2013-05-12, 19:38
I'm trying to identify a replacement for the old trusty Acer Revo r1600u ion nettop, and it occurs to me that there is probably a better way to assess XBMC options than anecdotal reviews or spec sheets.
For example, since I'm looking to replace the Revo that currently powers an only 720p capable projector, I'm sure I don't need something with AMD APU/8gb DDR3/SSD specs, which I use to power a 1080p flatscreen.
Like many I'm interested in the new crop of android TV devices, whether the quad core sticks, or the appleTV style boxes. But there isn't much objective data showing, for example, the framerate one gets on an RK3188 stick when playing back (as another example) a 1080p encoded clip of the migrating bird panorama from "Planet Earth", as compared to the same clip pushed to the same display on a Revo, an APU, Pivos, etc.
I'm willing to kick off that kind of testing with my various devices (the AMD APU system, Revo Ion, an MK802, Core I7 laptop, on 720p and 1080p displays etc) but it makes sense to establish a methodology that others agree to be worthwhile, and that people would be willing to actively add to in order to create something like a usable knowledge base. There will always be variables, many variables, but in the same way that dozens of sites put new hardware components through their in house benchmark routines, the deltas can still show differences when parsed from aggregate.
There are of course plenty of testing benchmarks for x86 cpu performance, video card performance, Android specific test suites, etc, but they wouldn't seem to be as valuable as something specific to XBMC.
I'm thinking the way to go would be a selection of commonly used benchmark video clips, with some means of recording the FPS graph over time (XBMC can display framerate on screen, IIRC, but I don't know that it can be recorded and analyzed after the fact?)
Any interest in this sort of thing?
For example, since I'm looking to replace the Revo that currently powers an only 720p capable projector, I'm sure I don't need something with AMD APU/8gb DDR3/SSD specs, which I use to power a 1080p flatscreen.
Like many I'm interested in the new crop of android TV devices, whether the quad core sticks, or the appleTV style boxes. But there isn't much objective data showing, for example, the framerate one gets on an RK3188 stick when playing back (as another example) a 1080p encoded clip of the migrating bird panorama from "Planet Earth", as compared to the same clip pushed to the same display on a Revo, an APU, Pivos, etc.
I'm willing to kick off that kind of testing with my various devices (the AMD APU system, Revo Ion, an MK802, Core I7 laptop, on 720p and 1080p displays etc) but it makes sense to establish a methodology that others agree to be worthwhile, and that people would be willing to actively add to in order to create something like a usable knowledge base. There will always be variables, many variables, but in the same way that dozens of sites put new hardware components through their in house benchmark routines, the deltas can still show differences when parsed from aggregate.
There are of course plenty of testing benchmarks for x86 cpu performance, video card performance, Android specific test suites, etc, but they wouldn't seem to be as valuable as something specific to XBMC.
I'm thinking the way to go would be a selection of commonly used benchmark video clips, with some means of recording the FPS graph over time (XBMC can display framerate on screen, IIRC, but I don't know that it can be recorded and analyzed after the fact?)
Any interest in this sort of thing?
