Intro:
Numbers:
binary (base 2), octal (base 8), decimal (base 10), hex (base
16):
Remember how to convert between these?
sign/magnitude, 1's complement, 2's complement:
S/M: first bit
sign, rest magnitude.
1's complement: if number is negative, invert all
bits.
2's complement: if number is negative, invert all bits and add 1.
Using n bits, how many numbers can be represented in 2's complement?
Total 2n numbers can be represented.
Half are positive,
half are negative.
But remember 0 is first positive number.
=>
-2n-1 to 2n-1 - 1
Fixed point: decimal point is fixed
Floating point: IEEE standard:
single precision: 1.8.23
sign: 1
bit, 1 for negative, 0 for positive numbers
exponent: 8 bits, bias of
127
significand: 23 bits, implied 1
value: -1sign *
(1.significand) * 2exp-127
special values:
Rounding: 4 IEEE options:
Overflow and underflow:
overflow: if both numbers are positive and
result is negative
underflow: if both numbers are negative and the result is
0 or positive.
saturating arithmetic: instead of rolling over on overflow or
underflow, the result is set to be the maximum or minimum possible
value.
Harddrive access time

Starting a program
Logical operations
Datapath, pipelining, super-scalar, out-of-order execution
System performance