57 research outputs found
GD-Wheel: A Cost-Aware Replacement Policy for Key-Value Stores
No full textVarious memory-based key-value stores, such as Memcached and Redis, are used to speed up dynamic web applications. Specifically, they are used to cache the results of computations, such as database queries. Currently, these key-value stores use either LRU or an LRU approximation as the replacement policy for choosing a key-value pair to be evicted from the store. However, if the cost of recomputing cached values varies significantly, as in the RUBiS and TPC-W benchmarks, then neither of these replacement policies are the best choice. When deciding what key-value pair to replace, it can be advantageous to take the cost of recomputation into consideration. To that end, this thesis proposes a new cost-aware replacement policy, GD-Wheel, which seamlessly integrates recency of access and cost of recomputation. This thesis applies GD-Wheel to Memcached and evaluates its performance using the Yahoo! Cloud Serving Benchmark. The evaluation shows that GD-Wheel, when compared to LRU, greatly reduces the total recomputation cost, as well as the average and 99th percentile read access latency for the application
The first complete chloroplast genome of Haymondia wallichii (Fabaceae) and its phylogenetic analysis
No full textHaymondia wallichii DC. was a scandent shrub. In this study, we sequenced the complete chloroplast (cp) genome of H. wallichii to investigate its phylogenetic relationship in Fabaceae. The total length of the cp genome was 153,668 bp, consisted of a large single copy (LSC) region of 84,310 bp, a small single copy (SSC) region of 17,918 bp, and a pair of inverted repeat regions (IRs) of 25,720 bp. The genome contained 132 genes, namely 37 tRNA genes, 87 protein-coding genes, and 8 rRNA genes. The overall GC content was 35.4%. The phylogenetic analysis indicated H. wallichii was closely related to Pueraria montana var. thomsonii and Pueraria montana var. lobata
Reducing DRAM Row Activations with Eager Read/Write Clustering
No full textThis article describes and evaluates a new approach to optimizing DRAM performance and energy consumption that is based on eagerly writing dirty cache lines to DRAM. Under this approach, many dirty cache lines are written to DRAM before they are evicted. In particular, dirty cache lines that have not been recently accessed are eagerly written to DRAM when the corresponding row has been activated by an ordinary, noneager access, such as a read. This approach enables clustering of reads and writes that target the same row, resulting in a significant reduction in row activations. Specifically, for a variety of applications, it reduces the number of DRAM row activations by an average of 42% and a maximum of 82%. Moreover, the results from a full-system simulator show compelling performance improvements and energy consumption reductions. Out of 23 applications, 6 have overall performance improvements between 10% and 20%, and 3 have improvements in excess of 20%. Furthermore, 12 consume between 10% and 20% less DRAM energy, and 7 have energy consumption reductions in excess of 20%
Improving Approximate Nearest Neighbor Search through Learned Adaptive Early Termination
No full textMolecular Cloning and Functional Analysis of IrUGT86A1-like Gene in Medicinal Plant Isodon rubescens (Hemsl.) Hara
No full textThe synthesis of secondary metabolites in plants often includes glycosylation modifications. Often, the final step of constructing plant secondary metabolites is completed by glycosylation transferases, which are also involved in many cell processes. In this study, a UDP-glycosyltransferase gene (UGT) was amplified from Isodon rubescens (Hemsl.) Hara with RT-PCR and named IrUGT86A1-like (GenBank: MZ913258). Here, we found that IrUGT86A1-like gene is 1450 bp in length and encodes for 479 amino acids. Bioinformatics analysis revealed that IrUGT86A1-like is a stable and hydrophilic protein, located in the cytoplasm with a transmembrane domain. Phylogenetic analysis showed that IrUGT86A1-like protein has the closest genetic relationship with the UDP-glycosyltransferase 86A1-like protein (XP_042054241.1) of Salvia splendens. RT-qPCR analysis demonstrated that the expression of IrUGT86A1-like gene varied in different tissues; leaves had the highest expression followed by flowers, stems, and roots had the lowest expression. This expression trend is similar to the distribution of oridonin content in different tissues of I. rubescens. Additionally, IrUGT86A1-like gene was found to be positively enhanced by NaCl and MeJA treatment, and in contrast was down-regulated by ABA treatment. Finally, the prokaryotic expression vector pEASY®-Blunt E1-IrUGT86A1 was successfully used to express about 53 KD of IrUGT86A1-like protein. This research builds a foundation for further investigation on the function of this gene in the synthesis and modification of secondary metabolites
Random-LTD: Random and Layerwise Token Dropping Brings Efficient Training for Large-scale Transformers
No full textLarge-scale transformer models have become the de-facto architectures for
various machine learning applications, e.g., CV and NLP. However, those large
models also introduce prohibitive training costs. To mitigate this issue, we
propose a novel random and layerwise token dropping method (random-LTD), which
skips the computation of a subset of the input tokens at all middle layers.
Particularly, random-LTD achieves considerable speedups and comparable accuracy
as the standard training baseline. Compared to other token dropping methods,
random-LTD does not require (1) any importance score-based metrics, (2) any
special token treatment (e.g., [CLS]), and (3) many layers in full sequence
length training except the first and the last layers. Besides, a new LayerToken
learning rate schedule is proposed for pretraining problems that resolve the
heavy tuning requirement for our proposed training mechanism. Finally, we
demonstrate that random-LTD can be applied to broader applications, including
GPT and BERT pretraining as well as ViT and GPT finetuning tasks. Our results
show that random-LTD can save about 33.3% theoretical compute cost and 25.6%
wall-clock training time while achieving similar zero-shot evaluations on
GPT-31.3B as compared to baseline.Comment: 22 page
Interband carrier recombination mechanism in Al-doped ZnO
No full textDue to strong nonlinear optical effects and ultrafast response properties, transparent conductive oxides are promising candidate materials for next-generation THz modulation devices and other photonic applications. However, the mechanisms of photon-induced ultrafast carrier relaxations in those materials have yet to be fully understood. Here we investigated the interband carrier dynamics in a prototype of this family, Al-doped ZnO, with different excitation photon energies through femtosecond optical transmittance measurements. We found that the interband relaxation of the photon-induced carriers is mainly through second-order recombination, which is essentially the direct band-to-band recombination with the assistance of intraband electron relaxation. The high density of the doped electrons provides superior intraband relaxation channels through electron-phonon interaction and electron-impurity scattering, which contribute to the overall ultrafast response, especially under indirect and band tail transitions. This study provides a new insight into the ultrafast response mechanism of Al-doped ZnO and the analyzing method may also be generalized to other heavily doped semiconductors.</p
The complete chloroplast genome of Aconitum piepunense (Ranunculaceae) and its phylogenetic analysis
No full textAconitum piepunense belonging to the family Ranunculaceae is an endangered herb species in southwestern China. In this study, the complete chloroplast genome of A. piepunense was sequenced, and the results revealed a typical quadripartite structure with a length of 155,836 bp, including a large single-copy region (LSC, 86,433 bp), a small single-copy region (SSC, 16,945 bp), and a pair of inverted repeat (IR) regions (IRa and IRb, 26,229 bp, respectively). A total of 130 genes were identified in the A. piepunense chloroplast genome, containing 85 protein-coding genes, 37 transfer RNA (tRNA) genes, and 8 ribosomal RNA (rRNA) genes. Phylogenetic analysis based on the maximum likelihood method indicated that A. piepunense formed a monophyletic group, which was sister to A. contortum and A. vilmorinianum
- …
