Notice the LEDs per strip is divided by 8, so those numbers are assuming you are feeding 8 strands, not a single strand with the other 7 inactive. I think you need tell it you have 8*8 columns as if you had 8 of these stacked up.
FastLED can do DMA but believe it also wants to feed 8 strands because it assumes you are wanting to maximise throughput.
Your speed question is tricky, since OctoWS2811 pixels take a fixed amount of time to update per pixel regardless of the software. The magic of the OctoWS library is that the code does not block while feeding the strand, and you can feed 8 pixels at a time if your pixels array allows. So your refresh rate is can be as high as 1/8 the time to refresh your total pixels with the next frame being prepped while the last frame is read out.
I think you are looking at 10ms refresh for the strand length, so you even with blocking code from another less clever library you could still get a reasonable refresh.