Journal articles on the topic 'Domain-Specific Accelerator'
To see the other types of publications on this topic, follow the link: Domain-Specific Accelerator.
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 50 journal articles for your research on the topic 'Domain-Specific Accelerator.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Browse journal articles on a wide variety of disciplines and organise your bibliography correctly.
1
Cong, Jason, Mohammad Ali Ghodrat, Michael Gill, Beayna Grigorian, and Glenn Reinman. "Architecture Support for Domain-Specific Accelerator-Rich CMPs." ACM Transactions on Embedded Computing Systems 13, no. 4s (July 2014): 1–26. http://dx.doi.org/10.1145/2584664.
Full text
2
Sotiriou-Xanthopoulos, Efstathios, Sotirios Xydis, Kostas Siozios, George Economakos, and Dimitrios Soudris. "A Framework for Interconnection-Aware Domain-Specific Many-Accelerator Synthesis." ACM Transactions on Embedded Computing Systems 16, no. 1 (November 3, 2016): 1–26. http://dx.doi.org/10.1145/2983624.
Full text
3
Feng, Fan, Li Li, Kun Wang, Yuxiang Fu, Guoqiang He, and Hongbing Pan. "Design and Application Space Exploration of a Domain-Specific Accelerator System." Electronics 7, no. 4 (March 29, 2018): 45. http://dx.doi.org/10.3390/electronics7040045.
Full text
4
Sunny, Febin P., Asif Mirza, Mahdi Nikdast, and Sudeep Pasricha. "ROBIN: A Robust Optical Binary Neural Network Accelerator." ACM Transactions on Embedded Computing Systems 20, no. 5s (October 31, 2021): 1–24. http://dx.doi.org/10.1145/3476988.
Full textAbstract:
Domain specific neural network accelerators have garnered attention because of their improved energy efficiency and inference performance compared to CPUs and GPUs. Such accelerators are thus well suited for resource-constrained embedded systems. However, mapping sophisticated neural network models on these accelerators still entails significant energy and memory consumption, along with high inference time overhead. Binarized neural networks (BNNs), which utilize single-bit weights, represent an efficient way to implement and deploy neural network models on accelerators. In this paper, we present a novel optical-domain BNN accelerator, named ROBIN , which intelligently integrates heterogeneous microring resonator optical devices with complementary capabilities to efficiently implement the key functionalities in BNNs. We perform detailed fabrication-process variation analyses at the optical device level, explore efficient corrective tuning for these devices, and integrate circuit-level optimization to counter thermal variations. As a result, our proposed ROBIN architecture possesses the desirable traits of being robust, energy-efficient, low latency, and high throughput, when executing BNN models. Our analysis shows that ROBIN can outperform the best-known optical BNN accelerators and many electronic accelerators. Specifically, our energy-efficient ROBIN design exhibits energy-per-bit values that are ∼4 × lower than electronic BNN accelerators and ∼933 × lower than a recently proposed photonic BNN accelerator, while a performance-efficient ROBIN design shows ∼3 × and ∼25 × better performance than electronic and photonic BNN accelerators, respectively.
5
Belda, María José, Katzalin Olcoz, Fernando Castro, and Francisco Tirado. "Optimization of a line detection algorithm for autonomous vehicles on a RISC-V with accelerator." Journal of Computer Science and Technology 22, no. 2 (October 17, 2022): e10. http://dx.doi.org/10.24215/16666038.22.e10.
Full textAbstract:
In recent years, autonomous vehicles have attracted theattention of many research groups, both in academiaand business, including researchers from leading com-panies such as Google, Uber and Tesla. This type ofvehicles are equipped with systems that are subjectto very strict requirements, essentially aimed at per-forming safe operations –both for potential passengersand pedestrians– as well as carrying out the process-ing needed for decision making in real time. In manyinstances, general-purpose processors alone cannotensure that these safety, reliability and real-time re-quirements are met, so it is common to implementheterogeneous systems by including accelerators. Thispaper explores the acceleration of a line detection ap-plication in the autonomous car environment using aheterogeneous system consisting of a general-purposeRISC-V core and a domain-specific accelerator. In par-ticular, the application is analyzed to identify the mostcomputationally intensive parts of the code and it isadapted accordingly for more efficient processing. Fur-thermore, the code is executed on the aforementionedhardware platform to verify that the execution effec-tively meets the existing requirements in autonomousvehicles, experiencing a 3.7x speedup with respect torunning without accelerator.
6
Vretenar, M., A. Mamaras, G. Bisoffi, and P. Foka. "Production of radioisotopes for cancer imaging and treatment with compact linear accelerators." Journal of Physics: Conference Series 2420, no. 1 (January 1, 2023): 012104. http://dx.doi.org/10.1088/1742-6596/2420/1/012104.
Full textAbstract:
Abstract Accelerator-produced radioisotopes are widely used in modern medicine, for imaging, for cancer therapy, and for combinations of therapy and diagnostics (theragnostics). Clinical trials are well advanced for several radioisotope-based treatments that might open the way to a strong request of specific accelerator systems dedicated to radioisotope production. While cyclotrons are the standard tool in this domain, we explore here alternative options using linear accelerators. Compared to cyclotrons, linacs have the advantage of modularity, compactness, and reduced beam loss with lower shielding requirements. Although in general more expensive than cyclotrons, linacs are competitive in cost for production of low-energy proton beams, or of intense beams of heavier particles. After a review of radioisotopes of potential interest, in particular produced with low-energy protons or helium, this paper presents two linac-based isotope production systems. The first is a compact RFQ-based system for PET (Positron Emission Tomography) isotopes, and the second is an alpha-particle linac for production of alpha-emitters. The accelerator systems are described, together with calculations of production yields for different targets.
7
Huang, Shizhen, Enhao Tang, Shun Li, Xiangzhan Ping, and Ruiqi Chen. "Hardware-friendly compression and hardware acceleration for transformer: A survey." Electronic Research Archive 30, no. 10 (2022): 3755–85. http://dx.doi.org/10.3934/era.2022192.
