Trace the memory hierarchy outward from the CPU, cache memory → main memory → secondary storage, and explain each level's role and how the levels differ in speed and capacity.
Put the standard names of the DRAM used for main memory in generation order: DDR3 → DDR4 → DDR5.
Tell the memory modules DIMM and SO-DIMM apart by size and by the devices that use them.
Cache memory (キャッシュメモリ) sits between the CPU and main memory and is a fast, small memory that holds copies of frequently used data and instructions.
CPU reads
Level 1 cache
Level 2 cache
Main memory
If neither has it, it reads from main memory and also puts a copy in the cache for next time.
4 / 4
The CPU runs far faster than main memory, so reading main memory every time would leave the CPU waiting. Cache memory therefore holds the data most likely to be needed, raising the share of reads that find their target there (a hit) and cutting the wait. Because it is small, a full cache has to evict something, such as the data that has gone unused the longest (the LRU method).
AnalogyLike a kitchen where any seasoning you just used stays next to the stove, and when that space fills up, the one left unused the longest goes back to the shelf at the back.
How to rememberThe English word cache means a hidden store or stash: a place to quietly keep supplies within reach.
Exam points
ExamCache memory is there to shorten, on average, the time the CPU spends waiting for main memory (主記憶) and so raise processing efficiency. Its capacity is smaller than that of main memory.
ExamCaches are stacked in levels by distance from the CPU: level 1 (1 次キャッシュ), then level 2. A read checks level 1 first, then level 2, and goes to main memory only if both miss.
ExamCache memory is built from SRAM, which is faster than the DRAM used for main memory. Its access time (アクセス時間) is shorter than that of main memory and of every level below it.
Main memory (主記憶) holds the programs and data currently running and is the central memory the CPU reads and writes directly.
Each storage location in main memory has a number called an address, and the CPU specifies an address to read or write instructions and data. It is built mainly from DRAM, which must be rewritten periodically (refresh) to keep what it stores. Main memory is volatile and loses its contents when power is cut, so data you want to keep is saved to secondary storage.
AnalogyA workbench where you spread out the papers you are working on: the bigger the bench, the more papers you can have open at once.
Exam points
ExamMain memory mainly uses DRAM. DRAM keeps its contents through refresh (リフレッシュ) and is volatile (揮発性) memory that loses them when power is cut.
ExamThe CPU fetches the instructions and data it needs by specifying an address (アドレス), the number assigned to each location in main memory.
Secondary storage (補助記憶) is non-volatile, large-capacity storage that keeps its contents when power is off and holds programs and data for the long term.
Devices such as HDDs, SSDs and optical discs belong here, and to run a program its contents are first loaded into main memory. The memory hierarchy puts the register, which holds values being calculated inside the CPU, at the top and runs register → cache memory → main memory → secondary storage, with each lower level slower but cheaper per unit of capacity. Secondary storage is the slowest level of all, yet it has the largest capacity.
AnalogyA filing cabinet: getting papers in and out takes time, but it holds a lot, and its contents stay put even when you switch off the room's lights.
How to rememberOn a PC spec sheet, "Memory 16GB" usually means main memory, and "Storage 512GB" usually means the capacity of secondary storage.
Exam points
ExamIn the memory hierarchy (記憶階層), levels closer to the CPU read and write faster, and secondary storage at the bottom is the slowest. The registers (レジスタ) inside the CPU sit at the very top.
ExamAn SSD is secondary storage that records data in flash memory (フラッシュメモリ); it reads and writes electrically and has no mechanism for spinning a recording medium.
ExamAn HDD is secondary storage that reads and writes by spinning a magnetic disk (磁気ディスク), and it supports random access (ランダムアクセス) by moving the head straight to the position to be read.
DDR3 SDRAM is a DRAM standard used for main memory and the third generation of the DDR SDRAM series.
SDRAM is DRAM that works in sync with a clock signal that sets the timing of operations. DDR (Double Data Rate) sends data on both the rising and falling edges of the clock, transferring twice per cycle, and DDR3 is its third generation.
AnalogyIf you take a step on both the tick and the tock of a metronome, you cover twice the steps at the same tempo.
How to rememberThe Double in DDR means "twice per beat". The number after it is the generation, not a capacity or a module count.
DDR4 SDRAM is the successor to DDR3 SDRAM and the fourth-generation DRAM standard of the DDR SDRAM series.
It transfers data faster than DDR3 and lowers the supply voltage to cut power consumption. Its module has a different pin count and a different notch position from DDR3, so it cannot be inserted into a DDR3 slot.
DDR5 SDRAM is the successor to DDR4 SDRAM and the fifth-generation DRAM standard of the DDR SDRAM series.
Generation
Standard supply voltage
Relation to the previous generation
DDR3
1.5V
Successor to DDR2
DDR4
1.2V
Faster and lower-voltage than DDR3
DDR5
1.1V
Even faster and lower-voltage than DDR4
It pushes transfer speed beyond DDR4 and lowers the supply voltage again. Its internal design was also reworked, for example by putting the power-management circuit on the module itself.
A DIMM is a memory module standard that mounts several DRAM chips on a small board, which plugs into a PC's memory slot to serve as main memory.
← Scroll the figure sideways →
The name stands for Dual Inline Memory Module and comes from the pins on the front and back of the board each carrying separate signals. DIMMs are used mainly in desktop PCs and servers, and each DRAM generation has its own dedicated shape.
AnalogyIf DRAM chips are the parts, a DIMM is the cartridge that lines them up so they can be plugged in.
How to rememberDual means "both sides". Its predecessor, the SIMM (Single), had pins on the front and back wired to the same signals.
Exam points
ExamMemory modules (メモリモジュール) such as DIMMs are installed by plugging them into a memory slot (メモリスロット) on the motherboard (マザーボード).
Mix-upNames like DDR3 and DDR4 give the generation of the DRAM on board, while names like DIMM and SO-DIMM give the module's form-factor standard; the two are separate axes.
SO-DIMM is a smaller memory module standard derived from the DIMM, used mainly as main memory in notebook PCs and small devices.
Its board is shorter and its pin count differs from a DIMM's, so even with the same DRAM generation it cannot go into a desktop DIMM slot. It also appears in small desktop PCs and other machines meant for places with limited space.
How to rememberSO stands for Small Outline, meaning a DIMM with a small outline.
DIMM vs. SO-DIMM
Aspect
DIMM
SO-DIMM
Key differenceBoard size
Standard size
Short and compact
Main devices
Desktop PCs and servers
Notebook PCs and small devices
Key points
Cache memory bridges the speed gap between the CPU and main memory and cuts the time the CPU waits for main memory.
Secondary storage keeps its contents without power and is the slowest but largest level of the memory hierarchy.
Main memory is where the CPU reads and writes directly by specifying addresses, and it is built mainly from DRAM.
The number after DDR is the generation: from DDR3 to DDR4 to DDR5, each newer one transfers faster at a lower supply voltage.
A memory module is a board carrying several DRAM chips, and plugging it into a motherboard memory slot adds main memory.