Cover of Code by Charles Petzold - Business and Economics Book

From "Code"

Author: Charles Petzold
Publisher: Microsoft Press
Year: 2000
Category: Computers

🎧 Free Preview Complete

You've listened to your free 10-minute preview.
Sign up free to continue listening to the full summary.

🎧 Listen to Summary

Free 10-min Preview
0:00
Speed:
10:00 free remaining
Chapter 21: Get on the Bus
Key Insight 2 from this chapter

Memory and Long-Term Storage Technologies

Key Insight

Random access memory (RAM) chips, like the 2102 static RAM (SRAM) from the mid-1970s, store data using address inputs, data inputs, data outputs, and a write signal. The 2102 chip stores 1024 bits and has a read access time of 350 to 1000 nanoseconds. To store 8-bit values for microprocessors, eight 2102 chips are wired in parallel, creating a 1024x8 RAM array (1 kilobyte). A 4-KB memory board might contain four such 'banks' of 8 chips, totaling 32 chips. Address signals A0-A9 connect to RAM chips, A10-A11 select banks, and A12-A15 determine the board's 4KB address range within a 64KB memory space, configurable using DIP switches and comparator circuits. A critical feature for bus operation is the 2102's 'tri-state output,' which allows its data output to enter a high-impedance state, effectively disconnecting it from the bus when deselected, ensuring only one device drives the data lines at a time.

SRAM (Static Random Access Memory) such as the 2102, retains its contents as long as power is supplied and typically uses 4 transistors per bit. Dynamic RAM (DRAM), however, requires only 1 transistor per bit but needs periodic 'refresh cycles' (hundreds of times per second) to prevent data loss, even with continuous power. Despite this complexity, DRAM has become the standard due to its higher density and increasing capacity over time, quadrupling approximately every three years since Intel introduced a 16384-bit DRAM chip in 1975. Modern computers use SIMMs or DIMMs containing multiple DRAM chips, with 128 megabyte DIMMs available for under $300 today. Since RAM is volatile, losing contents when power is off, computers require non-volatile 'long-term storage' like magnetic storage.

Magnetic storage evolved from Oberlin Smith's principles (1878) and Valdemar Poulsen's telegraphone (1898), which recorded sound on steel wire using an electromagnet ('head'). Fritz Pfleumer patented magnetic tape in 1928, using iron particles on paper (later cellulose acetate), becoming a popular recording medium. Remington Rand introduced the first commercial digital tape system in 1950, storing a few megabytes. Tape remains used for archiving but lacks fast random access. Magnetic disks, geometrically better for fast access, were first used for computer data by IBM's RAMAC in 1956, storing 5 megabytes on 50 2-foot metal disks. Disks became smaller and higher capacity, categorized into removable floppy disks (e.g., 3.5-inch storing 1440 kilobytes) and fixed hard disks (faster, higher capacity, e.g., 20 gigabyte drives for under $400 in 1999). Disk surfaces are organized into tracks and sectors (typically 512 bytes per sector). Hard disk interfaces like SCSI, ESDI, and IDE use DMA to transfer data directly between disk and RAM, bypassing the microprocessor. The terms 'memory' refer to volatile semiconductor RAM, while 'storage' refers to non-volatile media like disks and tape; the microprocessor directly addresses memory, but accessing storage requires additional program steps and DMA.

📚 Continue Your Learning Journey — No Payment Required

Access the complete Code summary with audio narration, key takeaways, and actionable insights from Charles Petzold.