How can a computer make decisions faster?
Image: MODIS Land Rapid Response Team, NASA GSFC, Public domain, via Wikimedia Commons
branch prediction does: guesses which way an if-statement will go to keep the pipeline full
How can a computer make decisions faster?
Imagine you're reading a mystery book and you have to decide whether to jump to the next chapter based on clues. Without guessing, you'd wait until the end of the chapter to see if the clues make sense, slowing down your reading.
A computer uses a branch predictor to guess whether to jump to the next chapter or continue reading, making the reading process faster by avoiding unnecessary stops.
Example
If the computer guesses correctly, it skips unnecessary chapters, just like guessing the next chapter in a mystery book.
Remember this
The branch predictor guesses the direction of a "branch" (like a decision point) to keep the computer's instruction pipeline moving smoothly.
Text adapted from Wikipedia, licensed under CC BY-SA 4.0.
Related concepts
ReLU and Leaky ReLU
Why do computers sometimes struggle with simple decisions?
Adam has bias correction: divides by (1-β^t) in early steps
Why do we sometimes need to fix mistakes in computer decisions?
Triton differs from CUDA
Why does a super-fast computer sometimes run slower than a regular one?
Overlapping subproblems
Ever calculated a huge Fibonacci sequence by hand?
Write the Bellman equation for reinforcement learning
Predicting tomorrow's weather with today's clues
most transformer operations are memory-bound, not compute-bound
Why do computers sometimes get tired?
Swipe through 100 ML concepts daily
Open Pocket Polymath