In the past couple of weeks, Nvidia has been quite the topic of conversation over its announcement of the upcoming RTX 30 series gaming graphics cards. However, it seems that Nvidia is also getting ready to release its Ampere-based workstation graphics cards, and with GTC 2020 coming up, there's the chance of them getting revealed there.
A screenshot of the Nvidia Control panel from Moore's Law is Dead showed an Ampere graphics card featuring 48GB of GDDR6 memory that was later confirmed to be a Quadro model. As per the screenshot, the alleged Ampere Quadro graphics will feature 10752 CUDA cores (256 more than the RTX 3090), a boost clock of 1860MHz, a memory clock of 16Gbps, and a memory interface of 384-bit. These numbers translate into almost 39TFLOPs of computing performance.
Even bigger Ampere cards are coming, and I have pics… pic.twitter.com/XxQK29eQ9i
— Moore's Law Is Dead (@mooreslawisdead) September 12, 2020
Nvidia's choice of using GDDR6 memory instead of GDDR6X memory might have been forced due to the current lack of ability of Micron producing 2GB modules of GDDR6X memory, forcing the company to go for GDDR6 memory. Nonetheless, the GDDR6 memory is clocked at an impressive 16Gbps, the highest clock speed available for this type of memory. This results in a maximum bandwidth of 768.1GBps, quite less than the RTX 3090 but still over 8GBps faster than the RTX 3080.
The naming scheme of the upcoming Ampere Quadro RTX cards isn't known yet, but they tend to follow a rather simplistic approach. As per the leaker @kopite7kimi, this one will be named A6000, with others expected to follow a similar pattern (A8000/5000/4000)
Nvidia is expected to introduce the upcoming Ampere Quadro RTX graphics cards during the GTC virtual event, scheduled for October 5th.
KitGuru says: Are you a user of workstation graphics cards? Would a 48GB GPU be a tempting upgrade?