AMD's next generation 7nm based graphics architecture code named Navi is expected to debut sometime between 2018 and 2019. Very little is known about Navi at this point as all eyes have been on AMD's Vega graphics chips which will debut in the gaming segment next week in the form of the RX Vega 64 and RX Vega 56.
Some info slipped regarding AMD's futuristic Navi architecture courtesy of Fudzilla, who report that the new graphics chip will feature AI acceleration circuitry. This could be anything from on-chip fabric, new ISA capabilities or even a dedicated AI acceleration engine akin to NVIDIA's tensor core which the company debuted with its Volta V100 accelerator earlier this year.
AMD's First Full Fledged AI Accelerators
Navi is AMD’s third generation graphics architecture designed under the umbrella of the Radeon Technologies Group. The graphics architecture will build on the power efficiency principles of Polaris & Vega. Navi is believed to be the first GPU microarchitecture conceived entirely under graphics guru and head of the Radeon Technolgoies Group, Raja Koduri. It will employ “next gen” memory, likely third generation stacked High Bandwidth Memory or GDDR6 depending on the product to boost performance and efficiency.
Navi is the first graphics architecture from the company built on GlobalFoundries’ upcoming 7nm FinFET process. Based on what GlobalFoundries has so far publicly disclosed regarding the 7nm FinFET process, Navi should land sometime between 2018 and 2019.
AMD Radeon Graphics Architectures
|GPU||Process||Transistors In Billions||Performance||Memory||Bandwidth|
|Southern Islands||2012||HD 7970 Ghz||Tahiti||28nm||4.3||4.1 TFLOPS||3GB GDDR5||264GB/s|
|Volcanic Islands||2013||R9 290X||Hawaii||28nm||6.2||5.6 TFLOPS||4GB GDDR5||320GB/s|
|Caribbean Islands||2015||R9 Fury X||Fiji||28nm||8.9||8.6 TFLOPS||4GB HBM1||512GB/s|
|POLARIS||2016||RX 480||Polaris 10||14nm||5.7||5.2 TFLOPS||8GB GDDR5||256 GB/s|
|VEGA||2017||RX Vega 64||Vega 10||14nm||12.5||13 TFLOPS||8GB HBM2||480GB/s|
|NAVI||2019||TBA||Navi 10||7nm||TBA||TBA||Nextgen Memory||TBA|