Dedicated vs integrated graphics
Classical desktop computer architecture with a distinct
graphics card over
. Typical bandwidths for given memory technologies, missing are the
between GPU and CPU is not possible
, since both have their distinct physical memories. Data must be copied from one to the other to be shared.
Integrated graphics with partitioned main memory
: a part of the system memory is allocated to the GPU exclusively. Zero-copy is not possible, data has to be copied, over the system memory bus, from one partition to the other.
Integrated graphics with unified main memory
, to be found
AMD "Kaveri" or
As an alternative to the use of a video card, video hardware can be integrated into the
motherboard or the
CPU. Both approaches can be called integrated graphics. Motherboard-based implementations are sometimes called "on-board video" while CPU-based implementations are known as
accelerated processing units (APUs). Almost all desktop computer motherboards with integrated graphics allow the disabling of the integrated graphics chip in
BIOS, and have a
PCI Express (PCI-E) slot for adding a higher-performance graphics card in place of the integrated graphics. The ability to disable the integrated graphics sometimes also allows the continued use of a motherboard on which the on-board video has failed. Sometimes both the integrated graphics and a dedicated graphics card can be used simultaneously to feed separate displays. The main advantages of integrated graphics include cost, compactness, simplicity and low energy consumption. The performance disadvantage of integrated graphics arises because the graphics processor shares system resources with the CPU. A dedicated graphics card has its own random access memory (
RAM), its own cooling system, and dedicated power regulators, with all components designed specifically for processing video images. Upgrading to a dedicated graphics card offloads work from the CPU and system RAM, so not only will graphics processing be faster, but the computer's overall performance may also improve.
Both of the dominant CPU makers, AMD and
Intel, are moving to APUs. One of the reasons is that graphics processors are powerful parallel processors, and placing them on the CPU die allows their parallel processing ability to be harnessed for various computing tasks in addition to graphics processing. (See
Heterogeneous System Architecture, which discusses AMD's implementation.) APUs are the newer integrated graphics technology and, as costs decline, will probably be used instead of integrated graphics on the motherboard in most future low and mid-priced home and business computers. As of late 2013, the best APUs provide graphics processing approaching mid-range mobile video cards and are adequate for casual gaming. Users seeking the highest video performance for gaming or other graphics-intensive uses should still choose computers with dedicated graphics cards. (See
Size of market and impact of accelerated processing units on video card sales, below.)
Beyond the enthusiast segment is the market for professional video cards for workstations used in the
special effects industry, and in fields such as design, analysis and scientific research. Nvidia is a major player in the professional segment. In November, 2013, AMD introduced a so-called "Supercomputing" graphics card "designed for
data visualization in finance,
aeronautics and automotive, design and engineering, geophysics, life sciences, medicine and defense."