Although Nvidia Ampere GPUs seem to be around the corner, the company may use an older architecture on its next generation of MX GPUs. By reading into the latest Nvidia driver's code, it seems the next MX300 GPUs will be using Pascal architecture, the same architecture used on GeForce 10-series GPUs, as well as the MX10 and MX20 series.
The MX series is aimed at entry-level and mid-level laptop systems. Both MX100 and MX200 had already used Pascal architecture on its GPUs, and it seems that this trend will continue. The spotted MX300 GPUs, the MX330 and the MX350, are rumoured to keep the same architecture as its predecessors in 2020.
Pascal is now a four years old architecture that debuted with the launch of Nvidia Geforce GTX 1080, which uses GP104 chips. This architecture used TSMC's 16nm FinFet process node, but it later jumped to Samsung's 14nm FinFet process node.
The lines of code added to the driver are the following:
- NVIDIA_DEV.1C94 = “NVIDIA GeForce MX350”
- NVIDIA_DEV.1D16 = “NVIDIA GeForce MX330”
According to Notebookcheck, the suffix “1C94” means that the MX350 won't use the GP108 chip, unlike the MX250 and MX150. The suffix “1C” points to either a GP107 or a GP106. But when taking into account that the GeForce GTX 1050 Max-Q has the “1C92” code, the GP107 is more likely the one to be used.
Regarding the MX330, the suffix “1D” that it carries is used for GPUs with GP108 chips. Based on this, it is possible the MX330 might be a refresh of the current MX250.
Notebookcheck also says the official announcement of the upcoming MX330 and MX350 is slated for February 20.
Discuss on our Facebook page HERE.
KitGuru says: If this is true, why do you think Nvidia would decide to use Pascal instead of Turing or Ampere? Would you buy a laptop in 2020 using a 4-year old GPU architecture?