Semiconductor Memories. Jan M. Rabaey Anantha Chandrakasan Borivoje Nikolic Paolo Spirito

Transcription

1 Semiconductor Memories Jan M. Rabaey Anantha Chandrakasan Borivoje Nikolic Paolo Spirito

2 Memory Classification

3 Memory Classification Read-Write Memory Non-Volatile Read-Write Memory Read-Only Memory Random Access Non-Random Access EPROM E 2 PROM Mask-Programmed Programmable (PROM) SRAM DRAM FIFO LIFO Shift Register CAM FLASH

4 Memory Classification

5 Memory access time

6 Memory timing: definitions

7

8 RAM memory Architecture M bits M bits S 0 S 1 Word 0 Word 1 Word 2 Storage cell A 0 A 1 S 0 Word 0 Word 1 Word 2 Storage cell Word N 2 A K 1 Word N 2 S N-1 Word N 1 Word N 1 K = log 2 N S N 1 Input-Output M bits Intuitive architecture for N x M memory Too many select signals: N words => N select signals Input-Output M bits Decoder reduces the number of select signals K = log 2 N

9 Matrix memory organization

10 Matrix memory organization A K A K+1 A K+2 A L 1 A A A A A 2 L K M = 2 K Amplify swing to rail-to-rail amplitude A 0.. A K 1 Selects appropriate word La word line abilita tutte le celle sulla riga consumo di potenza La bit line permette di leggere/scrivere la cella desiderata

11 Hierarchical memory architecture P 1 Advantages: 1. Shorter wires within blocks 2. Block address activates only 1 block => power savings

12 Hierarchical memory architecture

13 Memory array architecture

14 Power dissipation in memories

15

16 Memory organization

17 6-Transistor CMOS SRAM Cell

18 6-Transistor CMOS SRAM Cell WL V DD M 2 M 4 Q Q M 6 M 5 M 1 M 3 BL BL

19 SRAM Write operation: ex. Q = 1 WL V DD M 4 Q = 0 M 6 M 5 Q = 1 M 1 V DD BL = 1 BL = 0 IP: Q=1 WL=1 to activate M5 e M6 BL=0 M6 is ON Q goes to 0 since M4 turn OFF during the transient BL=1 M6 is ON Q goes to 0 since M1 turn OFF during the transient

20 SRAM Write operation

21 SRAM Read operation: ex. Q = 1 WL V DD BL M 4 Q = 0 Q = 1 M 6 M 5 BL V DD M 1 V DD V DD C bit C bit

22 Resistive-load SRAM cell WL V DD R L R L M 3 Q Q M 4 BL M 1 M 2 BL Static power dissipation: want R L large Bit lines precharged to V DD to address t p problem

23 SRAM Characteristic

24 Sense amplifiers small transition s.a. input output Il sense amplifier è un circuito in grado di amplificare le variazioni presenti sul segnale di ingresso un piccolo sbilanciamento delle tensioni su BL e BL viene amplificato.

25 Memory Classification 2 L K A K A K+1 A K+2 A L 1 A A A A A M = 2 K Amplify swing to rail-to-rail amplitude A 0.. A K 1 Selects appropriate word

26 Transposed-bit-line architecture BL C cross BL BL SA BL (a) Straightforward bit-line routing C cross BL k-1 BL k SA BL k BL k+1 (b) Transposed bit-line architecture

27

28 1-Transistor DRAM Cell X Write: C S is charged or discharged by asserting WL and BL. Read: Charge redistribution takes places between bit line and storage capacitance ΔV = V BL V PRE = (V X V PRE ) C S C S + C BL Voltage swing is small; typically around 250 mv.

