Image: SK Hynix
More memory and faster memory are always big deals when it comes to new graphics cards, and the next generation of GDDR (Graphics Double Data Rate memory) is going to be fast indeed—approximately double the rate of memory in current cards at up to 40 gigatransfers per second.
According to memory giant SK Hynix, mass production on these memory chips will start as soon as the third quarter of 2024. That’s this quarter, sometime before October. Neat!
Previous industry analysis had set GDDR7 chip availability at sometime in early 2025, which is when the RTX 5000 series of graphics cards from Nvidia is expected to drop. But with this newer, faster memory entering large-scale production before the end of the year according to SK Hynix (spotted by VideoCardz.com), it’s possible that next-gen GPUs could hit the market even earlier than expected.
Of course, memory is only one aspect of the incredibly advanced puzzle that is the modern graphics card. The other major factor? The processor fabrication itself. And there’s nothing stopping Nvidia from delaying a retail launch, though having tons of memory supply available might help to alleviate the usual supply crunch that happens whenever an in-demand GPU hits the market. (Those crunches have only gotten, um, crunchier now that massive AI data centers filled with power-hungry GPUs are becoming industrial necessities.)
SK Hynix and the other two members of the “big three” of the memory market—Micron and Samsung—have all shown off GDDR7 samples this year. While the fastest chips are “only” 32Gbps with 16GB modules, that would be an impressive boost over conventional designs even at launch.
For the sake of comparison, the standard RTX 4090 board layout uses 12 individual 2GB Micron GDDR6X chips running at 21Gbps. Just two new GDDR7 chips would be larger (32GB) and faster.
More concrete details on next-gen graphics cards from Nvidia, AMD, and Intel will have to wait. But if this new memory tech is any indication, they’re going to offer some absolutely wild performance. And with AI companies gobbling up every high-end graphics chip they can find, you can bet they won’t be coming cheap.
Author: Michael Crider, Staff Writer, PCWorld
Michael is a 10-year veteran of technology journalism, covering everything from Apple to ZTE. On PCWorld he’s the resident keyboard nut, always using a new one for a review and building a new mechanical board or expanding his desktop “battlestation” in his off hours. Michael’s previous bylines include Android Police, Digital Trends, Wired, Lifehacker, and How-To Geek, and he’s covered events like CES and Mobile World Congress live. Michael lives in Pennsylvania where he’s always looking forward to his next kayaking trip.