Full textAbstract:
<abstract> <p>The transformer model has recently been a milestone in artificial intelligence. The algorithm has enhanced the performance of tasks such as Machine Translation and Computer Vision to a level previously unattainable. However, the transformer model has a strong performance but also requires a high amount of memory overhead and enormous computing power. This significantly hinders the deployment of an energy-efficient transformer system. Due to the high parallelism, low latency, and low power consumption of field-programmable gate arrays (FPGAs) and application specific integrated circuits (ASICs), they demonstrate higher energy efficiency than Graphics Processing Units (GPUs) and Central Processing Units (CPUs). Therefore, FPGA and ASIC are widely used to accelerate deep learning algorithms. Several papers have addressed the issue of deploying the Transformer on dedicated hardware for acceleration, but there is a lack of comprehensive studies in this area. Therefore, we summarize the transformer model compression algorithm based on the hardware accelerator and its implementation to provide a comprehensive overview of this research domain. This paper first introduces the transformer model framework and computation process. Secondly, a discussion of hardware-friendly compression algorithms based on self-attention and Transformer is provided, along with a review of a state-of-the-art hardware accelerator framework. Finally, we considered some promising topics in transformer hardware acceleration, such as a high-level design framework and selecting the optimum device using reinforcement learning.</p> </abstract>
8
Fang, Jian, Yvo T. B. Mulder, Jan Hidders, Jinho Lee, and H. Peter Hofstee. "In-memory database acceleration on FPGAs: a survey." VLDB Journal 29, no. 1 (October 26, 2019): 33–59. http://dx.doi.org/10.1007/s00778-019-00581-w.
Full textAbstract:
Abstract While FPGAs have seen prior use in database systems, in recent years interest in using FPGA to accelerate databases has declined in both industry and academia for the following three reasons. First, specifically for in-memory databases, FPGAs integrated with conventional I/O provide insufficient bandwidth, limiting performance. Second, GPUs, which can also provide high throughput, and are easier to program, have emerged as a strong accelerator alternative. Third, programming FPGAs required developers to have full-stack skills, from high-level algorithm design to low-level circuit implementations. The good news is that these challenges are being addressed. New interface technologies connect FPGAs into the system at main-memory bandwidth and the latest FPGAs provide local memory competitive in capacity and bandwidth with GPUs. Ease of programming is improving through support of shared coherent virtual memory between the host and the accelerator, support for higher-level languages, and domain-specific tools to generate FPGA designs automatically. Therefore, this paper surveys using FPGAs to accelerate in-memory database systems targeting designs that can operate at the speed of main memory.
9
Hosseini, Morteza, and Tinoosh Mohsenin. "Binary Precision Neural Network Manycore Accelerator." ACM Journal on Emerging Technologies in Computing Systems 17, no. 2 (April 2021): 1–27. http://dx.doi.org/10.1145/3423136.
Full textAbstract:
This article presents a low-power, programmable, domain-specific manycore accelerator, Binarized neural Network Manycore Accelerator (BiNMAC), which adopts and efficiently executes binary precision weight/activation neural network models. Such networks have compact models in which weights are constrained to only 1 bit and can be packed several in one memory entry that minimizes memory footprint to its finest. Packing weights also facilitates executing single instruction, multiple data with simple circuitry that allows maximizing performance and efficiency. The proposed BiNMAC has light-weight cores that support domain-specific instructions, and a router-based memory access architecture that helps with efficient implementation of layers in binary precision weight/activation neural networks of proper size. With only 3.73% and 1.98% area and average power overhead, respectively, novel instructions such as Combined Population-Count-XNOR , Patch-Select , and Bit-based Accumulation are added to the instruction set architecture of the BiNMAC, each of which replaces execution cycles of frequently used functions with 1 clock cycle that otherwise would have taken 54, 4, and 3 clock cycles, respectively. Additionally, customized logic is added to every core to transpose 16×16-bit blocks of memory on a bit-level basis, that expedites reshaping intermediate data to be well-aligned for bitwise operations. A 64-cluster architecture of the BiNMAC is fully placed and routed in 65-nm TSMC CMOS technology, where a single cluster occupies an area of 0.53 mm 2 with an average power of 232 mW at 1-GHz clock frequency and 1.1 V. The 64-cluster architecture takes 36.5 mm 2 area and, if fully exploited, consumes a total power of 16.4 W and can perform 1,360 Giga Operations Per Second (GOPS) while providing full programmability. To demonstrate its scalability, four binarized case studies including ResNet-20 and LeNet-5 for high-performance image classification, as well as a ConvNet and a multilayer perceptron for low-power physiological applications were implemented on BiNMAC. The implementation results indicate that the population-count instruction alone can expedite the performance by approximately 5×. When other new instructions are added to a RISC machine with existing population-count instruction, the performance is increased by 58% on average. To compare the performance of the BiNMAC with other commercial-off-the-shelf platforms, the case studies with their double-precision floating-point models are also implemented on the NVIDIA Jetson TX2 SoC (CPU+GPU). The results indicate that, within a margin of ∼2.1%--9.5% accuracy loss, BiNMAC on average outperforms the TX2 GPU by approximately 1.9× (or 7.5× with fabrication technology scaled) in energy consumption for image classification applications. On low power settings and within a margin of ∼3.7%--5.5% accuracy loss compared to ARM Cortex-A57 CPU implementation, BiNMAC is roughly ∼9.7×--17.2× (or 38.8×--68.8× with fabrication technology scaled) more energy efficient for physiological applications while meeting the application deadline.
10
Gundi, Noel Daniel, Pramesh Pandey, Sanghamitra Roy, and Koushik Chakraborty. "Implementing a Timing Error-Resilient and Energy-Efficient Near-Threshold Hardware Accelerator for Deep Neural Network Inference." Journal of Low Power Electronics and Applications 12, no. 2 (June 6, 2022): 32. http://dx.doi.org/10.3390/jlpea12020032.
Full textAbstract:
Increasing processing requirements in the Artificial Intelligence (AI) realm has led to the emergence of domain-specific architectures for Deep Neural Network (DNN) applications. Tensor Processing Unit (TPU), a DNN accelerator by Google, has emerged as a front runner outclassing its contemporaries, CPUs and GPUs, in performance by 15×–30×. TPUs have been deployed in Google data centers to cater to the performance demands. However, a TPU’s performance enhancement is accompanied by a mammoth power consumption. In the pursuit of lowering the energy utilization, this paper proposes PREDITOR—a low-power TPU operating in the Near-Threshold Computing (NTC) realm. PREDITOR uses mathematical analysis to mitigate the undetectable timing errors by boosting the voltage of the selective multiplier-and-accumulator units at specific intervals to enhance the performance of the NTC TPU, thereby ensuring a high inference accuracy at low voltage. PREDITOR offers up to 3×–5× improved performance in comparison to the leading-edge error mitigation schemes with a minor loss in accuracy.
11
Priestley, Maria, Elena Simperl, Cristina Juc, and María Anguiano. "Measuring the impact of publicly funded open innovation programmes: the case of Data Market Services Accelerator." Open Research Europe 1 (July 28, 2022): 71. http://dx.doi.org/10.12688/openreseurope.13621.3.
