Just weeks away from the rumored public debut of AMD's new ATI Radeon 4000-series GPUs, AMD clues us in on a tantalizing tidbit about a key technology found in its next generation graphics cards. AMD's new lineup of graphics cards will not only use Graphics Double Data Rate 5 (GDDR5) memory, but AMD also claims that this will actually be "the first commercial implementation" of the technology.
"Today's GPU performance is limited by the rate at which data can be moved on and off the graphics chip, which in turn is limited by the memory interface width and die size. The higher data rates supported by GDDR5 – up to 5x that of GDDR3 and 4x that of GDDR4 – enable more bandwidth over a narrower memory interface, which can translate into superior performance delivered from smaller, more cost-effective chips. AMD's senior engineers worked closely with industry standards body JEDEC in developing the new memory technology and defining the GDDR5 spec."
AMD worked with Samsung, Hynix, and Qimonda to develop its implementation of GDDR5. It is the German-based Qimonda, however, who will be supplying the GDDR5 to AMD for the launch of the new Radeon graphics cards. Qimonda claims to have "already started mass production and the volume shipping of GDDR5 512Mbit components with a speed of 4.0Gbps to AMD", despite having produced the first product samples only six months ago.
"In addition to the potential for improved gaming and PC application performance, GDDR5 also holds a number of benefits for stream processing, where GPUs are applied to address complex, massively parallel calculations. Such calculations are prevalent in high-performance computing, financial and academic segments among others. AMD expects that the increased bandwidth of GDDR5 will greatly benefit certain classes of stream computations.
New error detection mechanisms in GDDR5 can also help increase the accuracy of calculations by identifying errors and re-issuing commands to get valid data. This capability is a level of reliability not available with other GDDR-based memory solutions today."
One of the more significant performance bottlenecks for GPUs is available graphics memory bandwidth. GDDR5's higher speed interface and more efficient transaction and error correction response should help to get data to and from the GPU quicker.
Some rumors are also starting to slip about NVIDIA's next-generation GPU, scheduled to be released this summer. While the details are still sketchy, there has been no official mention of an update to the memory architecture--such as a move to GDDR5. In fact, rumors are already circulating that NVIDIA's "June GeForce refresh is going to be limited to legacy GDDR3 technologies". Even if this statement is accurate, it doesn't necessarily preclude NVIDIA from incorporating GDDR5 into later revisions of graphics cards with their new GPUs.
It also bears mentioning that while a new memory architecture is certain to bring performance advances, it is by no means the only available architectural block that could offer a performance boosts. New core GPU designs from both ATI and NVIDIA are sure to offer significant performance advantages as well. What is most important is how the complete design--GPU, memory, interface, etc.--all function together in a balanced, efficient design. Needless to say, we are chomping at the bit to get our hands on the new graphics cards from both ATI and NVIDIA and see how they stack up against each other. More to come on that in the weeks ahead.