How to tell if your PC is using shared memory for the GPU without you knowing

¿How can you tell if your PC is using shared memory for the GPU without you knowing? If you've ever opened the Windows Task Manager and seen strange things like “Dedicated GPU memory” and “Shared GPU memory”You've probably been left with a blank stare. Adding to the confusion, sometimes your PC seems to be using shared memory for the GPU without you having changed anything in the BIOS or Windows. And of course, all the alarm bells start ringing: Is my RAM being "stolen"? Am I losing performance? Which applications are hogging the GPU without my knowledge?

To complicate matters further, many monitoring programs show video memory data is "allocated" but not always the memory that is actually in useThis is where many misunderstandings about VRAM, shared RAM, and what Windows displays in its graphs come from. In this guide, we'll break it all down calmly: how to tell if your PC is using shared memory for the GPU, how to check it by application, why it happens, what's going on with the processor's integrated GPU (iGPU), and what you can (or can't) change in the BIOS.

What exactly is GPU shared memory in Windows?

When you look at the GPU performance tab in Task Manager (Ctrl + Shift + Esc), you'll see two main values: “Dedicated GPU memory” and “Shared GPU memory”These names are not random, but reflect how Windows and the video memory manager (VidMm) organize graphics resources.

On the one hand, the Dedicated GPU memory is the physical VRAM of your graphics card: the 4, 6, 8, 12 GB, etc., that are soldered directly onto the card. This is a space reserved solely for the GPU and managed by the system through VidMm. In discrete GPUs (NVIDIA, dedicated AMD), it's the card's memory. In integrated GPUs, it's the remaining portion of the system RAM. locked exclusively for graphics.

On the other hand, the Shared GPU memory is system RAM that the GPU can use flexibly.It's not permanently "hijacked": the CPU can use it as regular memory, or the GPU as if it were video memory when needed. It's a shared resource pool where Windows decides who takes control at any given time based on the workload.

The important thing is to understand that The figure you see as "Total Shared GPU Memory" does not mean that amount is actually being used.This is usually the maximum amount Windows is willing to allocate to the GPU if it deems it necessary. Actual usage is shown in the "in use" value next to it, which is typically a few hundred MB in many cases.

Furthermore, Windows obtains that data directly from VidMm, the video memory managerAs Microsoft itself explains, The figures shown by Task Manager represent memory that is actually being used, not just the memory that has been requested or reserved.And this applies to DirectX 9, 11, 12, OpenGL, CUDA and other GPU-intensive APIs.

How to see if your PC is using shared memory for the GPU

To know for sure if your computer is using shared memory for the GPU, Windows 10 and 11 already include a very handy tool: the Task Manager itself, which makes a pretty decent “GPU monitor”.

The basic process is simple: open Task Manager with Ctrl + Shift + EscGo to the “Performance” tab and select “GPU” in the left column. There you will see several graphs (3D, Copy, Video Encoding, etc.) and below them a block with memory information for that GPU.

In that block, fields such as the following will appear: “Dedicated GPU memory” and “Shared GPU memory”along with the current usage. For example, you'll see something like "GPU Memory: 4,0 / 24,0 GB", "Dedicated GPU Memory: 8,0 GB", "Shared GPU Memory: 300 MB of 16,0 GB". That means your graphics card has 8 GB of physical VRAM and the system has up to 16 GB of RAM that it could use as shared memory for the GPU, of which only 300 MB are currently being used.

If your CPU has integrated graphics (iGPU) and you haven't disabled it, It's normal for Windows to reserve a certain amount of RAM as potential memory for that iGPUEven if you're using a dedicated graphics card. In many cases, the shared memory slot you see is closely tied to the processor's integrated GPU (iGPU), even if your dedicated GPU is handling the video output.

On some systems, you'll see that this shared memory either doesn't move or remains at ridiculously low values. That doesn't mean it's being wasted: It's simply available in case it's needed.But Windows doesn't mark it as truly busy until a graphics process actually needs it.

