Graphics chipmaker ATI unveiled a new technology it
said lets its visual chips share system memory for graphics processing.
HyperMemory uses the
high-speed, bi-directional data-transfer capabilities of PCI Express
loosely based on the original shared aspects of Accelerated Graphics Port
(AGP) technology, which introduced a dedicated point-to-point channel, so
that the graphics controller could directly access the main memory.
However, previous interconnect standards were not fast enough for
real-time graphics applications, so graphics cards have since shipped with
up to 256 megabytes of dedicated graphics memory to store textures and rendering
data required by the graphics processor.
ATI said HyperMemory increases the performance of system bus data
transfers, making accessing system memory faster and keeping the overall
system cost down.
“HyperMemory uses intelligent memory allocation algorithms to optimize
the use of available local memory and ensure critical components are placed
in fast local memory when required,” ATI said in its statement. “Optimal
assignment of data to local or system storage is determined dynamically.”
The company also said its new HyperMemory technology lets its board
partners make graphics cards with less on-board memory and instead use
system memory to handle the graphics storage requirements. ATI said graphics
cards featuring its HyperMemory technology would be announced later this
year.
ATI’s timing is good considering PCI Express is the next step, offering
transfer rates of 4.0 gigabytes per second over a x16 PCI Express lane for
graphics controllers. Intel and AMD
are
expected to have their respective ranges of PCI Express-compatible
products available next year. Already, high-end workstation users and gaming
programmers are looking at PCI Express as a way to reduce the load on their
Pentium and Athlon systems.
However, Jon Peddie, president of analyst firm Jon Peddie Research and
Tech Watch publisher, said the physical and electrical characteristics are
completely different from AGP, which may signify some slow adoption rates
for HyperMemory right out of the gate.
“What that means is users will not be able to move the graphics add-in
board (AIB) from their previous workstation to the new one if it has PCI
Express in it,” Peddie said in a recent newsletter. “Some workstations will
be built using what’s known as a bridge. The bridge will convert the PCI
Express signals into AGP8X signals and such systems will actually have two
AIB slots, but only one will be usable at a time.”
