A computer built for this work carries two processors and two kinds of working memory. That is a strange thing to build. Nobody puts two engines in a car and asks the driver to guess which one to use. So it is worth asking why the duplication exists, because the two answers are not the same answer, and knowing them is the difference between guessing at a spec sheet and reading one.
Start with the processors. The difference between a CPU and a GPU is not that one is newer or better. It is what kind of work each one is shaped for.
A CPU has a small number of lanes, and each one is extremely fast. That design suits work where the second step needs the answer from the first. Opening a file, deciding what to do next, following instructions in order. Most of what a computer does all day looks like that, which is why no machine has ever stopped needing a CPU.
A GPU is built the opposite way. It has thousands of lanes, each slower than a CPU lane, all performing the same operation on different numbers at the same instant. That arrangement is useless for work with a queue in it. It is enormously powerful when nothing has to wait.
Running a model is the second kind. Producing a single word means multiplying an enormous quantity of numbers together, and almost none of those multiplications depends on the result of another. They can all happen at once. Deciding the color of every pixel on a screen has exactly that shape too, which is why hardware designed for video games turned out to be hardware for artificial intelligence. Nobody planned that. The problems happened to match.
Now the memory, where the reason is completely different.
The GPU does not read the file once and remember it. Every single word a model produces requires reading through the entire set of numbers again. That is not occasional access. It is a constant, enormous flow, repeated for every word on the screen.
VRAM sits physically on the card, a short distance from the GPU, connected by something very wide. RAM sits elsewhere in the machine, further away, connected by something much narrower. Both hold things while they are being worked on. Only one of them can deliver at the rate this particular job demands. When the connection cannot keep up, the result is a GPU with enormous computing power sitting idle, waiting for numbers to arrive (APXML, n.d.).
This explains the thing that surprises people who try to run a model on a machine that is slightly too small. It often starts anyway. When a file will not fit in VRAM, some of it can be left in RAM and fetched as needed, an arrangement the card can technically manage (BentoML, 2026). What it produces is not a small penalty. It is somewhere between five and thirty times slower, which in practice means a person asks a question and gives up before the answer finishes (CORSAIR, 2026).
The reverse is also true and is worth knowing. Once a file has been loaded into VRAM and fits there, the narrow connection to the rest of the machine stops mattering almost entirely, because nothing needs to cross it anymore (Glukhov, 2025). The whole job is happening on the card.
So the two answers are these. There are two processors because some work has a queue in it and some does not. There are two memories because the card can only run at full speed on numbers it can reach without leaving home.
Storage sits underneath both and answers a third question, which is where things live when nobody is using them. It is the largest of the three and by far the slowest, and nothing is ever computed there. A file on storage is a file waiting.
Which is why the number that decides everything is VRAM. Not because larger is better in some general sense, but because fitting and not fitting are two different machines. Below the line the work happens on the card at full speed. Above it, the card waits.
A Chromebook has one processor and one memory. Neither of them is the wrong quality. They are the wrong kind, for this one job, and they are perfectly good for every other job the machine was sold to do.
APXML. (n.d.). LLM memory bandwidth importance. https://apxml.com/courses/llm-model-sizes-hardware/chapter-3-model-size-hardware-connection/memory-bandwidth
BentoML. (2026, April 15). What is GPU memory and why it matters for LLM inference. https://www.bentoml.com/blog/what-is-gpu-memory-and-why-it-matters-for-llm-inference
CORSAIR. (2026, April 28). Memory for local LLMs: How much RAM do you need? (and when speed matters). https://www.corsair.com/us/en/explorer/diy-builder/how-tos/memory-for-local-llms-how-much-ram-do-you-need-and-when-speed-matters/
Glukhov, R. (2025, June 12). LLM performance and PCIe lanes: Key considerations. Medium. https://medium.com/@rosgluk/llm-performance-and-pcie-lanes-key-considerations-db789241367d
The words, in plain terms
A machine has two of each, and the reason is that holding something and working on something are two different jobs.
Read the picture across, not down. The top row is the two places a file can sit while it is being used. The bottom row is the two things that can do the arithmetic. The left column is in the machine itself. The right column is on the card.
Storage runs underneath both columns because it belongs to neither. Nothing is ever worked on in storage. A file there is a file waiting.
The words in the boxes are the words you will see on a spec sheet, in a store, and in every guide about this you ever read.
The words, in plain terms
This is one machine drawn twice, at two different moments.
On the left the file is in storage. It will survive the power going off, the machine sitting in a drawer, and a year passing. Nothing is happening to it.
On the right the same file is in VRAM. Now it can be worked on. It also disappears completely the instant the power is cut, and it has to be put there again the next time.
There is deliberately no arrow between the two halves, because nothing travelled. Both pictures are true of the same machine at different times. The file did not go on a journey. It changed address.
The words, in plain terms
Two pictures of the same machine.
The picture on the left is deliberately wrong. It shows what the word load sounds like: an arrow carries the file up out of storage into VRAM, and there is a dashed outline where the file used to be.
The picture on the right is what actually happens. The file stays in storage, at full strength. A copy of it appears in VRAM. Two addresses, one file, and one of them forgets when the power goes off.
Load means copy, not move. Placed here on purpose, in grey and dashed, so the correct picture beside it has something to be correct against.
The words, in plain terms
Every bar is drawn to the same scale, and the red block is the same file every time. That is what makes the four comparisons honest.
The question is the same for each machine and it is answered with subtraction. Is the number on the machine larger than 6.63, or is it not.
