Talk:Image noise
What are we supposed to be doing here?
Are we supposed to be finding good hardware and tuning the random number generator to jack up fps? --Rdm 11:57, 2 October 2010 (UTC)
- It could be hard to get a constructive use of the FPS here, even if dividing the examples in to compiled/interpreted, Intel/AMD/etc., GPU X/Y/Z… Still the focus would be more on performance/hardware than the languages. I would like to suggest that the examples better fix the FPS a low value (1 FPS?) which should be of for most – no matter hardware. And then also, if possible, includes this 1-sec synchronization. --<Jofur> 12:51, 2 October 2010 (UTC)
- I get better than 1 fps with (by modern standards) modest hardware running Tcl. I'm not putting in my figures though; they're highly hardware-dependent. –Donal Fellows 16:04, 2 October 2010 (UTC)
Hardware information is not very important. It can be removed. It's just a reference. Improving FPS is always good. 30 FPS is enough to create a smooth animation. Any example running more than 30 FPS in most hardware is good, hence the optimization. --Guga360 16:14, 2 October 2010 (UTC)
- But why is that information requested anyway? Inter-language performance differences have never been the point of RC… –Donal Fellows 16:19, 2 October 2010 (UTC)
- OK. That's true. FPS is very hardware-dependent. We could test all examples in only 1 computer. Perfomance differences are not really point of RC. But we can still compare language perfomance, I think it's a good idea. --Guga360 16:34, 2 October 2010 (UTC)
- I still think the FPS should be rethought; Do you really want to set up the test on one computer; will then install 12 different C-compilers, LabView, etc.? Learning all compiler flags, libraries, etc?
- For me RC is more about the languages and in this task compiled languages that fits nicely into the CPU’s cache (ex. PureBasic, C++, etc.) will excel, while in a task where we communicate with a web-server and spend ~99.99% of the time waiting all languages will be equal fast. If performance should be a part of RC it should in some way be standardized before so that we have one common way to compare it – bad data is often much worse the no data. --<Jofur> 16:56, 2 October 2010 (UTC)
- OK. That's true. FPS is very hardware-dependent. We could test all examples in only 1 computer. Perfomance differences are not really point of RC. But we can still compare language perfomance, I think it's a good idea. --Guga360 16:34, 2 October 2010 (UTC)