29 Read operation

30 Write operation Avrò in transitorio la cui durata dipende dalla resistività di M1

31 Advanced 1-T DRAM cell Word line Insulating Layer Cell plate Capacitor dielectric layer Cell Plate Si Refilling Poly Transfer gate Storage electrode Isolation Si Substrate 2nd Field Oxide Trench Cell Stacked-capacitor Cell

32 Future DRAM Technology

33 1-Transistor DRAM Cell 1T DRAM requires a sense amplifier for each bit line, due to charge redistribution read-out. DRAM memory cells are single ended in contrast to SRAM cells. The read-out of the 1T DRAM cell is destructive refresh is necessary for correct operation. The 1T cell requires presence of a capacitance that must be explicitly included in the design. When writing a 1 into a DRAM cell, a threshold voltage is lost (M1 in pass transistor configuration) this loss can be circumvented by boot-strapping the word lines to a value higher than VDD

34 Noise sources in 1-T DRAM C WBL BL substrate Adjacent BL WL leakage C S electrode C cross

35 Read-Write RAM

36 Perifery Decoders Sense Amplifiers Input/Output Buffers Control & Timing Circuitry

37 8-bit AND Decoder Collection of 2 M complex logic gates organized in regular and dense fashion WL 0 = A 7 A 6 A 5 A 4 A 3 A 2 A 1 A 0 WL 1 = A 7 A 6 A 5 A 4 A 3 A 2 A 1 A 0 WL 255 = A 7 A 6 A 5 A 4 A 3 A 2 A 1 A 0 1 NAND with 8 inputs and 1 NOT very big area and very slow 2 stages Decoder

38 10-bit NOR Decoder WL 0 = A 7 +A 6 +A 5 +A 4 +A 3 +A 2 +A 1 +A 0 WL 1 = A 7 +A 6 +A 5 +A 4 +A 3 +A 2 +A 1 +A 0 WL 255 = A 7 +A 6 +A 5 +A 4 +A 3 +A 2 +A 1 +A 0 1 NOR with 8 inputs (16 T) 1 stage Decoder Number T: 256 x 16 = 4096 T

39 Hierarchical decoders WL 0 = A 7 A 6 A 5 A 4 A 3 A 2 A 1 A 0 WL 0 = A 7 A 6 A 5 A 4 A 3 A 2 A 1 A 0 WL 0 = (A 7 +A 6 )(A 5 +A 4 )(A 3 +A 2 )(A 1 +A 0 ) The address is partitioned into sections of 2 bits that are decoded in advance. The resulting signal are then combined using 4-input NAND gates 1 4-input NAND 4 2-input NOR

40 Hierarchical decoders Multi-stage implementation improves performance WL 1 WL 0 A 0 A 1 A 0 A 1 A 0 A 1 A 0 A 1 A 2 A 3 A 2 A 3 A 2 A 3 A 2 A 3 NAND decoder using 2-input pre-decoders A 1 A 0 A 0 A 1 A 3 A 2 A 3 A 2 Number T: (256x8T)+(4 x 4 x 4T) = 2112 T 52% smaller & faster

41 Hierarchical decoders Multi-stage implementation improves performance WL 1 WL 0 A 0 A 1 A 0 A 1 A 0 A 1 A 0 A 1 A 2 A 3 A 2 A 3 A 2 A 3 A 2 A 3 NAND decoder using 2-input pre-decoders A 1 A 0 A 0 A 1 A 3 A 2 A 3 A 2 Number T: (256x8T)+(4 x 4 x 4T) = 2112 T 52% smaller & faster

42 Column decoder: pass-transistor based BL 0 BL 1 BL 2 BL 3 A 0 S 0 S 1 S 2 A 1 S 3 Advantages: speed (t pd does not add to overall memory access time) Only one extra transistor in signal path Disadvantage: Large transistor count D

43 4-to-1 Tree-based column decoder BL 0 BL 1 BL 2 BL 3 A 0 A 0 A 1 A 1 D Number of devices drastically reduced, but delay increases quadratically with # of sections and becomes prohibitive for large decoders. Solutions: i) buffers; ii) progressive sizing; iii) combination of tree and pass-transistor approaches