# RAM and Memory Bandwidth

> What RAM holds, why it empties when the power goes off, and how much a computer needs. Why memory bandwidth can matter as much as size.

**Course:** [Computer Hardware](https://learn.benibix.com/learn/infrastructure-hardware)  
**Time:** 10 min

**After this lesson you can:**

- Say what RAM holds and what it does not
- Pick a sensible amount of RAM
- Explain what memory bandwidth limits

## What RAM does, and how fast it feeds the chip

RAM, or random access memory, is the fast memory a computer works from right now. Programs and their data are loaded into it from the drive, so the CPU can reach them quickly. It is much faster than a drive, but it is emptied when the power goes off. A laptop often has 8 to 32 gigabytes.

RAM is a desk and the drive is a filing cabinet. You take out the papers you need and spread them on the desk. When you go home, the desk is cleared, and only what went back in the cabinet is kept. That is why unsaved work is lost when the power cuts out.

When RAM fills up, the computer does not stop. It moves some of what is in memory out to the drive, to make room. Reading it back is far slower. So a machine with too little RAM feels slow and stutters, even with a fast chip.

Size is one number. The other is memory bandwidth: how much data can move from memory to the chip each second. It is counted in gigabytes a second. A desktop with two sticks of the newest memory moves about 90 gigabytes a second. A strong graphics card moves about 1,000.

Bandwidth matters when a job reads a lot of data again and again. Running an AI model is the clearest case. The chip reads every one of the model's numbers to write each word. If those numbers arrive slowly, the chip waits, however fast it is. So bandwidth often sets how fast the words come out.

Two memory sticks are usually faster than one stick of the same total size. Most boards read from two sticks side by side, which doubles the bandwidth. A laptop sold with one 16 gigabyte stick can be slower than one with two sticks of 8.

A common mistake is to think more RAM makes everything faster. It only helps until you have enough for what you run. Past that point, extra RAM sits empty. For most office work 16 gigabytes is enough. Video editing and AI models are where 32 or more pays off.

![A CPU joined to RAM by a thin pipe, and a GPU joined to its video memory by a thick pipe. The pipe width is the memory bandwidth.](https://learn.benibix.com/diagrams/hw-memory.svg)

### Example: Emma's laptop runs out of memory

Emma's laptop has 8 gigabytes of RAM. With 30 browser tabs, a spreadsheet and a video call open, it uses all 8 and starts moving data to the drive. Typing lags by a second.

She upgrades to two sticks of 16, for 32 gigabytes in all. The same set of apps uses 11 gigabytes, and the lag is gone.

**Words in this lesson:** [Random Access Memory (RAM)](https://learn.benibix.com/term/infrastructure/random-access-memory), [Memory Bandwidth](https://learn.benibix.com/term/infrastructure/memory-bandwidth)

**Question:** What happens to the data in RAM when the power goes off?

**Answer:** It is lost, unless it was saved to the drive first. RAM only holds data while it has power. Anything that matters has to be written to the drive, which keeps it when the power is off.

---

Source: https://learn.benibix.com/learn/infrastructure-hardware/ram-and-memory-bandwidth
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-19
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
