Approximate Computing with Unreliable Dynamic Memories

Embedded memories account for a large fraction of the overall silicon area and power consumption in modern SoC(s). While embedded memories are typically realized with SRAM, alternative solutions, such as embedded dynamic memories (eDRAM), can provide higher density and/or reduced power consumption. One major challenge that impedes the widespread adoption of eDRAM is that they require frequent refreshes potentially reducing the availability of the memory in periods of high activity and also consuming significant amount of power due to such frequent refreshes. Reducing the refresh rate while on one hand can reduce the power overhead, if not performed in a timely manner, can cause some cells to lose their content potentially resulting in memory errors. In this paper, we consider extending the refresh period of gain-cell based dynamic memories beyond the worst-case point of failure, assuming that the resulting errors can be tolerated when the use-cases are in the domain of inherently error-resilient applications. For example, we observe that for various data mining applications, a large number of memory failures can be accepted with tolerable imprecision in output quality. In particular, our results indicate that by allowing as many as 177 errors in a 16 kB memory, the maximum loss in output quality is 11%. We use this failure limit to study the impact of relaxing reliability constraints on memory availability and retention power for different technologies.

Children's serial recall errors: Implications for theories of short-term memory development

Three experiments examined developmental changes in serial recall of lists of 6 letters, with errors classified as movements, omissions, intrusions, or repetitions. In Experiments 1 and 2, developmental differences between groups of children aged from 7 to 11 years and adults were found in the pattern of serial recall errors. The errors of older participants were more likely to be movements than were those of younger participants, who made more intrusions and omissions. The number of repetition errors did not change with age, and this finding is interpreted in terms of a developmentally invariant postoutput response inhibition process. This interpretation was supported by the findings of Experiment 3, which measured levels of response inhibition in 7-, 9-, and 11-year-olds by comparing recall of lists with and without repeated items. Response inhibition remained developmentally invariant, although older children showed greater response facilitation (improved correct recall of adjacent repeated items). Group differences in the patterns of other errors are accounted for in terms of developmental changes in levels of output forgetting and changes in the efficiency of temporal encoding processes, (C) 2000 Academic Press.

Hand-related rather than goal-related source of gaze-dependent errors in memory-guided reaching

Mechanisms for visuospatial cognition are often inferred directly from errors in behavioral reports of remembered target direction. For example, gaze-centered target representations for reach were first inferred from reach overshoots of target location relative to gaze. Here, we report evidence for the hypothesis that these gaze-dependent reach errors stem predominantly from misestimates of hand rather than target position, as was assumed in all previous studies. Subjects showed typical gaze-dependent overshoots in complete darkness, but these errors were entirely suppressed by continuous visual feedback of the finger. This manipulation could not affect target representations, so the suppressed gaze-dependent errors must have come from misestimates of hand position, likely arising in a gaze-dependent transformation of hand position signals into visual coordinates. This finding has broad implications for any task involving localization of visual targets relative to unseen limbs, in both healthy individuals and patient populations, and shows that response-related transformations cannot be ignored when deducing the sources of gaze-related errors.

An orally available compound prevents deficits in memory caused by the Alzheimer amyloid-B oligomers.

Despite progress in defining a pathogenic role for amyloid beta protein (Abeta) in Alzheimer's disease, orally bioavailable compounds that prevent its effects on hippocampal synaptic plasticity and cognitive function have not yet emerged. A particularly attractive therapeutic strategy is to selectively neutralize small, soluble Abeta oligomers that have recently been shown to mediate synaptic dysfunction. METHODS: Using electrophysiological, biochemical, and behavioral assays, we studied how scyllo-inositol (AZD-103; molecular weight, 180) neutralizes the acutely toxic effects of Abeta on synaptic function and memory recall. RESULTS: Scyllo-inositol, but not its stereoisomer, chiro-inositol, dose-dependently rescued long-term potentiation in mouse hippocampus from the inhibitory effects of soluble oligomers of cell-derived human Abeta. Cerebroventricular injection into rats of the soluble Abeta oligomers interfered with learned performance on a complex lever-pressing task, but administration of scyllo-inositol via the drinking water fully prevented oligomer-induced errors. INTERPRETATION: A small, orally available natural product penetrates into the brain in vivo to rescue the memory impairment produced by soluble Abeta oligomers through a mechanism that restores hippocampal synaptic plasticity.

Gender influences the detection of spatial working memory deficits in bipolar disorder.

