News
News Categories

AMD will skip ahead to 14nm fab process for its next-gen GPUs

By Wong Chung Wee - on 24 Apr 2015, 11:42am

AMD will skip ahead to 14nm fab process for its next-gen GPUs

Currently, the AMD R9 290X is the most powerful single GPU discrete graphics card from the company.

AMD will skip ahead to the 14nm fabrication process for its next-generation GPUs, which have been given the codename of “Artic Islands”. According to a report by techPowerup, AMD will move from the current 28nm fabrication process to the 14nm one, bypassing the 20nm fab process.

This is due to alleged claims that the 20nm node isn’t suitable for “high performance discrete GPUs”, unlike the NVIDIA Tegra K1 SoC, which is based on the said node. As a result, AMD is allegedly going with 14nm FinFET transistor architecture for its upcoming “Arctic Island” GPUs. As noted by techPowerup, Samsung and Intel are only two companies with fab plants capable of 14nm technology with production capacity. The technology is used for CPU, SoC and NAND production.

Therefore, AMD may dissociate itself from TSMC as the company moves forward with its 14nm-based GPUs. Their nomenclature hints at the possible low TDPs of the graphics chips, as a direct benefit of the new FinFET technology. If things go according to plan, the "Greenland" GPU will be the first from the Artic Island series, and it will feature AMD’s “most advanced stream processor design”, with support for HBM2 memory. This second generation high-performance memory interface is touted to offer 57% more bandwidth, at 48% lower power requirements than the current GDDR5 technology. SK Hynix is supplying the HBM2 memory ICs to AMD for its Arctic Island GPUs. Click here for more information on High Bandwidth Memory (HBM) interface.

(Source: techPowerup, Fudzilla, Wikipedia)

Join HWZ's Telegram channel here and catch all the latest tech news!
Our articles may contain affiliate links. If you buy through these links, we may earn a small commission.