Please look at the program performance comparison between Cypress S29GL-T and Micron M29EW (if you use another part, please refer to its datasheet).
S29GL-T: 1.76 us / word when you use 512-byte buffer.
M29EW: 1.76 us / word when you use 1024-byte (512-word) buffer.
Thus, the program performance is equivalent according to the datasheet.
The performance values in datasheets do not contain command / data input cycles overhead because it depends on the host system's specification / configuration. To program 1024 byte, you need to perform 512-byte write buffer program sequence two times for S29GL-T. That means an extra command cycles overhead is added. The amount of the overhead depends on your system's configuration, however in typical systems, it should not be significant comparing program time like 450us (or 900us) per 512-byte (or 1024-byte).