Computation-in-memory (CIM) is a promising candidate to improve the energy efficiency of multiply-and-accumulate (MAC) operations of artificial intelligence (AI) chips. This work presents an static ...random access memory (SRAM) CIM unit-macro using: 1) compact-rule compatible twin-8T (T8T) cells for weighted CIM MAC operations to reduce area overhead and vulnerability to process variation; 2) an even-odd dual-channel (EODC) input mapping scheme to extend input bandwidth; 3) a two's complement weight mapping (C2WM) scheme to enable MAC operations using positive and negative weights within a cell array in order to reduce area overhead and computational latency; and 4) a configurable global-local reference voltage generation (CGLRVG) scheme for kernels of various sizes and bit precision. A 64 × 60 b T8T unit-macro with 1-, 2-, 4-b inputs, 1-, 2-, 5-b weights, and up to 7-b MAC-value (MACV) outputs was fabricated as a test chip using a foundry 55-nm process. The proposed SRAM-CIM unit-macro achieved access times of 5 ns and energy efficiency of 37.5-45.36 TOPS/W under 5-b MACV output.
This article presents a computing-in-memory (CIM) structure aimed at improving the energy efficiency of edge devices running multi-bit multiply-and-accumulate (MAC) operations. The proposed scheme ...includes a 6T SRAM-based CIM (SRAM-CIM) macro capable of: 1) weight-bitwise MAC (WbwMAC) operations to expand the sensing margin and improve the readout accuracy for high-precision MAC operations; 2) a compact 6T local computing cell to perform multiplication with suppressed sensitivity to process variation; 3) an algorithm-adaptive low MAC-aware readout scheme to improve energy efficiency; 4) a bitline header selection scheme to enlarge signal margin; and 5) a small-offset margin-enhanced sense amplifier for robust read operations against process variation. A fabricated 28-nm 64-kb SRAM-CIM macro achieved access times of 4.1-8.4 ns with energy efficiency of 11.5-68.4 TOPS/W, while performing MAC operations with 4- or 8-b input and weight precision.
We demonstrate the interplay of pure spin current, spin-polarized current, and spin fluctuation in 3d NixCu1–x. By tuning the compositions of the NixCu1–x alloys, we separate the effects due to the ...pure spin current and spin-polarized current. By exploiting the interaction of spin current with spin fluctuation in suitable Ni-Cu alloys, we obtain an unprecedentedly high spin Hall angle of 46%, about 5 times larger than that in Pt, at room temperature. Furthermore, we show that spin-dependent thermal transport via anomalous Nernst effect can serve as a sensitive magnetometer to electrically probe the magnetic phase transitions in thin films with in-plane anisotropy. As a result, the enhancement of spin Hall angle by exploiting spin current fluctuation via composition control makes 3d magnets functional materials in charge-to-spin conversion for spintronic application.
Despite its important role in understanding ultrafast spin dynamics and revealing novel spin/orbit effects, the mechanism of the terahertz (THz) emission from a single ferromagnetic nanofilm upon a ...femtosecond laser pump still remains elusive. Recent experiments have shown exotic symmetry, which is not expected from the routinely adopted mechanism of ultrafast demagnetization. Here, by developing a bidirectional pump-THz emission spectroscopy and associated symmetry analysis method, we set a benchmark for the experimental distinction of the THz emission induced by various mechanisms. Our results unambiguously unveil a new mechanismanomalous Nernst effect (ANE) induced THz emission due to the ultrafast temperature gradient created by a femtosecond laser. Quantitative analysis shows that the THz emission exhibits interesting thickness dependence where different mechanisms dominate at different thickness ranges. Our work not only clarifies the origin of the ferromagnetic-based THz emission but also offers a fertile platform for investigating the ultrafast optomagnetism and THz spintronics.
Co/Ru/L11-CoPt based thin film as pseudo spin valve prepared by DC magnetron sputtering deposition. The magnetic and magnetoresistance properties with the thickness change of L11-CoPt layer were ...investigated. The resulting spin valve exhibited well-different out-of-plane and in-plane hysteresis loops. The coercivities changed unpredictably at different thicknesses of L11-CoPt layer. A perpendicular giant magnetoresistance ratio is obtained about 59.70 % and 160 % with current-out-of-plane measurements for the pattern and cross-linking structures of Pt(20 nm)/Co(2.5 nm)/Ru(1.5 nm)/L11-CoPt8/Pt(20 nm)/glass substrate, respectively. It was ascribed this giant magnetoresistance to spin-dependent transmission of the conduction electrons between Co free layer and L11-CoPt reference layer through Ru spacer layer, and considered as the ferromagnetic interlayer exchange coupling.
