Memories and the Memory Subsystem The Memory Hierarchy

Memories and the Memory Subsystem; The Memory Hierarchy; Caching; ROM

Memory: Some embedded systems require large amounts; others have small memory requirements Often must use a hierarchy of memory devices Memory allocation may be static or dynamic Main concerns [in embedded systems]: make sure allocation is safe minimize overhead

Memory types: RAM DRAM—asynchronous; needs refreshing SRAM—asynchronous; no refreshing Semistatic RAM SDRAM—synchronous DRAM ROM—read only memory PROM—one time EPROM—reprogram (uv light) EEPROM—electrical reprogramming FLASH—reprogram without removing from circuit

Standard memory configuration: Memory a “virtual array” Address decoder Signals: address, data, control fig_04_01

ROM—usually read-only (some are programmable-”firmware”) fig_04_02 transistor (0 or 1)

SRAM—similar to ROM In this example— 6 transistors per cell (compare to flipflop? ) fig_04_04

SRAM read: precharge bi and not bi to a value halfway between 0 and 1; word line drives bi’s to stored value SRAM write: R/W line is driven low fig_04_05

Dynamic RAM: only 1 transistor per cell READ causes transistor to discharge; it must be restored each time refresh cycle time determined by part specification fig_04_06

DRAM read and write timing: fig_04_08

Comparison—SRAM / DRAM fig_04_08

Memory: typical organization (SRAM or DRAM): fig_04_09

Two important time intervals: access time and cycle time fig_04_11

Terminology: Block: logical unit of transfer Page—logical unit; a collection of blocks Bandwidth—word transition rate on the I/O bus (memory can be organized in bits, bytes, words) Latency—time to access first word in a sequence Block access time—time to access entire block fig_04_12 Block size

Memory interface: Restrictions which must be dealt with: Size of RAM or ROM width of address and data I/O lines

Memory example: 4 K x 16 SRAM Uses 2 8 -bit SRAMs (to achieve desired word size) Address: 10 bits within a block 2 bits to specify block— CS (chip select) fig_04_14 Uses 4 1 K blocks (to achieve desired number of words)

Write: 8 -bit bus, two cycles per word fig_04_16

Read: choose upper or lower bits to put on the bus fig_04_17

If insufficient I/O lines: must multiplex signals and store in registers until data is accumulated (common in embedded system applications) Requires MAR / MDR configuration typically

DRAM: Variations available: EDO, SDRAM, FPM— basically DRAMs trying to accommodate ever faster processors Techniques: --synchronize DRAM to system clock --improve block accessing --allow pipelining As with SRAM, typically there are insufficient I/O pins and multiplexing must be used

Terminology: RAS—row address strobe CAS—column address strobe note either leading or trailing edge can capture address RAS cycle time RAS to CAS delay Refresh period fig_04_19

Example: EDO—extended data output: one row, multiple columns fig_04_20

Refreshing the DRAM: overhead; must refresh all rows, not just those accessed; Will this be controlled internally or externally? Example: refresh one row at a time, external refresh 4 M memory: 4 K rows, 1 K columns: 22 address bits 12 I/O pins, 10 shared between row and column Refresh each row every 64 ms 2 -phase clock (for greater flexibility), 50 MHz source fig_04_21

Refresh timing: must refresh one row every 16 musec Use a 9 -bit counter incremented from either phase of the 25 MHz clock, refresh at count 384 (15. 36 musec)—this provides some timing margin fig_04_22

Refresh address— 12 bit binary counter—increment following the completion of each row refresh fig_04_23

Address selection: Read / write refresh fig_04_24

Refresh arbitration: avoid R/W conflicts 1. If normal R/W operation starts, allow it to complete 2. If refresh operation has started, remember R/W operation 3. For a tie, normal R/W operation has priority Required signals: R/W—has been initiated Refresh interval—has elapsed Normal request—by arbitration logic Refresh grant—by arbitration logic Normal active—R/W has started Refresh active—refresh has started fig_04_25

Row and column addresses (column only 10 bits): Generate row, column addresses on phase 1, RAS, CAS on phase 2: fig_04_25

Arbitration circuit: request portion followed by grant portion fig_04_27

Complete system: fig_04_28

R/W cycles: fig_04_29

Memory organization: Typical “Memory map” fig_04_30 For power loss

Memory hierarchy fig_04_31

Paging / Caching Why it typically works: locality of reference (spatial/temporal) Note: in real-time embedded systems, behavior may be atypical; but caching may still be a useful technique fig_04_32 “working set”

Typical memory system with cache: hit rate (miss rate) important fig_04_33

Basic caching strategies: Direct-mapped Associative Block-set associative questions: what is “associative memory”? what is overhead? what is efficiency (hit rate)? is bigger cache better?