So what does it matter if you find a Windows driver? Where you planning on using the Windows driver in your 8051 program, or exactly what was your point? Or did you hope to find a 8051 driver?

About Epson: "Serial and parallel ports are created into the LX-300+, and IBM 2380Plus emulation and ESC/P support are also offered for connectivity."
http://www.liutilities.com/device-dr...+ii-esc-p/

And the OKI printer I linked to emulates IBM 2480, Epson ESC/P and OKI MICROLINE.
http://www.facsimile.com/printers/...inter.html

What problem do you see with using the ESC/P protocol in the microcontroller?

Stop swaying all over the road.

Trying 45 Mbps synchronous transfers over a 12Mbps USB interface is not working. Bad design choice, and not relevant to the usefulnes of a PC.
Claiming multi-second to multi-minute hangs in USB transfers from a PC is something indicating a serious problem with your gear - not a limitatin built into a modern PC.
Traditional matrix printers with old-style protocols are existing at reasonable prices - windows drivers are irrelevant since you intended to drive with microcontroller.
There are a number of microcontrollers that can bridge from a PC and supply high-speed HDLC. Asking me to look for stand-alone HDLC chips is irrelevant to the usefulnes of a PC.
Linux can run on almost any PC, as long as you make sure you don't select hardware that the manufacturer have specifically kept the hw documentation secret for. Irrelevant to the usefulnes of a PC.
Any area where people once used a RS232 port on a PC now have multiple alternatives where people are using USB. And there are alternatives to USB-to-RS232 dongles.
Any area where people once used a Centronics port on a PC now have multiple alternatives where people are using USB. And there are alternatives to USB to get a Centronics port.
In the case where the latency of USB is hindering, the free processor core you get with most USB devices can take over the bit-toggling of individual bits.

How do you get something done, if you always spend time trying to do it the 1980 way, even when new products allows cheaper or quicker alternatives? The availability of ancient products will drop for the simple reason that people don't care to try to supply products that no one wants. Every new project should be started with an open mind - what components and circuit solutions (or complete modules) will optimize the project the most? Either by saving development time for small-series products, or cutting down on production costs for high-volume products. But that requires that you ask the question "what are the options" instead of claiming "there are no options".