Display omitted
•Preparing Co/Pt8 multilayer based pseudo spin valves (PSV) by deposition method.•Determining development of giant magnetoresistance (GMR) in the pattern structure.•Observing enhancement of GMR in the cross-linking structure.
We here demonstrate the multifunctional properties of atomically thin heterojunctions that are enabled by their strong interfacial interactions and their application toward self-powered sensors with ...unprecedented performance. Bonding between tin diselenide and graphene produces thermoelectric and mechanoelectric properties beyond the ability of either component. A record-breaking ZT of 2.43 originated from the synergistic combination of graphene’s high carrier conductivity and SnSe2-mediated thermal conductivity lowering. Moreover, spatially varying interaction at the SnSe2/graphene interface produces stress localization that results in a novel 2D-crack-assisted strain sensing mechanism whose sensitivity (GF = 450) is superior to all other 2D materials. Finally, a graphene-assisted growth process permits the formation of high-quality heterojunctions directly on polymeric substrates for flexible and transparent sensors that achieve self-powered strain sensing from a small temperature gradient. Our work enhances the fundamental understanding of multifunctionality at the atomic scale and provides a route toward structural health monitoring through ubiquitous and smart devices.
Computation-in-memory (CIM) is a promising avenue to improve the energy efficiency of multiply-and-accumulate (MAC) operations in AI chips. Multi-bit CNNs are required for high-inference accuracy in ...many applications 1-5. There are challenges and tradeoffs for SRAM-based CIM: (1) tradeoffs between signal margin, cell stability and area overhead; (2) the high-weighted bit process variation dominates the end-result error rate; (3) trade-off between input bandwidth, speed and area. Previous SRAM CIM macros were limited to binary MAC operations for fully connected networks 1, or they used CIM for multiplication 2 or weight-combination operations 3 with additional large-area near-memory computing (NMC) logic for summation or MAC operations.
The development of security-aware mobile devices using wide-input-output (IO) nonvolatile memory (NVM) is hindered by high peak current, large area overhead for high read bandwidth (BWR), and ...considerable energy consumption for data movement between NVM and logic blocks. Furthermore, data stored in NVM are vulnerable to reverse-engineering attacks. This work presents a high BWR security-aware near-memory-computing spin-transfer torque magnetic random-access memory (STT-MRAM) macro using a multi-bit current-mode sense amplifier (MB-CSA) to reduce peak current and energy consumption for wide-IO access, a near-memory shift-and-rotate functionality (NSRF) in conjunction with the MB-CSA to reduce area overhead and enable the completion of read and logic operations within a single cycle, and a reverse-engineering-proof XOR-based memory data protector to protect data stored in NVM against reverse-engineering attacks. A 1-Mb 1024-b read STT-MRAM macro with data protector fabricated using foundry embedded 22-nm STT-MRAM. This work achieved 42.67 GB/s for BWR and 0.23 pJ/b. Inclusion of the NSRF circuit reduced area overhead by 33.3% while increasing latency by only 170 ps.
The demagnetizing field associated with shape anisotropy originates from the magnetic field generated through magnetization of magnetic materials. The shape and demagnetizing effects must be ...considered as characteristics of a magnetic object in the presence of a magnetic field. In this study, we demonstrated the strong influence of the demagnetizing field on the spin current transports in Pt/YIG, including the spin Seebeck effect, spin Hall effect, spin Hall magnetoresistance, and planar Hall resistance. The calculated effective demagnetizing factors were closely related to the thickness- and width-dependent anomalous plateau behaviors of the spin-dependent electrical and thermal transports as well as magneto-optical measurements. Moreover, we demonstrated that the width of the plateau manifested from the shape of the YIG sample. Furthermore, through magnetic force microscopy, we directly visualized the evolution of the magnetic domains with an abrupt 90° magnetic rotation, which we further corroborated by the 90° resistance phase shift in the angular-dependent planar Hall effect.
Two-dimensional (2D) thermoelectrics have shown enhanced performance compared to bulk thermoelectrics but cannot easily be improved through nanostructuring, since phonon-boundary-scattering becomes ...ineffective under confined dimensions. To overcome this issue, we introduce a one-dimensional (1D) texturing approach that efficiently suppresses phonon transport between grains of 2D thermoelectrics (TE). Uniaxial graphene wrinkles were utilized to separate high-quality, ultrathin Bi2Se3 films in a highly directional manner. The TE properties of these textured samples exhibit a strongly anisotropic characteristic with a maximum ZT of 1.03 at room temperature, which represents the highest reported value for Bi2Se3. The texturing approach provides a route for the enhancement of 2D thermoelectrics for future applications.