Full textAbstract:
One of the current goals of the European Commission is to stimulate the development and uptake of data and AI technologies in the economy. Earlier foundations of this work included initiatives to promote the publication of open data in the public domain. More recently, the Commission’s attention has been shifting to open innovation programmes that help startups and small-medium enterprises (SMEs) to develop the capacity to engage with the latest technical and regulatory trends, and to share their data innovations with other organisations. In order to assess the efficacy and impact of such initiatives, each programme’s specific social and economic objectives must be taken into consideration. As an example of how this can be done in practice, our paper presents the motivating objectives and methodological approaches that were used to assess the impact of the Data Market Services Accelerator (DMS), an EU-funded initiative for data-centric companies. We evaluated the performance of the programme in terms of its effect on the market, fundraising capabilities of companies, innovation, and socio-economic aspects. In addition to assessing how DMS was able to meet its intended objectives, our examination also underscored current challenges related to specific outcomes that are meaningful to the European Commission, including data standardisation and long-term legal strategy. We conclude the paper with a series of recommendations to support the impact assessment efforts of other similar innovation programmes.
12
Curcio, Alessandro, Alessandro Cianchi, Gemma Costa, Francesco Demurtas, Michael Ehret, Massimo Ferrario, Mario Galletti, Danilo Giulietti, José Antonio Pérez-Hernández, and Giancarlo Gatti. "Performance Study on a Soft X-ray Betatron Radiation Source Realized in the Self-Injection Regime of Laser-Plasma Wakefield Acceleration." Applied Sciences 12, no. 23 (December 6, 2022): 12471. http://dx.doi.org/10.3390/app122312471.
Full textAbstract:
We present an analysis of the performance of a broadband secondary radiation source based on a high-gradient laser-plasma wakefield electron accelerator. In more detail, we report studies of compact and ultra-short X-ray generation via betatron oscillations in plasma channels. For the specific working point examined in this paper, determined by the needs of other experiments ongoing at the facility, at ∼0.02 Hz operation rate, we have found ≲106 photons emitted per shot (with a fluctuation of 50%) in the soft X-rays, corresponding to a critical energy of ∼0.8 keV (with a fluctuation of 40%). The source will be implemented for experiments in time-domain spectroscopy, e.g., biological specimens, and for other applications oriented to medical physics.
13
Reinehr Gobatto, Leonardo, Pablo Rodrigues, Mateus Saquetti Pereira de Carvalho Tirone, Weverton Luis da Costa Cordeiro, and José Rodrigo Furlanetto Azambuja. "Programmable Data Planes meets In-Network Computing: A Review of the State of the Art and Prospective Directions." Journal of Integrated Circuits and Systems 16, no. 2 (August 17, 2021): 1–8. http://dx.doi.org/10.29292/jics.v16i2.497.
Full textAbstract:
Improving network traffic in networks is one of the concerns between networking researchers and network operators since the architecture of modern networks still faces challenges to process large data traffic without the cost of consuming a significant amount of resources not related to computing specifically. On the other hand, network programmability has enabled the development of new applications and network services, from software-defined networking to domain-specific languages created to program network devices and specify their behavior. The development of programmable hardware and hardware accelerators like FPGAs, GPUs, and CPUs help this new paradigm go one step further. Use the artifact of programmability of these devices to solve problems, such as improve the processing of data traffic is the key of in-network computing. It offers the opportunity to execute programs typically running on end-hosts within programmable network devices already incorporated on the network, thus being capable of provides a reduction on the in-network processing load and requires no extra cost, since operations can be concluded using a fewer amount of devices of the network and no extra device are needed. In this paper, we survey in-network computing, as well as we suggest classifying related works to in-network computing according to the hardware accelerator used. Also, we discuss challenges and research directions.
14
Dally, William J., Yatish Turakhia, and Song Han. "Domain-specific hardware accelerators." Communications of the ACM 63, no. 7 (June 18, 2020): 48–57. http://dx.doi.org/10.1145/3361682.
Full text
15
Boigelot, Bernard. "Domain-specific regular acceleration." International Journal on Software Tools for Technology Transfer 14, no. 2 (July 20, 2011): 193–206. http://dx.doi.org/10.1007/s10009-011-0206-x.
Full text
16
Pudi, Dhilleswararao, Samuel Jigme Harrison, Dimitrios Stathis, Srinivas Boppu, Ahmed Hemani, and Linga Reddy Cenkeramaddi. "Methodology for Structured Data-Path Implementation in VLSI Physical Design: A Case Study." Electronics 11, no. 18 (September 19, 2022): 2965. http://dx.doi.org/10.3390/electronics11182965.
Full textAbstract:
State-of-the-art modern microprocessor and domain-specific accelerator designs are dominated by data-paths composed of regular structures, also known as bit-slices. Random logic placement and routing techniques may not result in an optimal layout for these data-path-dominated designs. As a result, implementation tools such as Cadence’s Innovus include a Structured Data-Path (SDP) feature that allows data-path placement to be completely customized by constraining the placement engine. A relative placement file is used to provide these constraints to the tool. However, the tool neither extracts nor automatically places the regular data-path structures. In other words, the relative placement file is not automatically generated. In this paper, we propose a semi-automated method for extracting bit-slices from the Innovus SDP flow. It has been demonstrated that the proposed method results in 17% less density or use for a pixel buffer design. At the same time, the other performance metrics are unchanged when compared to the traditional place and route flow.
17
Schmitt, Christian, Moritz Schmid, Sebastian Kuckuk, Harald Köstler, Jürgen Teich, and Frank Hannig. "Reconfigurable Hardware Generation of Multigrid Solvers with Conjugate Gradient Coarse-Grid Solution." Parallel Processing Letters 28, no. 04 (December 2018): 1850016. http://dx.doi.org/10.1142/s0129626418500160.
Full textAbstract:
Not only in the field of high-performance computing (HPC), field programmable gate arrays (FPGAs) are a soaringly popular accelerator technology. However, they use a completely different programming paradigm and tool set compared to central processing units (CPUs) or even graphics processing units (GPUs), adding extra development steps and requiring special knowledge, hindering widespread use in scientific computing. To bridge this programmability gap, domain-specific languages (DSLs) are a popular choice to generate low-level implementations from an abstract algorithm description. In this work, we demonstrate our approach for the generation of numerical solver implementations based on the multigrid method for FPGAs from the same code base that is also used to generate code for CPUs using a hybrid parallelization of MPI and OpenMP. Our approach yields in a hardware design that can compute up to 11 V-cycles per second with an input grid size of 4096[Formula: see text]4096 and solution on the coarsest using the conjugate gradient (CG) method on a mid-range FPGA, beating vectorized, multi-threaded execution on an Intel Xeon processor.
