Last January had officially ratified JEDEC specifications of the new GDDR5X standards to be adopted for the next generation of video cards higher end. Following the American Micron had reported that the development program of its GDDR5X memories was active, with first products targeted for upcoming partners to be made more available
That time has arrived: Micron has announced that it has begun shipments of GDDR5X memory to its partners, who will use these memories in conjunction with its new high-end video cards. We can not think about the proposals of the Pascal family NVIDIA and AMD the Polaris range as candidates to the use of GDDR5 memory, also due to the unique architecture of these solutions.
gddr5_schema_bandwidth.jpg (46547 bytes)
The GDDR5X standard does not differ from the point of view of design guidelines from what GDDR5; consequently its implementation for manufacturers of video cards is overall very simple. All this, however, comes with obvious benefits: the GDDR5X memories are able to provide up to a doubling of the maximum bandwidth compared to GDDR5 memories, arriving at values between 10:14 Gbytes per second of data rate per pin against a peak of 7 Gbytes the second of which they are capable GDDR5 memories.
Hence, even with the same GPU architecture, you can achieve interesting performance increases with the use of GDDR5X memories by removing at least part of the bottleneck given by the bandwidth of video memory, the more evident with increasing screen resolution and the complexity in qualitative settings chosen for the running game.
The GDDR5X standard is therefore expected to become the reference in the market of graphics cards with high performance during the second half of 2016, before the debut of HBM2 memories (High Bandwidth Memory second generation) that will be assigned to just the top of the range proposed such as NVIDIA AMD expected between late 2016 and early 2017.
That time has arrived: Micron has announced that it has begun shipments of GDDR5X memory to its partners, who will use these memories in conjunction with its new high-end video cards. We can not think about the proposals of the Pascal family NVIDIA and AMD the Polaris range as candidates to the use of GDDR5 memory, also due to the unique architecture of these solutions.
gddr5_schema_bandwidth.jpg (46547 bytes)
The GDDR5X standard does not differ from the point of view of design guidelines from what GDDR5; consequently its implementation for manufacturers of video cards is overall very simple. All this, however, comes with obvious benefits: the GDDR5X memories are able to provide up to a doubling of the maximum bandwidth compared to GDDR5 memories, arriving at values between 10:14 Gbytes per second of data rate per pin against a peak of 7 Gbytes the second of which they are capable GDDR5 memories.
Hence, even with the same GPU architecture, you can achieve interesting performance increases with the use of GDDR5X memories by removing at least part of the bottleneck given by the bandwidth of video memory, the more evident with increasing screen resolution and the complexity in qualitative settings chosen for the running game.
The GDDR5X standard is therefore expected to become the reference in the market of graphics cards with high performance during the second half of 2016, before the debut of HBM2 memories (High Bandwidth Memory second generation) that will be assigned to just the top of the range proposed such as NVIDIA AMD expected between late 2016 and early 2017.
Comments
Post a Comment