Advanced Micro Devices already uses TSMC's 28nm High-Performance manufacturing process for highly-complex chip designs, but its lesser chips have yet to progress to that node, something that will be corrected by next year.
Since we have mentioned Taiwan Semiconductor Manufacturing Company, we are, naturally, talking about AMD's Graphics Processing Units, not its CPUs. As far as we know, AMD's central processing units will get to the 22nm or 20nm node by 2014. Intel will have 14nm chips by that time, but we've already seen that a smaller node doesn't necessarily mean better chips, not if the processors are built properly. Keep in mind that AMD said, months ago, that it would no longer bother trying to push its manufacturing process forward at the same rate as Intel. Instead, it will focus R&D resources on optimizing design and security (even if it has to partner with ARM to do it). Still, as we said, we won't be talking about CPUs here. Instead, we will be relaying the information contained in this rumor.
Long story short, the Sunnyvale, California-based company will adopt the 28 nanometer bulk CMOS silicon fabrication process next year. 28nm bulk CMOS is used for less complex products that need to be manufactured in high volumes. For example, the Sea Islands GPUs will not be using it (they are made on the aforementioned high-performance process instead), but motherboard chipsets and low-end accelerated processing units will. In related news, the same report states that the Sea Islands themselves are already in tape-out stage, meaning that AMD probably has prototypes floating around. We aren't surprised, not with other sources exposing upcoming products, including the Radeon HD 8000 series (Venus, Mars and Oland).
That's it for AMD process news for now. We're sure some more leaks and reports will pop up as time passes, but this week will probably be all about the new Top500 supercomputer list and AMD's contribution to it.
0 comments:
Post a Comment