18
Liu, Xinyu, Chenhong Cao, and Shengyu Duan. "A Low-Power Hardware Architecture for Real-Time CNN Computing." Sensors 23, no. 4 (February 11, 2023): 2045. http://dx.doi.org/10.3390/s23042045.
Full textAbstract:
Convolutional neural network (CNN) is widely deployed on edge devices, performing tasks such as objective detection, image recognition and acoustic recognition. However, the limited resources and strict power constraints of edge devices pose a great challenge to applying the computationally intensive CNN models. In addition, for the edge applications with real-time requirements, such as real-time computing (RTC) systems, the computations need to be completed considering the required timing constraint, so it is more difficult to trade off between computational latency and power consumption. In this paper, we propose a low-power CNN accelerator for edge inference of RTC systems, where the computations are operated in a column-wise manner, to realize an immediate computation for the currently available input data. We observe that most computations of some CNN kernels in deep layers can be completed in multiple cycles, while not affecting the overall computational latency. Thus, we present a multi-cycle scheme to conduct the column-wise convolutional operations to reduce the hardware resource and power consumption. We present hardware architecture for the multi-cycle scheme as a domain-specific CNN architecture, which is then implemented in a 65 nm technology. We prove our proposed approach realizes up to 8.45%, 49.41% and 50.64% power reductions for LeNet, AlexNet and VGG16, respectively. The experimental results show that our approach tends to cause a larger power reduction for the CNN models with greater depth, larger kernels and more channels.
19
Parravicini, Daniele, Davide Conficconi, Emanuele Del Sozzo, Christian Pilato, and Marco D. Santambrogio. "CICERO: A Domain-Specific Architecture for Efficient Regular Expression Matching." ACM Transactions on Embedded Computing Systems 20, no. 5s (October 31, 2021): 1–24. http://dx.doi.org/10.1145/3476982.
Full textAbstract:
Regular Expression (RE) matching is a computational kernel used in several applications. Since RE complexity and data volumes are steadily increasing, hardware acceleration is gaining attention also for this problem. Existing approaches have limited flexibility as they require a different implementation for each RE. On the other hand, it is complex to map efficient RE representations like non-deterministic finite-state automata onto software-programmable engines or parallel architectures. In this work, we present CICERO , an end-to-end framework composed of a domain-specific architecture and a companion compilation framework for RE matching. Our solution is suitable for many applications, such as genomics/proteomics and natural language processing. CICERO aims at exploiting the intrinsic parallelism of non-deterministic representations of the REs. CICERO can trade-off accelerators’ efficiency and processors’ flexibility thanks to its programmable architecture and the compilation framework. We implemented CICERO prototypes on embedded FPGA achieving up to 28.6× and 20.8× more energy efficiency than embedded and mainstream processors, respectively. Since it is a programmable architecture, it can be implemented as a custom ASIC that is orders of magnitude more energy-efficient than mainstream processors.
20
Ruan, Aitong, Edison Chiu, and Shannon K. Oda. "Overcoming AML T cell therapy barriers with engineered proteins." Journal of Immunology 208, no. 1_Supplement (May 1, 2022): 122.27. http://dx.doi.org/10.4049/jimmunol.208.supp.122.27.
Full textAbstract:
Abstract Leukemia is the most common pediatric cancer and the second leading cause of cancer deaths in children. Adoptive cell therapy (ACT) has been validated in pediatric B cell acute lymphoblastic leukemia, where engineered cancer-targeting T cells can effectively eradicate tumor and tumor recurrence, generally more safely than drug and radiation therapy. However, ACT is not approved for patients with pediatric AML, other leukemias or solid tumors. Of note, pediatric AML patients have a worse 5-year survival rate than ALL. AML cells can severely diminish ACT efficacy by expressing ligands that activate T cell inhibitory receptors and by downregulating costimulatory ligand expression. To address these issues, we developed fusion proteins that combine the ectodomain of an inhibitory T cell receptor with an intracellular costimulatory signaling domain. We have shown that this effectively “replaces a brake with an accelerator” in CD8 T cells, and have now developed FPs with different costimulatory domains to confer different attributes to T cells. Despite improved outcomes with single FP T cell therapy, we hypothesized that combining differently costimulated T cells would provide a synergistically diverse T cell response and improve outcomes. We tested 1:1 pairwise combinations of differently costimulated T cells that are also transduced with a tumor-specific TCR in serial killing over-time (SKO) assays and in vivo persistence studies. In survival studies, dual T cell therapy resulted in enhanced survival. We propose that a synergistically diverse T cell therapy provides enhanced antitumor efficacy over a monogenic cell product.
21
Nakamura, Nobuhiro, Soh Yamazaki, Ken Sato, Akihiko Nakano, Masao Sakaguchi, and Katsuyoshi Mihara. "Identification of Potential Regulatory Elements for the Transport of Emp24p." Molecular Biology of the Cell 9, no. 12 (December 1998): 3493–503. http://dx.doi.org/10.1091/mbc.9.12.3493.
Full textAbstract:
To examine the possibility of active recycling of Emp24p between the endoplasmic reticulum (ER) and the Golgi, we sought to identify transport signal(s) in the carboxyl-terminal region of Emp24p. Reporter molecules were constructed by replacing parts of a control invertase-Wbp1p chimera with those of Emp24p, and their transport rates were assessed. The transport of the reporter was found to be accelerated by the presence of the cytoplasmic domain of Emp24p. Mutational analyses revealed that the two carboxyl-terminal residues, leucine and valine (LV), were necessary and sufficient to accelerate the transport. The acceleration was sequence specific, and the terminal valine appeared to be more important. The LV residues accelerated not only the overall transport to the vacuole but also the ER tocis-Golgi transport, suggesting its function in the ER export. Hence the LV residues are a novel anterograde transport signal. The double-phenylalanine residues did not affect the transport by itself but attenuated the effect of the anterograde transport signal. On the other hand, the transmembrane domain significantly slowed down the ER to cis-Golgi transport and effectively counteracted the anterograde transport signal at this step. It may also take part in the retrieval of the protein, because the overall transport to the vacuole was more evidently slowed down. Consistently, the mutation of a conserved glutamine residue in the transmembrane domain further slowed down the transport in a step after arriving at thecis-Golgi. Taken together, the existence of the anterograde transport signal and the elements that regulate its function support the active recycling of Emp24p.
