Programming languages cont.

Classification of programming languages:

CPU cores

To make CPUs faster, initial method was to increase frequency. But after we reached the physical limits of frequency, turned to CPU cores - add "more CPUs"

A single CPU has multiple cores

A CPU core is an independent instruction execution unit within the CPU

Think of CPU as having little sub-CPUs - those are its cores

Each core has its own registers, instruction pipeline, etc.

Terminology - dual-core, quad-core, octa-core...

In our course by default, we discuss single-core CPUs

(end of "reminder" section, now we begin the performance section)

Performance

Things that impact performance:

• Hardware efficiency
  • CPU clock - impacts frequency. As the frequency with which the clock sends synchronization signals increases, the performance of the machine increases
    • But frequency is not everything - used to be, but is not the only important thing anymore
  • Also important - CPU architecture. Good CPU design decisions make the CPU faster
  • Number of CPU cores - more cores is typically better (but a given program will only benefit from multi-core processing if it is written to take advantage of them)
  • Cache - speeds up average CPU-RAM interactions, making performance better
  • RAM - can be a bottleneck, so want fast RAM
  • Hard disk - also can be a bottleneck
    • Speed of hard disk defines how fast programs start execution
    • First step of program execution is to copy from hard disk to RAM
• Software efficiency
  • Program efficiency - more important than hardware!
  • Compiler efficiency - also important but performance differences between compilers are minimal compared to program efficiency improvements