How to see which applications are using the GPU and how much video memory they consume

Just as you can see which programs are using the most CPU, you can also see which processes are using the GPU and how much graphics memory each one is using. For this, you don't need to resort to MSI Afterburner or other external tools, because Windows 10/11 already offers a breakdown by process.

Within Task Manager, go to the "Details" tab. There, you can right-click on any column heading (for example, "Name") and click "Select columns." In that window, check the boxes related to GPUs, such as “GPU Engine”, “Dedicated GPU Usage” and “Shared GPU Memory Usage” (Depending on the Windows version, the names may vary slightly, but the idea is the same).

Once those columns are activated, you'll see for each process how much dedicated GPU memory and how much shared memory it is usingMicrosoft explains that this number reflects the total amount of memory used by that processand that the sum may exceed the total GPU memory because some blocks of graphics memory are shared between processes.

This means that the Task Manager's process view gives you a fairly reliable picture of Which application is using up VRAM and how much shared RAM is it using as support?If you're looking for "something like Task Manager but for the GPU," this view is currently the main reference integrated into the system.

Tools like MSI Afterburner, HWiNFO64, and similar programs are great for monitoring overall GPU usage, frequencies, temperatures, etc., but in many cases They are based on data from VRAM "allocated" by the game or driver, not strictly on the memory actually in use at any given moment.and other factors such as Windows may limit FPS even though the GPU isn't at 100%. That's why, for example, you might see 10 GB "used" in a game that, if you look closely at the Task Manager, isn't actually using that much effectively.

Dedicated vs. shared GPU memory: what they mean in practice

A common source of confusion is thinking that if Windows says "GPU Memory: 4 of 24 GB", that means your magic 8 GB card has become a 24 GB VRAM card. It doesn't work that way.What happens is that the system is adding the physical VRAM of the GPU (for example, 8 GB) with the RAM it could use as shared memory (for example, 16 GB), and is showing you the total potential graphics memory.

At the bottom of the GPU panel you will first see the “Dedicated GPU memory” (the actual VRAM of your card). Right below, the “Shared GPU memory”where it indicates something like “300 MB of 16,0 GB allocated”. Those 16 GB aren't actually taken from your RAM from the start, but rather represent the maximum amount of RAM that Windows could make available to the GPU if the workload required it.

It is very likely that, if your processor has an iGPU and it is not disabled in the BIOS, You'll see those shared memory values ​​even if you never connect a monitor to the motherboard's output.From the operating system's point of view, that iGPU is still alive and needs to have a memory buffer in case it is used, so Windows pre-configures that "virtual reserve".

Actually, That shared RAM is a mix of reserved memory and virtual memoryIt's not extra physical VRAM, nor will it turn your graphics card into a gaming powerhouse. It serves as support, especially for integrated graphics or for specific computing applications with GPGPUs, where the GPU needs to read from and write to the same memory as the CPU.

In fact, if you force the iGPU as the primary graphics card in the BIOS but connect a monitor to a dedicated graphics card, a curious situation arises: Windows may end up using the dedicated GPU's VRAM as a backup for the iGPUFrom the outside it looks like a mess, but what the system is doing is playing with several memory pools (local VRAM and system RAM) so that the CPU and GPU can synchronize data without friction.

The processor's iGPU, BIOS, and shared memory

Many people worry when they see the system "taking" part of their RAM to use as graphics memory, especially when they already have a powerful dedicated graphics card with its own VRAM. The first thing to clarify is that Windows isn't stealing 16 GB of "real" RAM for the GPUWhat you see is the maximum capacity that could be used, not a constant consumption.

This shared memory pool is closely linked to the integrated graphics (iGPU) of the processorIf your CPU doesn't have integrated graphics, you probably won't even see this. But if your processor does have an iGPU (like Intel processors with UHD Graphics or many AMD APUs), Windows leaves that iGPU active unless you explicitly disable it in the BIOS/UEFI.