22
Clark, N. T., Hongtao Zhong, and S. A. Mahlke. "Automated Custom Instruction Generation for Domain-Specific Processor Acceleration." IEEE Transactions on Computers 54, no. 10 (October 2005): 1258–70. http://dx.doi.org/10.1109/tc.2005.156.
Full text
23
Soldavini, Stephanie, and Christian Pilato. "A Survey on Domain-Specific Memory Architectures." Journal of Integrated Circuits and Systems 16, no. 2 (August 19, 2021): 1–9. http://dx.doi.org/10.29292/jics.v16i2.509.
Full textAbstract:
The never-ending demand for high performance and energy efficiency is pushing designers towards an increasing level of heterogeneity and specialization in modern computing systems. In such systems, creating efficient memory architectures is one of the major opportunities for optimizing modern workloads (e.g., computer vision, machine learning, graph analytics, etc.) that are extremely data-driven. However, designers demand proper design methods to tackle the increasing design complexity and address several new challenges, like the security and privacy of the data to be elaborated.This paper overviews the current trend for the design of domain-specific memory architectures. Domain-specific architectures are tailored for the given application domain, with the introduction of hardware accelerators and custom memory modules while maintaining a certain level of flexibility. We describe the major components, the common challenges, and the state-of-the-art design methodologies for building domain-specific memory architectures. We also discuss the most relevant research projects, providing a classification based on our main topics.
24
Frolov, Vladimir, Vadim Sanzharov, Vladimir Galaktionov, and Alexander Shcherbakov. "Development in Vulkan: a domain-specific approach." Proceedings of the Institute for System Programming of the RAS 33, no. 5 (2021): 181–204. http://dx.doi.org/10.15514/ispras-2021-33(5)-11.
Full textAbstract:
In this paper we propose a high-level approach to developing GPU applications based on the Vulkan API. The purpose of the work is to reduce the complexity of developing and debugging applications that implement complex algorithms on the GPU using Vulkan. The proposed approach uses the technology of code generation by translating a C++ program into an optimized implementation in Vulkan, which includes automatic shader generation, resource binding, and the use of synchronization mechanisms (Vulkan barriers). The proposed solution is not a general-purpose programming technology, but specializes in specific tasks. At the same time, it has extensibility, which allows to adapt the solution to new problems. For single input C++ program, we can generate several implementations for different cases (via translator options) or different hardware. For example, a call to virtual functions can be implemented either through a switch construct in a kernel, or through sorting threads and an indirect dispatching via different kernels, or through the so-called callable shaders in Vulkan. Instead of creating a universal programming technology for building various software systems, we offer an extensible technology that can be customized for a specific class of applications. Unlike, for example, Halide, we do not use a domain-specific language, and the necessary knowledge is extracted from ordinary C++ code. Therefore, we do not extend with any new language constructs or directives and the input source code is assumed to be normal C++ source code (albeit with some restrictions) that can be compiled by any C++ compiler. We use pattern matching to find specific patterns (or patterns) in C++ code and convert them to GPU efficient code using Vulkan. Pattern are expressed through classes, member functions, and the relationship between them. Thus, the proposed technology makes it possible to ensure a cross-platform solution by generating different implementations of the same algorithm for different GPUs. At the same time, due to this, it allows you to provide access to specific hardware functionality required in computer graphics applications. Patterns are divided into architectural and algorithmic. The architectural pattern defines the domain and behavior of the translator as a whole (for example, image processing, ray tracing, neural networks, computational fluid dynamics and etc.). Algorithmic pattern express knowledge of data flow and control and define a narrower class of algorithms that can be efficiently implemented in hardware. Algorithmic patterns can occur within architectural patterns. For example, parallel reduction, compaction (parallel append), sorting, prefix sum, histogram calculation, map-reduce, etc. The proposed generator works on the principle of code morphing. The essence of this approach is that, having a certain class in the program and transformation rules, one can automatically generate another class with the desired properties (for example, the implementation of the algorithm on the GPU). The generated class inherits from the input class and thus has access to all data and functions of the input class. Overriding virtual functions in generated class helps user to carefully connect generated code to the other Vulkan code written by hand. Shaders can be generated in two variants: OpenCL shaders for google “clspv” compiler and GLSL shaders for an arbitrary GLSL compiler. Clspv variant is better for code which intensively uses pointers and the GLSL generator is better if specific HW features are used (like hardware ray tracing acceleration). We have demonstrated our technology on several examples related to image processing and ray tracing on which we get 30-100 times acceleration over multithreaded CPU implementation.
25
Sever, Sanja, Hanna Damke, and Sandra L. Schmid. "Dynamin:Gtp Controls the Formation of Constricted Coated Pits, the Rate Limiting Step in Clathrin-Mediated Endocytosis." Journal of Cell Biology 150, no. 5 (September 4, 2000): 1137–48. http://dx.doi.org/10.1083/jcb.150.5.1137.
Full textAbstract:
The GTPase dynamin is essential for receptor-mediated endocytosis, but its function remains controversial. A domain of dynamin, termed the GTPase effector domain (GED), controls dynamin's high stimulated rates of GTP hydrolysis by functioning as an assembly-dependent GAP. Dyn(K694A) and dyn(R725A) carry point mutations within GED resulting in reduced assembly stimulated GTPase activity. Biotinylated transferrin is more rapidly sequestered from avidin in cells transiently overexpressing either of these two activating mutants (Sever, S., A.B. Muhlberg, and S.L. Schmid. 1999. Nature. 398:481–486), suggesting that early events in receptor-mediated endocytosis are accelerated. Using stage-specific assays and morphological analyses of stably transformed cells, we have identified which events in clathrin-coated vesicle formation are accelerated by the overexpression of dyn(K694A) and dyn(R725A). Both mutants accelerate the formation of constricted coated pits, which we identify as the rate limiting step in endocytosis. Surprisingly, overexpression of dyn(R725A), whose primary defect is in stimulated GTP hydrolysis, but not dyn(K694A), whose primary defect is in self-assembly, inhibited membrane fission leading to coated vesicle release. Together, our data support a model in which dynamin functions like a classical GTPase as a key regulator of clathrin-mediated endocytosis.
26
Smith, Jesse J., Erica K. Evans, Monica Murakami, Mary B. Moyer, M. Arthur Moseley, George Vande Woude, and Sally Kornbluth. "Wee1-Regulated Apoptosis Mediated by the Crk Adaptor Protein in Xenopus Egg Extracts." Journal of Cell Biology 151, no. 7 (December 25, 2000): 1391–400. http://dx.doi.org/10.1083/jcb.151.7.1391.
