Just to keep things straight ...

The 10 MHz rate is determined by the FIFO, which has a 100 ns cycle time, not by the USB. If the PC is busy, it might take a second or two, but who cares? The samples are stored in a local SRAM, or whatever, then, when the sample is complete, even if they're sampled at 500 MHz, they're saved and transferred asynchronously via the USB, which isn't asynchronous, but it is used asynchronously with respect to the sampling.

Full-speed, < 1.2 Mbps, USB is probably capable of 1 MHz transfers, but even if they're 1 KHz transfers, it has no impact on the sample rate.

The sequence, ideally, would be, (1) set up trigger specification via USB, 1 hz PC-to-sampler transfer rate is adequate, well, almost; (2) perform sampling task, whatever that involves, resulting in data stored in buffer memory; (3) transfer resulting data to PC for processing and display, at whatever rate the USB channel and very slow, awkward, PC will provide. There's no reason, BTW, that this has to be done under Windows, as DOS can do it.

RE