If you enter the BIOS/UEFI and go to the advanced options, many motherboards offer a setting of the type “Internal Graphics”, “iGPU”, “IGD” or similarIn that section, depending on the manufacturer, you can choose things like “Enabled”, “Disabled” or “Auto”; and sometimes even set the maximum amount of shared memory for the integrated graphics (256 MB, 512 MB, 1 GB, 2 GB…).

Let's be clear: Not all motherboards and processors allow you to adjust that shared memory.To do this, the processor must include an integrated GPU (iGPU), the chipset and motherboard must support this configuration, and the motherboard manufacturer must have exposed the options in the BIOS; you can check if your motherboard needs updating for that purpose.

Although you could disable the iGPU in the BIOS to stop that shared memory from appearing, it's not usually the best option. If your dedicated graphics card breaks down or fails one dayYou'd be left without video and without an emergency way to boot your PC, diagnose problems, or continue working. Keeping the iGPU available is a kind of free insurance that's very welcome when you need it.

Are you "losing" RAM because of shared video memory?

A very common question is whether, if Windows shows several GB as "shared GPU memory", That RAM is no longer available for the operating system or gamesThat impression is misleading: what actually happens is that the system marks a strip of RAM as potentially usable by the GPU, but if the CPU needs it, it will use it without hesitation.

In other words, if you see that your computer has 8 GB of dedicated VRAM and "X GB" of shared GPU memory, it doesn't mean that those X GB are unusable. The RAM is only effectively allocated to the GPU when there is a specific workload that requires it.Meanwhile, Windows continues to use it for all other tasks without you losing anything.

This idea can be summarized as follows: Shared video memory is not a permanent reduction of your RAMRather, it's a way the system organizes resources so that, if needed, the GPU can have a little extra help. That's why, when a user sees "3921 MB lost," they aren't actually losing them: they're still available to the system if needed.

Furthermore, for demanding games, what truly makes the difference isn't shared RAM, but rather a powerful dedicated graphics card with ample VRAM and adequate bandwidth. Integrated graphics with shared memory are fine for light tasks and some very modest games, but they aren't designed for serious gaming, no matter how much more RAM you allocate to them in the BIOS.

Also note that Windows uses other virtual memory systems besides shared GPU memory. For example, It can use disk space (page file) as an extension of RAMThis means you'll see "confirmed memory" figures much higher than the physical RAM installed. It's not magic: it's simply another layer of virtual memory that the system uses to better handle configurations with less DRAM.

GPU virtual memory, GPGPU, and graphics RAM usage

The graphics memory you see as shared isn't just used for "painting" things on the screen. These days, many applications use the GPU as a calculation engine to accelerate all kinds of tasks: AI, video editing, rendering, compression, simulations, etc. This idea is often called GPGPU (General-Purpose computing on Graphics Processing Units).

For the CPU and GPU to work together efficiently under these types of workloads, it is very useful that they can access the same coherent memory spaceThis is where system RAM comes into play. The GPU can operate with two different memory "pools": its local VRAM (fast but separate) and system RAM (shared with the CPU, accessible via PCI Express and other mechanisms).

When a program initiates a calculation using Compute Shaders or other compute APIs, the CPU sends the job to the GPU, The GPU processes the data and then writes the results back into the system RAMThis is achieved through the DMA units and the PCIe bus. While physical VRAM remains crucial for textures, framebuffers, and high-frequency data, shared RAM allows both processors to share results without unnecessary intermediate copies.

Many algorithms improve their performance not by 10 or 20%, but in orders of magnitude when processes switch from running on the CPU to running on the GPU. To take advantage of this without having to include absurd amounts of VRAM on each card (something that would significantly increase their cost and wouldn't always be necessary), Windows and manufacturers have opted for treat part of the system RAM as a “virtual extension” of the video memory.