Full textAbstract:
Many of the biochemical reactions of apoptotic cell death, including mitochondrial cytochrome c release and caspase activation, can be reconstituted in cell-free extracts derived from Xenopus eggs. In addition, because caspase activation does not occur until the egg extract has been incubated for several hours on the bench, upstream signaling processes occurring before full apoptosis are rendered accessible to biochemical manipulation. We reported previously that the adaptor protein Crk is required for apoptotic signaling in egg extracts (Evans, E.K., W. Lu, S.L. Strum, B.J. Mayer, and S. Kornbluth. 1997. EMBO (Eur. Mol. Biol. Organ.) J. 16:230–241). Moreover, we demonstrated that removal of Crk Src homology (SH)2 or SH3 interactors from the extracts prevented apoptosis. We now report the finding that the relevant Crk SH2-interacting protein, important for apoptotic signaling in the extract, is the well-known cell cycle regulator, Wee1. We have demonstrated a specific interaction between tyrosine-phosphorylated Wee1 and the Crk SH2 domain and have shown that recombinant Wee1 can restore apoptosis to an extract depleted of SH2 interactors. Moreover, exogenous Wee1 accelerated apoptosis in egg extracts, and this acceleration was largely dependent on the presence of endogenous Crk protein. As other Cdk inhibitors, such as roscovitine and Myt1, did not act like Wee1 to accelerate apoptosis, we propose that Wee1–Crk complexes signal in a novel apoptotic pathway, which may be unrelated to Wee1's role as a cell cycle regulator.
27
Jaru-Ampornpan, Peera, Thang X. Nguyen, and Shu-ou Shan. "A Distinct Mechanism to Achieve Efficient Signal Recognition Particle (SRP)–SRP Receptor Interaction by the Chloroplast SRP Pathway." Molecular Biology of the Cell 20, no. 17 (September 2009): 3965–73. http://dx.doi.org/10.1091/mbc.e08-10-0989.
Full textAbstract:
Cotranslational protein targeting by the signal recognition particle (SRP) requires the SRP RNA, which accelerates the interaction between the SRP and SRP receptor 200-fold. This otherwise universally conserved SRP RNA is missing in the chloroplast SRP (cpSRP) pathway. Instead, the cpSRP and cpSRP receptor (cpFtsY) by themselves can interact 200-fold faster than their bacterial homologues. Here, cross-complementation analyses revealed the molecular origin underlying their efficient interaction. We found that cpFtsY is 5- to 10-fold more efficient than Escherichia coli FtsY at interacting with the GTPase domain of SRP from both chloroplast and bacteria, suggesting that cpFtsY is preorganized into a conformation more conducive to complex formation. Furthermore, the cargo-binding M-domain of cpSRP provides an additional 100-fold acceleration for the interaction between the chloroplast GTPases, functionally mimicking the effect of the SRP RNA in the cotranslational targeting pathway. The stimulatory effect of the SRP RNA or the M-domain of cpSRP is specific to the homologous SRP receptor in each pathway. These results strongly suggest that the M-domain of SRP actively communicates with the SRP and SR GTPases and that the cytosolic and chloroplast SRP pathways have evolved distinct molecular mechanisms (RNA vs. protein) to mediate this communication.
28
Gu, Yu, Robert Tinn, Hao Cheng, Michael Lucas, Naoto Usuyama, Xiaodong Liu, Tristan Naumann, Jianfeng Gao, and Hoifung Poon. "Domain-Specific Language Model Pretraining for Biomedical Natural Language Processing." ACM Transactions on Computing for Healthcare 3, no. 1 (January 31, 2022): 1–23. http://dx.doi.org/10.1145/3458754.
Full textAbstract:
Pretraining large neural language models, such as BERT, has led to impressive gains on many natural language processing (NLP) tasks. However, most pretraining efforts focus on general domain corpora, such as newswire and Web. A prevailing assumption is that even domain-specific pretraining can benefit by starting from general-domain language models. In this article, we challenge this assumption by showing that for domains with abundant unlabeled text, such as biomedicine, pretraining language models from scratch results in substantial gains over continual pretraining of general-domain language models. To facilitate this investigation, we compile a comprehensive biomedical NLP benchmark from publicly available datasets. Our experiments show that domain-specific pretraining serves as a solid foundation for a wide range of biomedical NLP tasks, leading to new state-of-the-art results across the board. Further, in conducting a thorough evaluation of modeling choices, both for pretraining and task-specific fine-tuning, we discover that some common practices are unnecessary with BERT models, such as using complex tagging schemes in named entity recognition. To help accelerate research in biomedical NLP, we have released our state-of-the-art pretrained and task-specific models for the community, and created a leaderboard featuring our BLURB benchmark (short for Biomedical Language Understanding & Reasoning Benchmark) at https://aka.ms/BLURB .
29
Campos, Fabiana V., Baron Chanda, Paulo S. L. Beirão, and Francisco Bezanilla. "α-Scorpion Toxin Impairs a Conformational Change that Leads to Fast Inactivation of Muscle Sodium Channels." Journal of General Physiology 132, no. 2 (July 28, 2008): 251–63. http://dx.doi.org/10.1085/jgp.200809995.
Full textAbstract:
α-Scorpion toxins bind in a voltage-dependent way to site 3 of the sodium channels, which is partially formed by the loop connecting S3 and S4 segments of domain IV, slowing down fast inactivation. We have used Ts3, an α-scorpion toxin from the Brazilian scorpion Tityus serrulatus, to analyze the effects of this family of toxins on the muscle sodium channels expressed in Xenopus oocytes. In the presence of Ts3 the total gating charge was reduced by 30% compared with control conditions. Ts3 accelerated the gating current kinetics, decreasing the contribution of the slow component to the ON gating current decay, indicating that S4-DIV was specifically inhibited by the toxin. In addition, Ts3 accelerated and decreased the fraction of charge in the slow component of the OFF gating current decay, which reflects an acceleration in the recovery from the fast inactivation. Site-specific fluorescence measurements indicate that Ts3 binding to the voltage-gated sodium channel eliminates one of the components of the fluorescent signal from S4-DIV. We also measured the fluorescent signals produced by the movement of the first three voltage sensors to test whether the bound Ts3 affects the movement of the other voltage sensors. While the fluorescence–voltage (F-V) relationship of domain II was only slightly affected and the F-V of domain III remained unaffected in the presence of Ts3, the toxin significantly shifted the F-V of domain I to more positive potentials, which agrees with previous studies showing a strong coupling between domains I and IV. These results are consistent with the proposed model, in which Ts3 specifically impairs the fraction of the movement of the S4-DIV that allows fast inactivation to occur at normal rates.
