Leti shrinks SRAMs by 30%, proposes new memory designs

December 07, 2017 // By Julien Happich
In a paper titled “Advanced Memory Solutions for Emerging Circuits and Systems” it presented at IEDM 2017, French microelectronics research centre Leti combined FD-SOI technology with its 3D CoolCube monolithic stacking technology to create 4T SRAM bitcells with the same functionality level of 6T bitcells, reducing the memory footprint by 30 percent. 

Another memory design breakthrough detailed in the paper include an embedded Content-Addressable Memory (CAM) dynamically configurable into a Static Random-Access Memory (SRAM) depending on the application running on the SoC.

1) ReCAM organization.

Leti proposes a high-speed 6T-ReCAM/SRAM (Reconfigurable CAM/SRAM) compatible with compact 6T SRAM foundry bitcells, with the CAM and SRAM words orthogonal to each other. Fabricating an 8Kb test-macro in 28nm FD-SOI technology, the authors highlight that their reconfigurable CAM/SRAM outperforms the state of the art, with operations clocked at 1.56GHz (at 0.9V), drawing as little as 0.13fJ/bit energy per search.

TFET DRAM bitcell storing logical '0' and '1'.

TFET DRAM bitcell storing logical '0' and '1'.

On another front, Leti explored the use of advanced Tunnel Field-Effect Transistors (TFET) for the design of a 1T1C refresh-free DRAM and a 2T1C SRAM. The core of these two circuits consists in a reverse-biased (drain-to-source) TFET and a capacitor see figures 2 and 3. Compared to the state of the art, the proposed bitcells show similar density, lower capacitance value, competitive latency and a significantly improved leakage (orders of magnitude lower at under 1fA/bit), note the authors.

“Enabling TFET-based DRAM and integrating TFET standard cells into CMOS designs will allow circuit designers to take advantage of the best features of both technologies”, explained Bastien Giraud, lead author of the paper.


Vous êtes certain ?

Si vous désactivez les cookies, vous ne pouvez plus naviguer sur le site.

Vous allez être rediriger vers Google.