Glossary

Here is an alphabetical list of terms with links to where they are explained in the text.

*

• Local** refers to the physical machine you are working on, i.e. your desktop or laptop.

`

• LMOD modules

A

• address space
• anti-pattern
• array of structures

B

• bandwidth limited
• bandwith saturated

C

• Cache coherence
• cache coherent non-uniform memory architecture
• cache line
• cache-oblivious
• ccNUMA
• cell-based Verlet list construction
• code modernization
• Code optimization
• Common sense optimizations
• communication
• computational complexity
• computational intensity
• compute limited
• compute node

D

• Debugging
• Distributed memory parallelization

E

• embarrassingly parallel

H

• hardware thread
• Hybrid memory parallelization

I

• instruction pipelining

L

• L1 cache
• L2 cache
• L3 cache
• loop fusion

M

• main memory
• memory bandwidth
• multi-core
• multi-processor

N

• node
• NUMA

P

• parallel program
• Partitioned Global Address Space
• pattern
• peak performance
• pipeline stalls
• process
• profiling

R

• registers

S

• sequential
• serial
• shared memory parallelization
• short-ranged
• SIMD vectorisation
• software thread
• Spatial locality
• structure of Arrays

T

• Temporal locality
• thread
• tiling
• time to solution

V

• Verlet list