64-bit computing

64-bit computing describes a specific type of computer architecture. It is defined by a word size of 64 bits. This size influences the central processing unit (CPU), the memory, and the software. In these systems, the registers are 64 bits wide. A register is a tiny, fast storage area inside the processor. These registers hold data units, such as integers or memory addresses.

Using 64-bit architecture allows a computer to handle much larger amounts of information at once. It also provides a massive capacity for addressing memory.

The mechanism of 64-bit computing relies on how the processor manages data and addresses. The CPU uses its integer or address registers to point to locations in memory. These registers act like a map for the computer. Because the registers are 64 bits wide, they can store enormous numbers. A single 64-bit register can hold 2 to the 64th power of different values. This number is over 18 quintillion, or 1.8 x 10^19. When using an unsigned binary representation, the range is 0 through 18,446,744,073,709,551,615. If the computer uses two's complement, the range spans from negative 9,223,372,036,854,775,808 to positive 9,223,372,036,854,775,807. This massive range allows a processor to directly access 16 exabytes (EB) of byte-addressable memory.

Processor registers are organized into several distinct functional groups. Some registers are for integers, while others are for floating-point math. Floating-point registers handle numbers with decimal points. There are also SIMD (single instruction, multiple data) registers and control registers. Many designs also include special registers for address arithmetic, such as index or base registers. In modern designs, general-purpose integer registers often perform many of these roles. However, only integer or address registers can usually be used to find data in memory. This means the size of these specific registers limits the total addressable memory. Even if a computer has wider floating-point registers, the address registers set the limit.

History shows that 64-bit technology existed long before personal computers. Supercomputers have used 64-bit words since the 1970s. The Cray-1, released in 1975, was a famous 64-bit architecture. In the 1960s, the IBM System/360 was a 32-bit computer. It used 32-bit integer registers but only used 24 bits for addresses. This limited its address space to 16 MiB. By the mid-1980s, 32-bit microprocessors like the Intel 80386 became the standard. A 32-bit address register can only reference 4 GB of random-access memory (RAM). As memory became cheaper, computers needed more than 4 GB of space. This demand led companies like MIPS and DEC to develop 64-bit architectures for servers.

64-bit computing eventually moved from specialized machines to consumer products. In the mid-1990s, several companies developed 64-bit workstation systems. Interestingly, 64-bit microprocessors appeared in consumer electronics before PCs. The Nintendo 64 and the PlayStation 2 used 64-bit processors. In 2003, 64-bit computing reached the mainstream PC market. This happened when AMD released x86-64 processors. At the same time, Apple introduced the PowerPC G5. These developments allowed regular desktop users to access more powerful computing capabilities. Later, Intel released the Core 2 Duo to bring these advantages to mobile computers.

There are important technical differences in how different 64-bit systems handle addresses. While the architecture is called 64-bit, not all systems use the full 64 bits for virtual memory. For example, x86-64 and AArch64 only support 48 bits of virtual address. In these cases, the remaining 16 bits must be all zeros or all ones. Some instruction sets also support fewer than 64 bits of physical memory address. Additionally, a CPU might have external data buses that differ from its internal registers. For instance, the 32-bit Pentium actually had a 64-bit data bus. These variations show that "64-bit" can describe different parts of a single system.

Today, 64-bit computing is the standard for almost all modern technology. The capacity of 32-bit systems was quickly outpaced by real-world needs. In 2023, laptop computers commonly used 16 GB of memory. Servers often started with 64 GB or more. A 32-bit system simply cannot address this much memory. 64-bit computing connects to the broader growth of data science and high-performance computing. It provides the necessary headroom for massive databases and complex applications. Without this wide address space, modern memory capacities would be impossible to use.