QCA Equations

# of Clock Cycles

CPI ⋅ # of Instructions

CPU time/Execution time

CPI ⋅ # of Instructions ⋅ Clock Cycle Time

CPU time/Execution time

(CPI ⋅ # of Instructions)/Clock Rate

Throughput

# of Instructions/CPU time

Speedup

Old/New

Speedup

1/((1 - f) + f/s)

Speedup

1/((1 - ∑f) + ∑(f/s))

Average CPI

∑(percentage of workload ⋅ CPI of workload)

Energy

Execution Time ⋅ Average Power

Response Time

task/time to complete task

Address bits

log₂(Memory Size in bytes)

Number of blocks

Memory Size/Block Size

Cache Size

Number of blocks ⋅ Block Size

Number of Sets

Total Cache Blocks/Associativity

Cache Size

Number of Sets ⋅ Associativity ⋅ Block Size

Total Cache Blocks

Number of Sets ⋅ Associativity

Address bits

Tag bits + Index bits + Block Offset bits

Block Offset bits

log₂(Block Size)

Index bits

log₂(Number of Sets)

Tag bits

Address bits - Block Offset bits - Index bits

Cache Size bits

Number of Cache blocks(block size ⋅ 8 + Tag bits + 1 aka valid bit

Hit Rate

Number of Hits/Total
Accesses

Miss Rate

1 - Hit Rate

AMAT

Hit Rate + (Miss Rate ⋅ Miss Penalty)

Execution Time for No pipeline

Number of Instruction ⋅ 5 aka Number of Pipeline Stages ⋅ clock cycle time

Execution Time for Pipeline

(Number of Instruction + 5 aka Number of Pipeline Stages - 1) ⋅ clock cycle time

Execution Time for Pipeline (average)

CPI ⋅ clock cycle time

Speedup

clock frequency/CPI

Additional CPI branch misprediction

Branch frequency * misprediction rate *misprediction penalty

effective CPI

Additional CPI branch misprediction + CPI