I know technically the integrated graphics support the features I need, I was just looking for some confirmation that people have had problem free usage with XBMC GPU acceleration. So thanks guys for the input, really good to know!
A 2100T is extremely low power for the performance it offers and should be a considerable boost over the e3200/e350 systems I've been using lately. I don't intend to use a dedicated GPU if the IGP can provide performance and quality that I like, fewer components brings down the power consumption, heat output, and bill of parts...so all things I like to focus on :p
The 2105 will be interesting as it should pack the HD 3000 graphics instead of the HD 2000, but if the 2000 is sufficient for XBMC then I'm probably not going to bother waiting. The 2100T has a 35w TDP as opposed to the 65w TDP of the rest in the class, and that is pretty impressive. The Core based Celeron 430 I had used prior to upgrading to the e3200 was a 35w part, and that was just a 1.8ghz single core. AMD has some solid 45w parts, including the 600e series Athlon II X4 parts, but given the load this system typically sees, I don't think I can really justify the 4 cores, so for lighter or single-threaded loads the sandybridge architecture is far superior.
As for the 2100T vs the 2100, yea there will be a notable jump in performance jumping from 2.5ghz to 3.1, but it will be sufficiently fast enough. The classic e3200/Athlon II X2 builds run things as smooth as can be really, so I don't think I need to pursue the maximum performance, rather I'd like to focus on performance per watt. Plus it will be an interesting comparison, taking a fast chip and not stressing it to hell and back might actually be more efficient than taking a slower chip and forcing it to work much harder. It won't beat atom or zacate, but it might just be the sweet spot I'm looking for between those and 65w mainstream chips.
*edit
It seems before I even got the chance to test my theories, xbit labs beat me to it! Check out their comparison
here. If xbmc can utilize quicksync then it could actually be a lower power platform than zacate or atom. Thanks to the advanced power gating idle power is actually lower as well. But under typical loads, atom and zacate are clear winners. Shouldn't be a big surprise, but it still manages to out perform everything in it's class without any real draw backs.
Also note some of the complaints about the gigabyte board, apparently it has rather shotty power delivery circuitry, lending to it's poor efficiency compared to better designed boards like the MSI e350 boards. Idle power on the MSI platform was as low as
7.3wcompared to 15.8w. 1080p power consumption was also shotty, 22.4 vs 14.9.
So NOT all e350 platforms are made equal. Cheap and inefficient components really can bring down all of AMDs hard work. Also note they used a standard atx psi, which is far less efficient at these loads so I guess these results may not be 100% accurate, but the difference can't be entirely based on that fact. Certainly I still think it's significant, and would be advising against the gigabyte on those grounds. Anyone looking to shave as much power as possible should be advised of this.