Integrated versus Dedicated Graphics

From MEPIS Documentation Wiki

Jump to: navigation, search

Integrated Graphics / Dedicated Graphics

  • This is primarily a History page to clarify the differences between Integrated Graphics and Dedicated graphics. These categories are separate from Graphics Accelerators and Graphical Processing units. For the differences between those, visit Graphics Accelerators versus Graphics Processing Units.

Integrated Graphics

As the name implies, Integrated Graphics are generally part of the system. Integrated Graphics derive their name from being integrated onto the motherboard of the system. Typically, there are two ways to attach Integrated Graphics into a system

  • 1: Separate Chip on the Motherboard itself
    • This is typically seen on workstation motherboards today that still use the relatively cool running Rage128 from ATi.
    • This is also typically seen on older consumer motherboards that also used the Rage128 graphics, such as those from Gateway.
  • 2: Integration into the Northbridge
    • In this design method, the Graphics Chip is placed on the same piece of silicon as the Memory Controller and other I/O Functions
      • This is the typical design found in Intel Motherboards
      • AMD motherboard designs before the Athlon64 also were often setup in this manner
    • In the Athlon64 motherboard designs, the memory controller is placed on the processor
      • The result is that the Northbridge will generally hold the Integrated GPU and all of the I/O functions, cutting down on heat, and making the motherboard less expensive to produce, which means motherboards are generally cheaper.
        • This can create some issues when attempting to write drivers for these types of Graphics Chips, as the use of HyperTransport[1] to access the controller may not be fully defined

In both implementations of the integrated graphics, the Graphics chip gets it's video memory, or framebuffer, from the system memory.

  • Example : Your computer has 512megs of RAM. Your Integrated Graphics Chip can use 256megs of RAM. If you tell your Graphics chip to use it's 256megs of RAM, your system memory will be cut in half. Only 256megs will be made available to the processor.

The result is that Integrated Graphics are slow performers. Because they have to share memory with the system, they also have to share the same memory controller as the processor.

Not that there haven't been ways to get around the inherent performance problems in Integrated Graphics. Nvidia has a technology called TurboCache[2], and AMD/ATi has a technology called HyperMemory[3]. Both of these technologies do about the same thing. They place a small amount of Dedicated memory into the system for the use of the Graphics Chip. Then, beyond that originally small framebuffer, additional memory is shared from the Main System Memory. The result is that in most cases where Integrated graphics are used, the small framebuffer is enough.

  • Small note on this: HyperMemory and Turbocache were originally designed for extremely low-end Dedicated cards, however the Radeon Xpress 200 motherboard chipset appears to have this technology enabled. AMD 690 based motherboards also appear to use a HyperMemory setup.


Dedicated Graphics

Where-as Integrated Graphics refers to having a Graphics Processor that is integrated into the motherboard, Dedicated Graphics refer to separate add-in cards that contain the Graphics Chip and all of the support electronic components for the Graphics chip. Dedicated Graphics generally come in 3 different types of add in cards: PCI | AGP | PCI-Express

Dedicated Graphics cards contain the Graphics Chip itself, the memory for the framebuffer, the Digital to analog converters for VGA or TV output, and the physical ports for the display as well.

As Dedicated Graphics are add-in cards, they offer several advantages over Integrated graphics:

  • System Performance : Where integrated graphics borrow from the main system memory to provide the frame buffer, what actually holds the content of what is displayed, Dedicated Graphics have their own frame buffer. This results in the graphics card being able to make better use of it's resources.
  • Not enough power? Change cards : With integrated graphics, there is no upgrade path. What you buy is what you get. With Dedicated graphics, if your current graphics card is not powerful enough, buy a new one.
  • Extra display ports / Multiple displays : Most integrated graphics only have an output for one display device at one time. It has only been within the past couple of years that motherboards have started carrying two different types of display ports, generally a VGA and an TV out, and more recently, a VGA and DVI output. With an add-in Dedicated card, you can get various types of inputs and outputs.


There are, however, some disadvantages

  • Cost : An add in card will increase the price of a system if you are building one from the ground up.
  • Something more to go wrong : A Dedicated Graphics card does increase the complexity of a computer, so when something goes wrong, it can be another factor in a malfunction.
  • Heat : Since most dedicated graphics cards don't have to concern themselves with fitting into a particular thermal envelope (maximum thermal output) that a motherboard does, they can be much faster. More performance comes at a price. Some dedicated graphics cards will have fans that take up to two normal slots on a motherboard.
  • Noise : some dedicated graphics, like the Geforce FX 5800 series, have earned nicknames like Dustbuster. Not all dedicated graphics have obnoxious fans, and some don't have any fans at all, but you may not know how your dedicated graphics card will sound until you turn it on.


(saving this to the Wiki)

Personal tools