30
Chou, Chun-An, Xiaoning Jin, Amy Mueller, and Sarah Ostadabbas. "Multimodal Data Fusion-Moving From Domain-Specific Algorithms to Transdomain Understanding for Accelerated Solution Development." IEEE Sensors Letters 3, no. 1 (January 2019): 1–4. http://dx.doi.org/10.1109/lsens.2018.2886544.
Full text
31
Hoffman, Enrico Mingo, and Nikos G. Tsagarakis. "The Math of Tasks: A Domain Specific Language for Constraint-Based Task Specification." International Journal of Humanoid Robotics 18, no. 03 (June 2021): 2150008. http://dx.doi.org/10.1142/s0219843621500080.
Full textAbstract:
This paper proposes a Domain Specific Language to describe in a synthetic and comprehensive way, complex control problems for robotic systems. The proposed language, named Math of Tasks, abstracts from the mathematical description of the problem, which heavily depends on the particular algorithm chosen to solve it, relying on a set of operators and entities which instead, composed together, describes the behavior of the control problem. The Math of Tasks can describe any type of instantaneous controller regardless of the particular controlled variable (e.g., joint velocity, joint acceleration, contact forces). This paper shows the convenience of the proposed formalism using examples from classical control problems for different types of robotic platforms such as manipulators, humanoid bipeds, and quadrupeds.
32
Khasanov, Robert, Julian Robledo, Christian Menard, Andrés Goens, and Jeronimo Castrillon. "Domain-specific Hybrid Mapping for Energy-efficient Baseband Processing in Wireless Networks." ACM Transactions on Embedded Computing Systems 20, no. 5s (October 31, 2021): 1–26. http://dx.doi.org/10.1145/3476991.
Full textAbstract:
Advancing telecommunication standards continuously push for larger bandwidths, lower latencies, and faster data rates. The receiver baseband unit not only has to deal with a huge number of users expecting connectivity but also with a high workload heterogeneity. As a consequence of the required flexibility, baseband processing has seen a trend towards software implementations in cloud Radio Access Networks (cRANs). The flexibility gained from software implementation comes at the price of impoverished energy efficiency. This paper addresses the trade-off between flexibility and efficiency by proposing a domain-specific hybrid mapping algorithm. Hybrid mapping is an established approach from the model-based design of embedded systems that allows us to retain flexibility while targeting heterogeneous hardware. Depending on the current workload, the runtime system selects the most energy-efficient mapping configuration without violating timing constraints. We leverage the structure of baseband processing, and refine the scheduling methodology, to enable efficient mapping of 100s of tasks at the millisecond granularity, improving upon state-of-the-art hybrid approaches. We validate our approach on an Odroid XU4 and virtual platforms with application-specific accelerators on an open-source prototype. On different LTE workloads, our hybrid approach shows significant improvements both at design time and at runtime. At design-time, mappings of similar quality to those obtained by state-of-the-art methods are generated around four orders of magnitude faster. At runtime, multi-application schedules are computed 37.7% faster than the state-of-the-art without compromising on the quality.
33
Wang, Zhi, Xiao Hu, Jian Zhang, Zhao Lv, and Yang Guo. "AIM: Annealing in Memory for Vision Applications." Symmetry 12, no. 3 (March 20, 2020): 480. http://dx.doi.org/10.3390/sym12030480.
Full textAbstract:
As the Moore’s law era will draw to a close, some domain-specific architectures even non-Von Neumann systems have been presented to keep the progress. This paper proposes novel annealing in memory (AIM) architecture to implement Ising calculation, which is based on Ising model and expected to accelerate solving combinatorial optimization problem. The Ising model has a symmetrical structure and realizes phase transition by symmetry breaking. AIM draws annealing calculation into memory to reduce the cost of information transfer between calculation unit and the memory, improves the ability of parallel processing by enabling each Static Random-Access Memory (SRAM) array to perform calculations. An approximate probability flipping circuit is proposed to avoid the system getting trapped in local optimum. Bit-serial design incurs only an estimated 4.24% area above the SRAM and allows the accuracy to be easily adjusted. Two vision applications are mapped for acceleration and results show that it can speed up Multi-Object Tracking (MOT) by 780× and Multiple People Head Detection (MPHD) by 161× with only 0.0064% and 0.031% energy consumption respectively over approximate algorithms.
34
Popat, Rakesh, Sagar Lonial, Peter M. Voorhees, Simona Degli Esposti, Ira Gupta, Joanna Opalinska, Sandhya Sapra, et al. "DREAMM-2: Single-Agent Belantamab Mafodotin (Belamaf) Effects on Patient-Reported Outcome (PRO) Measures in Patients with Relapsed/Refractory Multiple Myeloma (RRMM)." Blood 136, Supplement 1 (November 5, 2020): 27–28. http://dx.doi.org/10.1182/blood-2020-140013.
