Home > Article > System Tutorial > An introduction to understanding the differences between GPU shared memory in win10
Most of my friends must have checked the value of shared memory in the task manager, but they don’t know what the difference between them is. Today I have brought you a detailed introduction to the difference. Let’s take a look. .
Answer: To put it simply, the dedicated memory is used by the GPU itself and will not be distributed to other applications, while the shared memory can be used by all applications, but the GPU Priority use.
Proprietary GPU memory:
The memory of proprietary GPU is generally divided into two types by default: independent graphics memory and integrated graphics memory.
Independent graphics memory means a single GPU PCIe card, which has independent GDDR memory.
And proprietary GPU memory refers to the memory contained on the GPU graphics card. It can only be used by the GPU, and has high bandwidth and very low latency.
Integrated graphics are more complicated in comparison, but this does not include alternative CPUs such as AMD GPUs integrated into Intel CPUs. Take Intel integrated graphics as an example.
The dedicated GPU memory above means that the BIOS replaces the system memory specifically used by the integrated graphics GPU in the system memory.
In other words, the BIOS actually saves a part of the memory after initialization and uses it for the GPU, called "Stolen Memory".
The size generally ranges from 16M to 1024M. The reserved memory supported by integrated graphics of different eras is different. For example, HD4000,
The maximum video memory it supports is 256M, which is not You can be as big as you want.
Shared GPU memory:
The memory here refers to the memory taken out from the system memory, which is given priority to the GPU.
Suppose there is 32G of memory and 16G of shared memory, and there are two GPUs. In fact, not all of the system memory is occupied. 16G is used by two GPUs together, not 16G for each.
Shared memory is not only shared by multiple GPUs, but also shared by the GPU and other applications, but the GPU priority is higher.
And Windows will also try to use dedicated GPU memory, and shared GPU memory can be used by other applications after they use a large amount of memory.
And this value cannot be set and needs to be set by Windows according to the system memory size.
The bandwidth and latency of memory are limited by PCIe and are lower than proprietary memory. This is one of the important reasons why Windows has limited use of proprietary GPU memory.
If you have other questions, you can check it out to learn more about the use of win10gpu~
The above is the detailed content of An introduction to understanding the differences between GPU shared memory in win10. For more information, please follow other related articles on the PHP Chinese website!