Older mini and mainframe computers generally operated on a line-at-a-time or screen-at-a-time basis. The line was sent to the computer by pressing the return, enter, send or transmit key, depending on what the key was called.
Editing of the typed-in line was done on the screen via the delete or backspace key, and the entire line was transmitted to the computer when the send key was pressed. Mainframe computers such as IBM's 370 and the System 38 used terminals such as the IBM 3270 which transmitted an entire screen of information all at once as opposed to a line at a time.
Mini and mainframe computers tended to serve many users at one time, and using line-at-a-time or screen-at-a-time input is more efficient for the computer. Most of the activity of a user of one of these machines is either reading output sent to them or waiting for input to be sent back to the running program, and thus other users could be served while the specific user was composing input. Terminals connected to these machines had limited editing capability, and thus were sometimes referred to as dumb terminals.
Minicomputers such as the PDP-11, and microcomputers ("personal computers") such as the TRS-80, Apple II and IBM-PC were able to implement delimiterless input as part of their design specification. Microcomputers, generally having only one user connected to them, could afford to support the less efficient delimiterless input capability as keyboard input could be handled by less expensive circuitry than that used on more expensive machines. This would then lead to the development of much more powerful games for personal computers. While games existed even on some of the largest computers, most of them were written, same as any other program, to either operate on line-at-a-time or screen-at-a-time input.