Memory

Question 1

what is the main memory trade-off

Answer 1

capacity vs cost per bit vs access time

Question 2

what are the 3 performance parameters

Answer 2

1. access time
2. memory cycle time
3. transfer rate

Question 3

what is the access time

Answer 3

time between presenting the address and recieving or storing the valid data

Question 4

what is the memory cycle time

Answer 4

access time + recovery time

Question 5

what is the transfer rate

Answer 5

the rate at which data can be moved

Question 6

what is a unit of transfer

Answer 6

number of data lines into and out of a memory module

Question 7

what is the internal unit of transfer

Answer 7

data bus width

Question 8

what is the external unit of transfer

Answer 8

block

Question 9

what is an addressable unit

Answer 9

smallest location which can be uniquely addressed

Question 10

how does sequentiual access access memory

Answer 10

starts at the beginning of the memory and accesses records in order

Question 11

what are the 3 ways to access memory

Answer 11

1. sequential
2. direct
3. random

Question 12

how is access time determined in sequential access

Answer 12

depends on the location of the data & the previous location
- highly variable

Question 13

how does direct access access memory

Answer 13

moves to a general region of memory then uses sequential access

Question 14

how is access time determined in direct access

Answer 14

• indiviual blocks have unique addresses
• still depends on location and previous location

Question 15

how does random access access memory

Answer 15

allows any selected memory location to be directly addressed and accessed

Question 16

how is access time determined in random access

Answer 16

• independent of location of previous access
• unique physical address identifies locations exactly

Question 17

what are the 3 types of storage

Answer 17

1. semiconductor
2. magnetic
3. optical

Question 18

4 physical characteristics of memory

Answer 18

1. information can decay naturally or be lost without electrical power
2. volatile/non-volatile
3. non-readable memory cannot be altered - ROM
4. power consumption - influences cost

Question 19

describe RAM

Answer 19

• read/write
• volatile - useful for temporary storage
• can be static or dynamic

Question 20

describe dynamic RAM

Answer 20

• periodically refreshes cells as charges can leak, even when powered
• simpler & smaller
• denser & v expense
• suited to large memory requirements
• requires a refresh option - chip disabled while refreshing

Question 21

describe static RAM

Answer 21

• cells store bits using the same on/off logic elements used in the processor, no charges to leak so no refreshing needed
• requires less circuitry
• faster
• suited for cache

Question 22

describe ROM

Answer 22

• provides permanent data which cannot be changed
• non-volatile

Question 23

what are the 2 types of memory errors

Answer 23

1. hard
2. soft

Question 24

what is a hard failure

Answer 24

permanent physical affects

Question 25

what is a soft failure

Answer 25

random events alter the stored information but no permanent damage

Question 26

how can memory failures be detected & corrected

Answer 26

1. parity bits 2. hamming codes

Question 27

describe parity bits

Answer 27

- to detect errors in data communication and processing, an additional bit is sometimes added to a binary code word to define its parity - a parity bit is the extra bit included to make the total number 1's in the resulting code word either even or odd

Question 28

how are parity bits generated & checked

Answer 28

- an even bit is generated at the sending end for all 7-bit ASCII characters - 8-bit characters including the parity bits are transmitted to their destination - Parity of each character is then checked at the receiving end - if the parity of the received character is not even (assuming even parity is used), at least one bit has changed its value during transmission

Question 29

pro of parity bits

Answer 29

uses only a single bit, resulting in the least overhead

Question 30

cons of parity bits

Answer 30

- if the number of changes is even, the error will not be detected - it can only detect 1, 3, or any odd number of errors in each character - parity does not indicate which bit contained the error - only detect - data must be discarded enitrely, and retransmitted from scratch

Question 31

describe hamming codes

Answer 31

- k parity bits are added to an n-bit data word, forming a new word of n + k bits - those positions numbered with powers of 2 are reserved for parity bits - code can be used with words of any length

Question 32

how is an additional parity bit used with hamming codes

Answer 32

- extra parity bit applies to all bits after the hamming code check bits have been added - then all single-bit, two-bit & three-bit errors can be detected

Question 33

how does an additional parity bit with hamming codes distinguish even-bit errors

Answer 33

- no parity error is detected - hamming code indicates there is an error - hamming codes are capable of correcting an error or detecting 2 but not capable of doing both simultaneously

Question 34

hamming code rule

Answer 34

d + p + 1 <= 2^p - where d = data bits, p = parity bits

Question 35

how are hamming codes checked

Answer 35

- when the 12 bits are read from emory, they are checked again for errors - C = C8C4C2C1 - C1 = XOR(1, 3, 5, 7, 9, 11) - C2 = XOR(2, 3, 6, 7, 10, 11) - C4 = XOR(4, 5, 6, 7, 12) - C8 = XOR(8, 9, 10, 11, 12) - if C = 0, there is no error detected - if C != 0, the 4 bit vinary number formed by the check bits gives the postion of the erroneous bit if only a single bit is in error

Question 36

what causes hard failures

Answer 36

- harsh environment - manufacturing defect - wear & tear

Question 37

what causes soft failures

Answer 37

- power supply issues - alpha particles

Question 38

draw a venn diagram for 4-bits and 3-parity bits

Question 39

what are syndromes

Answer 39

partity bits XOR with their own bits - i.e - Pa XOR b2 XOR b1 XOR b0

Question 40

if SEC is 0 this indicates

Answer 40

no error

Question 41

if SEC is 1 this indicates

Answer 41

error in check bit

Question 42

if SEC is >= 2 this indicates

Answer 42

error in data bit

Question 43

what is the formula for the number of bits needed for SEC codes

Answer 43

2^k - 1 >= M + k where: - M = num data bits - k = num parity bits