Does RandomX use a lot of memory?

Yes. RandomX, the system Monero uses for mining, deliberately uses a lot of memory: its fast mode needs a 2,080 MB dataset, plus additional working memory. For you, that means roughly 2–3 GB of free RAM is a sensible starting point for the mining software alone—not for your whole computer.

What uses all that memory?

RandomX makes your CPU, your computer’s main processor, work with a large collection of data while it tries to earn mining rewards. This collection is called a dataset.

Fast mode uses two main memory structures:

• Dataset: 2,080 MB. This is the large pool of data used during mining.
• Cache: 256 MB. This smaller collection helps build the dataset and is also used by light mode.

These numbers are not a complete budget for your computer. The miner needs some additional working space. Your operating system, browser, and other apps need RAM too. RAM is your computer’s temporary working memory.

The memory use is intentional. It helps RandomX favor ordinary CPUs and makes it harder to gain a huge advantage with specialized mining chips.

More mining threads do not each need 2 GB

A mining thread is one stream of mining work running on your CPU. You can often run several at once.

In a typical native miner—a mining program installed directly on your computer—threads share the large dataset. So eight threads do not normally require eight separate 2,080 MB datasets.

Each thread does need its own small working area. CPU cache matters here too. L3 cache is a small, very fast memory area built into the processor. A useful RandomX guideline is about 2 MB of L3 cache per mining thread.

That is different from ordinary RAM. Adding more RAM does not increase your CPU’s L3 cache. If you run too many threads, they can compete for that fast memory, so using every available thread may not give you the best speed.

Light mode saves memory, but sacrifices speed

RandomX also has a light mode, which uses the 256 MB cache instead of keeping the full dataset in memory.

That sounds attractive on a low-memory device, but there is a trade-off: it has to do extra computation to replace missing dataset lookups. This makes it much slower for mining.

Light mode is useful when keeping memory use down matters more than mining speed, such as checking mining results. It is generally not the better choice if you want efficient mining.

What this means in your browser

Browser mining still faces RandomX’s memory demands. Running inside a webpage does not make the dataset smaller.

Your actual experience depends on how the browser miner is built. Memory sharing, browser limits, and other open tabs can all affect whether it runs smoothly.

Watch for:

• Tabs freezing or closing.
• Your computer becoming unusually sluggish.
• Heavy disk activity as your system moves data out of crowded RAM.

Keep spare memory available and stop mining if your device becomes unstable. More RAM helps when memory is tight, but extra unused RAM alone will not make mining faster.

The bottom line

RandomX is memory-heavy by design. Plan around its 2,080 MB fast-mode dataset, leave room for your other software, and remember that CPU cache matters alongside RAM. Light mode uses less memory, but it is much slower.