Basic Terminology

    From Laptop Knowledge Database

    General Processing[edit]

    Clock speed[edit]

    How many times the component cycles per second, completing instructions with every cycle. If a CPU has a faster clock speed, it can perform more tasks within one second.

    IPC: Instructions Per Cycle[edit]

    How many instructions a component can complete within one cycle. If two CPUs have the same clock speed, the one which can perform more instructions per cycle can perform more instructions overall, leading to better performance.

    Power Limit[edit]

    how many watts the component is allowed to use. For example, a CPU with a 45w power limit will be limited to 45w of power max.


    The amount of power any given component is consuming. This is important to consider when analyzing thermals. Power consumption to heat output varies between microarchitectures/processes/etc (Renoir 7nm won't have the same exact thermal characteristics as Intel IceLake 10nm or Cometlake 14nm) so keep that in mind.

    Higher wattages are better than lower wattages on the same chip. The higher the wattages you can reach while maintaining sustainable temps the better. For example, 60w at 85C would be okay, but 60w at 100C and thermal throttling wouldn't be good.

    Generally speaking, any machine that can achieve and maintain 45w on a 45w tdp chip would be labeled acceptable. Anything closer to 50w or more would be great. Intel chips at 15w are okay, but devices that can maintain good temps at cTDP-up 25w are ideal. Renoir 25w is where the performance is at for the ulv processors. Meeting or exceeding tdp for the 35/45w tdp processors is great, since those are already industry leading in multicore performance for multiple reasons. Keep in mind, a machine is only really sustaining these wattages if the GPU is also under full load. This can be ignored for devices without dGPUs, but it's worth considering for individuals interested in gaming on UMA machines.


    CPU: Central Processing Unit[edit]

    The brain of the computer. The CPU executes tasks given to it. For example if you want to render a scene in a game, the CPU will look at the instructions given to it (by the game code) and then tell the GPU what to render, like telling the GPU to display certain objects on a part of the screen.

    CPU Cores[edit]

    Basically how many "parts" there are in a CPU. If a CPU has 4 cores, you can sorta think of it as having 4 different brains. Each core can do a different task. This is beneficial because instead of giving the CPU an instruction and waiting for it to do each instruction one at a time, you can simply give the CPU multiple instructions and not have to wait for tasks to be done one at a time. More cores can do many tasks faster than less cores, but this depends on the program. If the program is only designed to send instructions to one core, then the rest of the cores will be chilling and not be effective.

    SMT: Simultaneous MultiThreading (or Hyperthreading for Intel CPUs)[edit]

    You may see this term sometimes. In essence, it is splitting the workload of the core into two parts (typically only 2). So if you have 4 cores, you may see 8 threads. This works by basically using up the remaining bit of core power left to perform another task. So each core will switch its attention between two tasks rapidly. This is possible because the core often waits for instructions on what to do, so giving it two tasks will give the core something to do instead of twiddling its thumbs while waiting for the next instruction


    GPU: Graphics Processing Unit[edit]

    Renders what you see on the screen. If you play a beautiful game with lots of details, this will increase the load on the GPU, because the GPU must render each part of the scene. It is easier to render a cube than a complex rounded shape with lots of different shadows, as an example.


    RAM: Random access memory[edit]

    For computers, this is extremely fast volatile (loses data without constant power) memory which stores data currently in use. This can be thought of as the short-term/working memory of a human brain, which stores data on the current task. This is why multiple chrome tabs use more ram; because each tab must be stored in memory so the page doesn't need to be refreshed every time you switch tabs.

    Dual channel RAM: two sticks of ram working at the same time to store memory. If single channel is like having one person carrying packages back and forth between two locations, dual channel ram is like having two people. This means that the CPU does not have to wait as long for data from the RAM to arrive, improving performance for memory sensitive programs.