Objective: Despite evidence that gender may influence neurocognitive functioning, few studies have examined its effects in bipolar disorder (BD) a priori. The aim of this study was to examine how gender influences executive-type functions, which are potentially useful as endophenotypes for BD. Methods: The performance of 26 euthymic patients(12 males, 14 females) with DSM-IV BD (20 BD type I and six BD type II) was compared to that of 26 controls (12 males, 14 females) on tests of executive function. Controls were matched to patients on an individual basis for sex, age and premorbid IQ. Tests assessed spatial working memory (SWM), planning, attentional set-shifting and verbal fluency. Results: Overall, patients showed deficits in SWM strategy (p < 0.001) and made more SWM errors relative to controls (p < 0.001). These deficits were more apparent in male-only comparisons (both p < 0.001) than in female-only comparisons (both p < 0.05). When examined in isolation, male controls were significantly better at performing the SWM task than female controls (both p < 0.05). This pattern was not observed in the patient cohort: male patients had poorer strategy scores than female patients (p < 0.05), but made a similar number of SWM errors. Conclusions: These findings provide evidence that gender can influence the detection of SWM deficits in the euthymic phase of BD, as the sex-related disequilibrium in SWM identified in healthy controls was disrupted in BD. © 2008 The Authors Journal compilation © 2008 Blackwell Munksgaard.

Errorless learning and memory performance in schizophrenia

There is evidence that patients with schizophrenia have impaired explicit memory and intact implicit memory. The present study sought to replicate and extend that of O'Carroll et al. [O'Carroll, R.E., Russell, H.H., Lawrie, S.M. and Johnstone, E.C., 1999. Errorless learning and the cognitive rehabilitation of memory-impaired schizophrenic patients. Psychological Medicine 29, 105-112.] which reported that for memory-impaired patients with schizophrenia performance on a (cued) word recall task is enhanced using errorless learning techniques (in which errors are prevented during learning) compared to errorful learning (the traditional trial-and-error approach). Thirty patients with a DSM-IV diagnosis of schizophrenia and fifteen healthy controls (HC) participated. The Rivermead Behavioural Memory Test was administered and from their scores, the schizophrenic patients were classified as either memory-impaired (MIS), or memory-unimpaired (MUS). During the training phase two lists of words were learned separately, one using the errorless learning approach and the other using an errorful approach. Subjects were then tested for their recall of the words using cued recall. After errorful learning training, performance on word recall for the MIS group was impaired compared to the MUS and HC groups. However, after errorless learning training, no significant differences in performance were found between the three groups. Errorless learning may play an important role in remediation of cognitive deficits for patients with schizophrenia. (c) 2007 Elsevier Ireland Ltd. All rights reserved.

Probed serial recall in Williams syndrome: Lexical influences on phonological short-term memory

Williams syndrome is a genetic disorder that, it has been claimed, results in an unusual pattern of linguistic strengths and weaknesses. The current study investigated the hypothesis that there is a reduced influence of lexical knowledge on phonological short-term memory in Williams syndrome. Fourteen children with Williams syndrome and 2 vocabulary la matched control groups, 20 typically developing children and 13 children with learning difficulties, were tested on 2 probed serial-recall tasks. On the basis of previous findings, it was predicted that children with Williams syndrome would demonstrate (a) a reduced effect of lexicality on the recall of list items, (b) relatively poorer recall of list items compared with recall of serial order, and (c) a reduced tendency to produce lexicalization errors in the recall of nonwords. in fact, none of these predictions were supported. Alternative explanations for previous findings and implications for accounts of language development in Williams syndrome are discussed.

Logic and Memory Design Based on Unequal Error Protection for Voltage-scalable, Robust and Adaptive DSP Systems

In this paper, we propose a system level design approach considering voltage over-scaling (VOS) that achieves error resiliency using unequal error protection of different computation elements, while incurring minor quality degradation. Depending on user specifications and severity of process variations/channel noise, the degree of VOS in each block of the system is adaptively tuned to ensure minimum system power while providing "just-the-right" amount of quality and robustness. This is achieved, by taking into consideration block level interactions and ensuring that under any change of operating conditions, only the "less-crucial" computations, that contribute less to block/system output quality, are affected. The proposed approach applies unequal error protection to various blocks of a system-logic and memory-and spans multiple layers of design hierarchy-algorithm, architecture and circuit. The design methodology when applied to a multimedia subsystem shows large power benefits ( up to 69% improvement in power consumption) at reasonable image quality while tolerating errors introduced due to VOS, process variations, and channel noise.