Nice results. I think I have the same exact setup you have. I went with the dedicated card as well. I only did one test. I bet if I were to do it again I may be able to get in the 1,000s. I know that I wanted the 1TB because I heard it runs with 4 channels vs 2 on the lower GB models. I have heard the 1TB is the only way you can get these super fast read and write speeds.
That's correct. The 1TB uses a 4-lane PCIe channel, while the rest all use a 2-lane PCIe channel.
However, the 512GB is no slouch either (both my 27" iMac and 13" rMBP have the 512GB), which operates at 750MB/s read and 720MB/s write. The difference between 512GB and 1TB isn't noticeable unless you're doing I/O intensive stuff.
Meanwhile, the Samsung 256GB (SM0256F) is also very fast, almost on par with the 512GB. I have that in my 21.5" iMac and it runs at 720MB/s read and 670MB/s write. There's no noticeable difference either between the SM0256F and the 512GB and 1TB. Note that Samsung is the sole supplier for 512GB and 1TB.
There's also the SanDisk 256GB (SD0256F), which performs at around 700MB/s read and 550MB/s write. There is a slight noticeable difference between this and the SM0256F, where boot up times are perhaps 3 seconds slower on the SD0256F.
Meanwhile, the 128GB performs the slowest, with the SD0128F performing at 320MB/s write and 720MB/s read. I don't have the figures for the SM0128F though, but I heard that it's faster, as Samsung parts tend to be.