How to Enter Machine Code on the Sinclair ZX81

This article explores the manual methods used to input machine code into the Sinclair ZX81, a popular home computer from the early 1980s. It details the reliance on BASIC POKE commands for memory insertion and the USR function for execution, highlighting the technical constraints of the era. Readers will learn about memory addressing, hexadecimal conversion, and the precautions necessary to prevent system crashes during this delicate process.

The Challenge of the ZX81 Architecture

The Sinclair ZX81 was designed as an affordable entry point into computing, which meant it shipped with only 1KB of RAM and a built-in BASIC interpreter. Unlike modern systems or even some contemporaries, the ZX81 did not include a built-in machine code monitor or assembler. This lack of tooling meant that developers and hobbyists who wanted to write faster, more efficient programs in machine code had to rely on manual entry methods. The processor, a Zilog Z80, required specific hexadecimal values to be placed into exact memory addresses to function correctly, a task that had to be managed through the existing BASIC environment.

Using POKE Commands for Memory Insertion

The primary method for entering machine code was through the BASIC POKE command. This command allowed the user to write a specific byte value into a specific memory address. To load a routine, a user would need a listing of the machine code alongside its corresponding memory addresses. The process involved typing a series of lines into BASIC, such as POKE 16514, 254, repeatedly for every byte of the program. Because machine code routines could be hundreds of bytes long, this was a tedious and error-prone process. A single mistyped digit would corrupt the routine, often leading to unpredictable behavior or a system crash. Some users utilized paper tape readers or cassette interfaces to load pre-saved machine code blocks, but manual POKEing remained the fundamental method for learning and debugging small routines.

Executing Code with the USR Function

Once the machine code bytes were successfully placed into memory, the program needed a way to transfer control from the BASIC interpreter to the new routine. This was achieved using the BASIC USR function. The USR command required a specific argument, usually the starting memory address of the machine code routine. For example, if the code was loaded starting at address 16514, the user would type PRINT USR 16514 to execute it. The ZX81 would then jump to that memory location and begin processing the machine instructions directly. The routine would eventually need to return control back to BASIC, typically by executing a specific return instruction, otherwise, the computer would freeze and require a power cycle.

Risks and Best Practices

Entering machine code manually carried significant risks for the stability of the ZX81. The memory map was tightly packed, with system variables and the display file occupying specific regions. Writing machine code into protected areas could overwrite essential system data, causing the screen to garble or the machine to lock up entirely. To mitigate this, users had to carefully calculate safe memory locations, often found in the unused RAM above the BASIC program area. It was common practice to save the BASIC loader program to cassette before attempting to run the machine code, ensuring that the tedious work of POKEing the values did not have to be repeated if the system crashed. Despite the difficulty, mastering this process provided users with a deep understanding of the computer’s hardware and memory management.