitter and it might be useful. But until there is a properly functioning blitter without added artefacts there is absolutely no way of saying whether it would be faster or better than the current blitters; it's often the corner cases that take the most time.bokkie wrote:But OpenTTD developers have experimented with this F1 engine (http://wiki.openttd.org/wiki/index.php/Blitter) so what's the reason behind that? (Your reply seems to indicate that it's useless).
I dare to say that these are heavily populated maps with a lot of rails, trains etc, that suffer most from blitter performance. If I load such game on my machine, performance in paused mode is quite similiar to that in unpaused mode.Furthermore previous attempts to do the actual blitting have shown that the really blitting doesn't take much CPU; it's a few percent faster on multi[thread|core] CPUs and more than it is faster percent slower on single core/thread CPUs. Meaning that the blitter doesn't give most people an instant performance increase of significant magnitudes. This analysis has been made on a quite populated, but not 100% CPU game, however... one can always make their statistics meet their expectations by e.g. using an unpopulated map with a few moving industries and max-zoom-out on fast-forward, but is that a realistic scenario?
Why are you so sure that is due to blitter performance? Have you profiled openttd with the result of the blitter taking most of the cpu time or is it just a guess? In the latter case my guess is that the problem is not because of the blitter, but due to the pahtfinders, tile loop, etc.CommanderZ wrote:I dare to say that these are heavily populated maps with a lot of rails, trains etc, that suffer most from blitter performance. If I load such game on my machine, performance in paused mode is quite similiar to that in unpaused mode.Furthermore previous attempts to do the actual blitting have shown that the really blitting doesn't take much CPU; it's a few percent faster on multi[thread|core] CPUs and more than it is faster percent slower on single core/thread CPUs. Meaning that the blitter doesn't give most people an instant performance increase of significant magnitudes. This analysis has been made on a quite populated, but not 100% CPU game, however... one can always make their statistics meet their expectations by e.g. using an unpopulated map with a few moving industries and max-zoom-out on fast-forward, but is that a realistic scenario?
If the pathfinders are causing slowdown in paused mode then openttd has bigger problems.Yexo wrote:Why are you so sure that is due to blitter performance? Have you profiled openttd with the result of the blitter taking most of the cpu time or is it just a guess? In the latter case my guess is that the problem is not because of the blitter, but due to the pahtfinders, tile loop, etc.CommanderZ wrote:I dare to say that these are heavily populated maps with a lot of rails, trains etc, that suffer most from blitter performance. If I load such game on my machine, performance in paused mode is quite similiar to that in unpaused mode.
Indeed. I should read better next timeThief^ wrote:If the pathfinders are causing slowdown in paused mode then openttd has bigger problems.
Wouldn't GL_ARB_texture_non_power_of_two help (if of course the partial updates were sorted)?petern wrote:..."non-powers of 2" sized sprites.