The first bar is different from the other three, and the difference matters. The Chromebook’s bar is RAM, the working memory in the machine itself. The other three bars are VRAM, working memory on a card. The Chromebook does not have a small card. It has no card at all, which is why its RAM is the only thing that could possibly hold the file, and why it comes up short by 2.63 gigabytes.
Notice what the question never needed. Not the speed of the machine, not the year it was made, not the price. One subtraction settled all four.
Three machines in the world can hold this file. None of them is in this room.
The words, in plain terms
Every bar is drawn to the same scale, and the red block is the same file every time. That is what makes the four comparisons honest.
The question is the same for each machine and it is answered with subtraction. Is the number on the machine larger than 6.63, or is it not.
The first bar is different from the other three, and the difference matters. The Chromebook’s bar is RAM, the working memory in the machine itself. The other three bars are VRAM, working memory on a card. The Chromebook does not have a small card. It has no card at all, which is why its RAM is the only thing that could possibly hold the file, and why it comes up short by 2.63 gigabytes.
Notice what the question never needed. Not the speed of the machine, not the year it was made, not the price. One subtraction settled all four.
Three machines in the world can hold this file. None of them is in this room.
The words, in plain terms
Every bar is drawn to the same scale, and the red block is the same file every time. That is what makes the four comparisons honest.
The question is the same for each machine and it is answered with subtraction. Is the number on the machine larger than 6.63, or is it not.
The first bar is different from the other three, and the difference matters. The Chromebook’s bar is RAM, the working memory in the machine itself. The other three bars are VRAM, working memory on a card. The Chromebook does not have a small card. It has no card at all, which is why its RAM is the only thing that could possibly hold the file, and why it comes up short by 2.63 gigabytes.
Notice what the question never needed. Not the speed of the machine, not the year it was made, not the price. One subtraction settled all four.
Three machines in the world can hold this file. None of them is in this room.
The words, in plain terms
Every bar is drawn to the same scale, and the red block is the same file every time. That is what makes the four comparisons honest.
The question is the same for each machine and it is answered with subtraction. Is the number on the machine larger than 6.63, or is it not.
The first bar is different from the other three, and the difference matters. The Chromebook’s bar is RAM, the working memory in the machine itself. The other three bars are VRAM, working memory on a card. The Chromebook does not have a small card. It has no card at all, which is why its RAM is the only thing that could possibly hold the file, and why it comes up short by 2.63 gigabytes.
Notice what the question never needed. Not the speed of the machine, not the year it was made, not the price. One subtraction settled all four.
Three machines in the world can hold this file. None of them is in this room.
The words, in plain terms
Every bar is drawn to the same scale, and the red block is the same file every time. That is what makes the four comparisons honest.
The question is the same for each machine and it is answered with subtraction. Is the number on the machine larger than 6.63, or is it not.
The first bar is different from the other three, and the difference matters. The Chromebook’s bar is RAM, the working memory in the machine itself. The other three bars are VRAM, working memory on a card. The Chromebook does not have a small card. It has no card at all, which is why its RAM is the only thing that could possibly hold the file, and why it comes up short by 2.63 gigabytes.
Notice what the question never needed. Not the speed of the machine, not the year it was made, not the price. One subtraction settled all four.
Three machines in the world can hold this file. None of them is in this room.
- APeriod 1
- BPeriod 2
- CPeriod 4
- ABecause one of them is a backup in case the other fails
- BBecause some work has steps that wait on each other and some does not
- CBecause a GPU cannot do anything a CPU can do
- DBecause the machine runs two programs at the same time
- AWork where the second step needs the answer from the first
- BWork that can all happen at the same instant
- CHolding files when the power is off
- DProducing images on a screen
- AWhen the work involves large numbers
- BWhen the machine is old
- CWhen the work has a queue in it and each step waits on the one before
- DWhen there is not enough storage
- AOnce, when the file is first loaded
- BOnce per sentence
- COnce for every single word it produces
- DOnly when the answer is long
- AVRAM is a newer invention than RAM
- BVRAM sits on the card, close to the GPU, on a much wider connection
- CVRAM holds more than RAM in every machine
- DRAM empties when the power goes off and VRAM does not
- AThe machine refuses to start and shows an error
- BPart of the file stays in RAM and gets fetched as needed, which is five to thirty times slower
- CThe file is automatically made smaller
- DThe GPU shuts off and the CPU takes over at full speed
- AIt becomes more important
- BIt stops mattering almost entirely, because nothing needs to cross it
- CIt stays exactly as important as before
- DThe reading does not say
- AStorage is where things live when nobody is using them
- BStorage is broken on most machines
- CStorage is only for photos and music
- DStorage is faster than VRAM but harder to reach
- AThey are too old to be useful for anything
- BThey are the wrong kind for this one job, and fine for everything else
- CThey are the right kind but there is not enough of either
- DThey were installed incorrectly at the factory
- ANewer computers are better than older computers
- BFitting and not fitting are two different machines, not two speeds
- CA GPU is a faster version of a CPU
- DEvery computer can run a model if you wait long enough
- ARAM and CPU
- BVRAM and GPU
- CStorage and RAM
- DStorage and GPU
- AStorage is physically larger than the other parts
- BNothing is ever worked on in storage, so it belongs to neither side
- CStorage is optional on most machines
- DThe diagram ran out of room
- AThe Chromebook’s RAM is a small version of VRAM
- BThe Chromebook has no card at all, so its RAM is the only thing that could hold the file
- CVRAM and RAM are two words for the same thing
- DThe Chromebook’s card is too old to be listed
The first item is your name. Answer it — that is what puts you on the work now that email collection is off.