3D Ferroelectric NAND In-Storage Processing Architecture for Mass Spectrometry

Open modification search (OMS) is a critical step in mass spectrometry (MS) analysis and proteomics for identifying peptides in protein samples. However, large- scale OMS is highly data-intensive that requires hours to days of computation. In this work, we present a reconfigurable architecture built upon 3D Ferroelectric NAND (FeNAND) in-storage processing (ISP), leveraging heterogeneous integration to accelerate mass spectrum data processing. We fabricate a 2D FeNAND device and verify the feasibility of 3D FeNAND through TCAD and 3D netlist SPICE simulations. Experimental results indicate that, with appropriate hardware configurations, the proposed 3D FeNAND ISP architecture achieves a 14.3× to 24.2× speedup over the GPU baseline, while reducing energy consumption by five orders of magnitude through the elimination of large-scale data transfers. The proposed design offers an energy-efficient, high-performance ISP solution for emerging large-scale OMS applications in proteomics.