Background: Today’s generative AI compute tasks are putting immense pressure on memory access by requiring model weights to be loaded continuously from a a remote (DRAM) memory. The only way to provide sufficiently large (GBs) memory on chip is to start exploiting the 3rd scaling dimension using vertically stacked data concepts. In this context, Vertical Compute is developing a unique magnetic Vertical integrated Memory device concept poised to reshape how data and compute integrate together on a chip. Thanks to the inherent high speed of magnetics and the vertical data stacking, both high performance and (very) high density can be achieved.
About what you will do: This PhD project supports the exploration of Vertical Compute’s vertical magnetic device concept. The mission will be to explore, understand and develop the fundamental physics of device operation. This will require extensive magnetic and electrical characterization of 300mm fabricated devices, potentially with support of micromagnetic simulation. In this course, alternative materials and methods will be developed using SPINTEC fabrication facilities. As such, this thesis will be stationed mostly at SPINTEC in Grenoble, the PhD program being registered at Université Grenoble Alpes. You will also be part of the Vertical Compute Grenoble team, which has offices nearby.
Check the detailed offer.
Contacts at SPINTEC: Olivier Fruchart, Kevin Garello, Aurélien Masseboeuf