EFB to RAM is especially demanding in OpenGL – I don't exactly know why it's worse there (due to inherent issues with the API or just with how we're using it?), but (warning, not-completely-proven theories of mine ahead) on all backends EFB to RAM is bottlenecked by how fast your GPU can render a frame, how fast your CPU can do whatever it does with the frame, and, most importantly, how fast the frame can be shuffled between the CPU and GPU memory.
EFB to RAM is supposedly much faster on integrated GPUs (which makes sense, but I've found that it also depends on drivers since I've found it to be much less of a slowdown on Linux than on OS X), and AMD and Nvidia are both working on unified memory (e.g. the CUDA stuff Nvidia just announced today), but for now the only real thing to do about it is to overclock your CPU/GPU/RAM. (I don't actually know how effective each of those are, I don't happen to own anything overclockable.)
EFB to RAM is supposedly much faster on integrated GPUs (which makes sense, but I've found that it also depends on drivers since I've found it to be much less of a slowdown on Linux than on OS X), and AMD and Nvidia are both working on unified memory (e.g. the CUDA stuff Nvidia just announced today), but for now the only real thing to do about it is to overclock your CPU/GPU/RAM. (I don't actually know how effective each of those are, I don't happen to own anything overclockable.)
<@skid_au> fishing resort is still broken: http://i.imgur.com/dvPiQKg.png
<@neobrain> dafuq
<+JMC47> no dude, you're just holding the postcard upside down
----------------------------------------
<@Lioncash> pauldachz in charge of shitposting :^)
<@neobrain> dafuq
<+JMC47> no dude, you're just holding the postcard upside down
----------------------------------------
<@Lioncash> pauldachz in charge of shitposting :^)