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.
Testing Disagreement Models CHANG, YEN‐CHENG; HSIAO, PEI‐JIE; LJUNGQVIST, ALEXANDER ...
The Journal of finance (New York),
August 2022, Letnik:
77, Številka:
4
Journal Article
Recenzirano
Odprti dostop
ABSTRACT
We provide plausibly identified evidence for the role of investor disagreement in asset pricing. Our natural experiment exploits the staggered implementation of the Electronic Data ...Gathering, Analysis, and Retrieval (EDGAR) system, which induces a reduction in investor disagreement. Consistent with models of investor disagreement, EDGAR inclusion helps resolve disagreement around information events, leading to stock price corrections. The reduction in disagreement following EDGAR inclusion also reduces stock price crash risk, especially among stocks with binding short‐sale constraints and high investor optimism.
Convolutional neural network (CNN) offers significant accuracy in image detection. To implement image detection using CNN in the Internet of Things (IoT) devices, a streaming hardware accelerator is ...proposed. The proposed accelerator optimizes the energy efficiency by avoiding unnecessary data movement. With unique filter decomposition technique, the accelerator can support arbitrary convolution window size. In addition, max-pooling function can be computed in parallel with convolution by using separate pooling unit, thus achieving throughput improvement. A prototype accelerator was implemented in TSMC 65-nm technology with a core size of 5 mm 2 . The accelerator can support major CNNs and achieve 152GOPS peak throughput and 434GOPS/W energy efficiency at 350 mW, making it a promising hardware accelerator for intelligent IoT devices.
Abstract
Field-induced ionic motions in all-inorganic CsPbBr
3
perovskite quantum dots (QDs) strongly dictate not only their electro-optical characteristics but also the ultimate optoelectronic ...device performance. Here, we show that the functionality of a single Ag/CsPbBr
3
/ITO device can be actively switched on a sub-millisecond scale from a resistive random-access memory (RRAM) to a light-emitting electrochemical cell (LEC), or vice versa, by simply modulating its bias polarity. We then realize for the first time a fast, all-perovskite light-emitting memory (LEM) operating at 5 kHz by pairing such two identical devices in series, in which one functions as an RRAM to electrically read the encoded data while the other simultaneously as an LEC for a parallel, non-contact optical reading. We further show that the digital status of the LEM can be perceived in real time from its emission color. Our work opens up a completely new horizon for more advanced all-inorganic perovskite optoelectronic technologies.
This work presents a novel photo‐electrochemical architecture based on the 3D pyramid‐like graphene/p‐Si Schottky junctions. Overcoming the conventional transfer technique by which only planar ...graphene/Si Schottky junctions are currently available, this work demonstrates the 3D pyramid‐like graphene/p‐Si Schottky junction photocathode, which greatly enhances light harvesting efficiency and exhibits promising photo‐electrochemical performance for hydrogen generation. The formation of 3D pyramid‐like graphene/p‐Si Schottky junctions exhibits enhanced electrochemical activity and promotes charge separation efficiency compared with the bare pyramid Si surface without graphene. The inherent chemical inertness of graphene significantly improves the operational stability of 3D graphene/p‐Si Schottky junction photo‐electrochemical cells. The 3D pyramid‐like graphene/p‐Si Schottky junction photocathode delivers an onset potential of 0.41 V and a saturated photocurrent density of −32.5 mA cm−2 at 0 V (vs RHE) with excellent stability comparable to values reported for textured or nanostructured p‐Si photocathodes coated with ultrathin oxide layers by the conventional atomic layer deposition technique. These results suggest that the formation of graphene/Si Schottky junctions with a 3D architecture is a promising approach to improve the performance and durability of Si‐based photo‐electrochemical systems for water splitting or solar‐to‐fuel conversion.
This work demonstrates a novel 3D pyramid‐like graphene/p‐Si Schottky junction photocathode for H2 production based on the unique advantages of excellent carrier transport, high transparency, and superior corrosion protection of graphene. The formation of graphene/Si Schottky junctions with 3D architecture is a promising approach to improve the performance and durability of Si‐based photo‐electrochemical systems for water splitting or solar‐to‐fuel conversion.
Abstract
Most chemical vapor deposition methods for transition metal dichalcogenides use an extremely small amount of precursor to render large single-crystal flakes, which usually causes low ...coverage of the materials on the substrate. In this study, a self-capping vapor-liquid-solid reaction is proposed to fabricate large-grain, continuous MoS
2
films. An intermediate liquid phase-Na
2
Mo
2
O
7
is formed through a eutectic reaction of MoO
3
and NaF, followed by being sulfurized into MoS
2
. The as-formed MoS
2
seeds function as a capping layer that reduces the nucleation density and promotes lateral growth. By tuning the driving force of the reaction, large mono/bilayer (1.1 mm/200 μm) flakes or full-coverage films (with a record-high average grain size of 450 μm) can be grown on centimeter-scale substrates. The field-effect transistors fabricated from the full-coverage films show high mobility (33 and 49 cm
2
V
−1
s
−1
for the mono and bilayer regions) and on/off ratio (1 ~ 5 × 10
8
) across a 1.5 cm × 1.5 cm region.
