A new report suggests that NVIDIA’s upcoming Ampere GPUs will use the TSMC forge’s 7 nm process node, while the next-gen Hopper GPUs will shift towards Samsung’s announced 5 nm EUV process node technology.
The report comes courtesy of China Times. The outlet says that NVIDIA already stands as one of the most significant companies lining up to secure TSMC’s 7nm process, which hints that the Ampere GPU’s will be built on the tech. Or at the very least, Ampere’s data center and HPC lineup will use TSMC’s 5nm process.
Alongside Samsung’s Q1 earnings report confirmed that the Korean giant would kick start production of its very own 5nm manufacturing process in Q2 2020.
The report suggests Samsung is already fielding orders from manufacturers, notably for GPUs and CPUs presumed to use the 5nm process node destined for the consumer market. The report points towards NVIDIA as one of Samsung’s new customers placing orders for 5nm process node products.
As it stands, this remains unconfirmed, but with NVIDIA primed to touch on its Ampere plans in a pre-recorded GTC 2020 keynote speech from CEO Jensen Huang on May 14th, we can expect to hear a lot more very soon. The keynote usually touches on significant innovations and announcements for upcoming products. With the tagline ‘Get Amped,’ it’s relatively clear what NVIDIA is alluding to. We can expect to discover the new lineup of Ampere GPUs as well as the specifications for these.
Further backing up the imminent announcement of Ampere GPUs are reports that AIB partners are acting fast to liquidate existing stocks of GeForce RTX GPUs by slashing prices to make way for what is presumed to be a new stock of Ampere GPUs later this year. It’s widely expected that NVIDIA will launch its consumer GeForce RTX 3000-series products at the rescheduled Computex 2020 expo in Q3 2020.
What’s clear is that rumors and reports are coming in left, right, and center – there’s no smoke without fire.