It does have Optimus and the battery life is about where yours is - maybe 2.5 hours.
I'm running the nVidia chip exclusively, but when I get around to it, I'll probably switch, as I'm running an openbox desktop that just doesn't need that kind of graphics acceleration. I have yet to figure out a painless way of switching between the two, but that's probably because I haven't really worked on it much.
There's no current way to switch painlessly, but there is something that makes it slightly nicer than having to restart all of your programs. Check out Bumblebee [1]. It does it by running two X servers, one for the lower power system and another for the NVIDIA binary blob and plays with environment variables to get things working. Then it takes the NVIDIA server and copies the images over to the slower card. In a round about way it's what windows systems do for it, but not quite as sane and not quite as performant. There's been some work to create the sharing needed so that drivers can share the buffers directly and eventually there won't need to be the second X server. However this also depends on if that work either gets its license changed or NVIDIA GPLs the kernel bits to their driver.
Comments
It does have Optimus and the battery life is about where yours is - maybe 2.5 hours.
I'm running the nVidia chip exclusively, but when I get around to it, I'll probably switch, as I'm running an openbox desktop that just doesn't need that kind of graphics acceleration. I have yet to figure out a painless way of switching between the two, but that's probably because I haven't really worked on it much.
There's no current way to switch painlessly, but there is something that makes it slightly nicer than having to restart all of your programs. Check out Bumblebee [1]. It does it by running two X servers, one for the lower power system and another for the NVIDIA binary blob and plays with environment variables to get things working. Then it takes the NVIDIA server and copies the images over to the slower card. In a round about way it's what windows systems do for it, but not quite as sane and not quite as performant. There's been some work to create the sharing needed so that drivers can share the buffers directly and eventually there won't need to be the second X server. However this also depends on if that work either gets its license changed or NVIDIA GPLs the kernel bits to their driver.
[1] https://wiki.archlinux.org/index.php/Bumblebee
There's a bios setting for switching between the two