Computing-in-memory (CIM) based on embedded nonvolatile memory is a promising candidate for energy-efficient multiply-and-accumulate (MAC) operations in artificial intelligence (AI) edge devices. ...However, circuit design for NVM-based CIM (nvCIM) imposes a number of challenges, including an arealatency-energy tradeoff for multibit MAC operations, patterndependent degradation in signal margin, and small read margin. To overcome these challenges, this article proposes the following: 1) a serial-input non-weighted product (SINWP) structure; 2) a down-scaling weighted current translator (DSWCT) and positive-negative current-subtractor (PN-ISUB); 3) a currentaware bitline clamper (CABLC) scheme; and 4) a triple-margin small-offset current-mode sense amplifier (TMCSA). A 55-nm 1-Mb ReRAM-CIM macro was fabricated to demonstrate the MAC operation of 2-b-input, 3-b-weight with 4-b-out. This nvCIM macro achieved T MAC = 14.6 ns at 4-b-out with peak energy efficiency of 53.17 TOPS/W.
Electronic nicotine delivery systems (ENDS) or e-cigarettes have emerged as a popular recreational tool among adolescents and adults. Although the use of ENDS is often promoted as a safer alternative ...to conventional cigarettes, few comprehensive studies have assessed the long-term effects of vaporized nicotine and its associated solvents, propylene glycol (PG) and vegetable glycerin (VG). Here, we show that compared with smoke exposure, mice receiving ENDS vapor for 4 months failed to develop pulmonary inflammation or emphysema. However, ENDS exposure, independent of nicotine, altered lung lipid homeostasis in alveolar macrophages and epithelial cells. Comprehensive lipidomic and structural analyses of the lungs revealed aberrant phospholipids in alveolar macrophages and increased surfactant-associated phospholipids in the airway. In addition to ENDS-induced lipid deposition, chronic ENDS vapor exposure downregulated innate immunity against viral pathogens in resident macrophages. Moreover, independent of nicotine, ENDS-exposed mice infected with influenza demonstrated enhanced lung inflammation and tissue damage. Together, our findings reveal that chronic e-cigarette vapor aberrantly alters the physiology of lung epithelial cells and resident immune cells and promotes poor response to infectious challenge. Notably, alterations in lipid homeostasis and immune impairment are independent of nicotine, thereby warranting more extensive investigations of the vehicle solvents used in e-cigarettes.
Background
Reconstruction of total or near‐total nasal defects is challenging and requires the re‐creation of three nasal layers. Fasciocutaneous free flaps have been used effectively for restoring ...the vascularized nasal lining and staged forehead flap for nasal skin replacement, which is a long process. The aim of this study is to share our experience of combination of preliminary free tissue transfer with paramedian forehead flap reconstruction in the same stage of complex nasal reconstruction.
Methods
From December 2015 to July 2021, 10 patients underwent nasal reconstruction with free flaps including 4 medial sural artery perforator (MSAP) flaps, 5 anterolateral thigh (ALT) flaps and 1 radial forearm flap for lining and forehead flaps for skin coverage simultaneously for total or subtotal nasal defects. Nasal obstruction symptoms evaluation (NOSE) score was utilized to evaluate the functional outcome and the aesthetic results were evaluated with the last follow‐up photos with score 1–5 by 5 plastic surgeon and 5 laypersons.
Results
The size of the free flaps ranged from 3 cm x 6 cm to 6 cm x 13 cm. After excluding one patient who expired before forehead flap division due to comorbidities, the average duration between combination surgery and the division of the forehead flap pedicle of the remaining patients was 5.7 months (range, 2–12). For patients without any postoperative events, the duration was 2.2 months (range, 2–3). One free flap had partial necrosis due to infection. The average follow‐up duration was 29.6 months (range, 12–64). The NOSE score was 5.9 (range, 0–10) and the aesthetic score is 4.1 (range, 3–5) in average.
Conclusions
The combination of preliminary free tissue transfer for nasal lining restoration with a paramedian forehead flap for nasal skin replacement in the same stage may shorten the long process and achieve satisfactory reconstruction in complex nasal reconstruction.
Abstract
We show that analyst behavior changes in response to a randomly assigned shock that exogenously varies the timeliness and cost of accessing mandatory disclosures in the cross-section of ...investors: analysts reduce coverage and issue less optimistic, more accurate, less bold, and less informative forecasts. Our evidence indicates that analysts reduce a strategic component of their behavior: the changes are stronger among analysts with more strategic incentives like affiliated or retail-focused analysts. We conclude that mandatory disclosure can substitute for analyst information production, which is constrained by investors’ ability to verify forecasts using corporate filings.
Authors have furnished an Internet Appendix, which is available on the Oxford University Press Web site next to the link to the final published paper online.