It is true that in sequential read and write operations, the system's DDR4/DDR5 can outperform the graphics card's GDDR in some aspects, but VRAM is optimized for massive copy operations and parallel access, where It significantly impacts the system's DRAM usage.Therefore, shared memory is a complement to, not a replacement for, real VRAM, and is used in a very controlled manner.

“Allocated VRAM” vs VRAM actually used: why there is so much confusion

In the gaming world, there has been considerable confusion surrounding the concepts of "Allocated" VRAM, "reserved" VRAM, and VRAM actually used because of games. Some OSD tools, such as certain configurations of MSI Afterburner or HWiNFO64, report on the memory that the game has requested from the system, but they don't always differentiate how much of that memory is filled with active data at any given time.

This means that, for example, you might see absurd VRAM usage figures for a game (10, 12 GB “in use” at 1440p) when, checking the information provided by Windows through VidMm, the The memory that is actually being used is lessGraphics engines, by design, usually request more memory than they need "just in case," to avoid stuttering when loading new textures or scenes.

The Task Manager's own "Dedicated GPU memory usage" tab is fed directly from VidMmMicrosoft has clarified that this data represents used memory, not simply reserved memory. The same applies to the shared memory displayed in that panel, which reflects the portion of the system's RAM that is currently acting as video memory.

When analyzing a game, it's interesting to see What is the video memory consumption at different resolutions and settings?For example, by comparing 1080p, 1440p, and 4K, or by increasing texture quality. The key is to combine the information provided by the OSD tools with what you see in the Task Manager to distinguish between the VRAM allocated on paper and what the engine is actually using.

If you'd like to delve deeper into a specific title, you can start collecting your own measurements and noting how much VRAM and shared memory each game uses on your computer. Sharing this data with other users also helps to dispel myths and provide a more realistic idea of ​​what each game requires on different configurations.

Why does your dedicated GPU seem to "share" VRAM with the integrated one?

One particularly perplexing case is when you have, for example, a dedicated NVIDIA GTX 1080 and, at the same time, an Intel UHD iGPU in the processorAnd you see in the Task Manager that it seems as if both "share" VRAM or that you can take one monitor from the motherboard and another from the card as if it were all a single graphics system.

What happens internally is that Windows treats all system GPUs (integrated and discrete) as graphics adapters that can access certain common memory resourcesThe system RAM can act as a shared area that both GPUs access, either to display desktops, to calculate things in the background, or for distributed GPGPU tasks.

If you connect one monitor to the motherboard and another to the dedicated graphics card in a modern setup (for example, with an Intel i5-12600K processor and a GTX 1080 on a Z690 motherboard), Windows is capable of using both GPUs at the same timeThe iGPU can handle one screen, the dedicated one another, and even delegate certain video tasks (encoding, decoding) to one or the other as needed.

That's why, when you look at the Task Manager, you get the impression that the 1080 is "sharing VRAM with the Intel graphics". It's not that they're physically distributing the GTX 1080 memory chips.but both are accessing a memory space managed by Windows (system RAM and GPU virtual memory) that they can use as support in addition to their own local memory.

This architecture, although it may seem somewhat magical, is designed so that the system can to better distribute the load between CPU, iGPU and dedicated GPUAnd so that you don't lose functionality if one of the two becomes unavailable, as long as the BIOS keeps the iGPU active and there is enough RAM to play with these reserves.

From the outside, it might sound like a bit of chaos, but in day-to-day life, what you're interested in is knowing that You are not wasting VRAM or RAM because Task Manager refers to "shared" memory.What Windows is doing is leveraging each piece of hardware to compensate for shortcomings in another when needed, without you having to constantly reconfigure everything.

Ultimately, if you understand that your GPU's dedicated VRAM is the primary reference point for games and demanding graphics tasks, that shared memory is a flexible backup taken from RAM, and that the Task Manager provides fairly reliable data at both the process and system levels, You'll be able to interpret those figures you see on the screen much better. And you'll avoid thinking that your PC is "hiding" memory or using it behind your back in a dangerous way.