Full textAbstract:
Introduction: Belamaf (GSK2857916), a B-cell maturation antigen-targeting antibody-drug conjugate (ADC), demonstrated deep and durable responses with a manageable safety profile as a single agent in patients with heavily pretreated RRMM in the pivotal DREAMM-2 study (NCT03525678; Lonial ASCO 2020 Poster 436). Health-related quality of life (HRQoL) was evaluated via the cancer-specific European Organisation for Research and Treatment of Cancer quality of life questionnaire (EORTC-QLQ-C30; Popat EHA 2020 Poster EP1746), a PRO used extensively in oncology/MM studies to evaluate symptoms, functioning, and QoL. The EORTC-QLQ-MY20 module was used to assess MM symptoms. Corneal events are expected during belamaf treatment, as with other monomethyl auristatin F-containing ADCs, so two ophthalmic vision-related PRO questionnaires (National Eye Institute Visual Function Questionnaire-25 item [NEI-VFQ-25] and Ocular Surface Disease Index [OSDI]) were used to characterize the impact of corneal events on patient symptoms and visual function. Meaningful within-patient changes in OSDI and NEI-VFQ-25 scores were estimated to better interpret outcomes in this population (Eliason ISPOR 2020). We report here the results of the PRO analyses in DREAMM-2 according to measures used in the trial. Methods: In DREAMM-2, patients who received single-agent belamaf (2.5 or 3.4 mg/kg, every 3 weeks [Q3W]) completed PRO questionnaires electronically at baseline and Q3W during treatment. Group-level, mean change from baseline over time was evaluated on EORTC domains. We also evaluated the percentage of patients with ≥10-point meaningful change threshold for improvement (Osoba J Clin Oncol 1998) on EORTC domains over time. Meaningful change thresholds in ocular PROs measuring treatment-related corneal events were estimated using recommended anchor and distribution-based methods, with 12.5-16.6 points estimated as meaningful in this population, depending on the domain (Eliason ISPOR EU 2020). We report results of the PRO data analysis for patients in the 2.5-mg/kg group selected for clinical development. Results: At Weeks 7 and 13, 46% (21/46) and 41% (12/29) of patients who completed PROs improved ≥10 points in the EORTC-QLQ-C30 Fatigue domain score, respectively; 30% (14/46) and 31% (9/29) improved their General Pain domain score. For EORTC-QLQ-C30, there were trends toward improvement in Fatigue at some time points on treatment; Global Health Status (GHS)/QoL, Role Functioning, and Physical Functioning domain scores remained relatively stable. The EORTC-QLQ-MY20 Disease Symptoms domain score (representing pain in different locations) showed a general trend toward improvement over time, with improvements of ≥10 points at Weeks 7 and 13 for 38% (17/45) and 29% (8/28) of participants. Ocular PRO data were available for 95% (92/97) of patients. Based on the OSDI vision-related functioning domain, a total of 49.5% of patients experienced a ≥12.5-point worsening from baseline (median time to worsening: 44 days). Meaningful improvement of these changes (based on defined 12.5-point thresholds) from worst severity post baseline was seen in 72% of patients (median time to improvement: 24 days). Importantly, even among patients with meaningful worsening in visual functioning, patient-reported QoL/GHS, Physical Functioning, and Role Functioning domains of the EORTC-QLQ-C30 remained stable while on treatment (Figure). Conclusions: Disease symptoms, functioning, and QoL did not worsen over time in these heavily pretreated patients receiving belamaf in DREAMM-2. Patients showed a general improvement in fatigue, which is often a difficult-to-manage symptom for patients with RRMM. Group-level, meaningful worsening in vision-related PRO domains was observed, which improved in the majority of patients. Despite ocular symptoms, even in patients with meaningful worsening, EORTC-QLQ-C30 data suggest that overall HRQoL and patient functioning remained stable while on treatment. These PRO results demonstrate a balance between overall QoL/functioning and vision-related impacts that, together with its clinical efficacy, supports the use of belamaf in the treatment of patients with RRMM. Funding: GSK (study 205678); drug linker technology licensed from Seattle Genetics; mAb produced using POTELLIGENT Technology licensed from BioWa. Figure 1 Disclosures Popat: GSK: Consultancy, Honoraria, Other: TRAVEL, ACCOMMODATIONS, EXPENSES (paid by any for-profit health care company); Celgene: Consultancy, Honoraria; Bristol Myers Squibb: Consultancy, Honoraria; Takeda: Consultancy, Honoraria, Other: Travel support, Research Funding; Janssen: Consultancy, Honoraria, Other: TRAVEL, ACCOMMODATIONS, EXPENSES (paid by any for-profit health care company); AbbVie: Consultancy, Honoraria. Lonial:Novartis: Consultancy, Honoraria, Other: Personal fees; Takeda: Consultancy, Other: Personal fees, Research Funding; Amgen: Consultancy, Honoraria, Other: Personal fees; Sanofi: Consultancy; Karyopharm: Consultancy; GSK: Consultancy, Honoraria, Other: Personal fees; Abbvie: Consultancy; Merck: Consultancy, Honoraria, Other: Personal fees; TG Therapeutics: Membership on an entity's Board of Directors or advisory committees; JUNO Therapeutics: Consultancy; Millennium: Consultancy, Honoraria; Onyx: Honoraria; Genentech: Consultancy; Janssen: Consultancy, Honoraria, Other: Personal fees, Research Funding; BMS: Consultancy, Honoraria, Other: Personal fees, Research Funding. Voorhees:Adaptive Biotechnologies: Other: Personal fees; Levine Cancer Institute, Atrium Health: Current Employment; TeneoBio: Other: Personal fees; Oncopeptides: Other: Personal fees; Novartis: Other: Personal fees; Janssen: Other: Personal fees; Celgene: Other: Personal fees; Bristol-Myers Squibb: Other: Personal fees. Degli Esposti:Moorfields Eye Hospital: Current Employment; GlaxoSmithKline: Consultancy, Honoraria. Gupta:GlaxoSmithKline: Current Employment, Current equity holder in publicly-traded company; Novartis: Current equity holder in publicly-traded company. Opalinska:GSK: Current Employment, Current equity holder in publicly-traded company. Sapra:GSK: Current Employment, Current equity holder in publicly-traded company. Gorsh:GSK: Current Employment, Current equity holder in publicly-traded company. He:GSK: Current Employment, Current equity holder in publicly-traded company. Kleinman:Triphase Accelerator U.S Corporation: Consultancy; ONL Therapeutics, Inc: Consultancy; Revolution Medicines, Inc: Consultancy; Editas Medicine, Inc: Consultancy; Cleave Therapeutics, Inc: Consultancy; Coherus Biosciences, Inc: Consultancy; Synergy Research Inc: Consultancy; GSK: Consultancy; Eyeon Therapeutics, LLC.: Current equity holder in private company; Zenith Epigenetics Ltd: Consultancy. Schaumberg:Evidera, Inc: Current Employment; GSK: Consultancy; Novaliq: Consultancy; SilkTech: Consultancy; University of Utah School of Medicine: Current Employment. Loubert:Modus Outcomes: Current Employment. Meunier:Modus Outcomes: Current Employment. Regnault:Modus Outcomes: Current Employment. Eliason:GSK: Current Employment, Current equity holder in publicly-traded company.
35
Caporale, Alfredo Leandro, Catalina M. Gonda, and Lucía Florencia Franchini. "Transcriptional Enhancers in the FOXP2 Locus Underwent Accelerated Evolution in the Human Lineage." Molecular Biology and Evolution 36, no. 11 (July 29, 2019): 2432–50. http://dx.doi.org/10.1093/molbev/msz173.
Full textAbstract:
Abstract Unique human features, such as complex language, are the result of molecular evolutionary changes that modified developmental programs of our brain. The human-specific evolution of the forkhead box P2 (FOXP2) gene-coding region has been linked to the emergence of speech and language in the human kind. However, little is known about how the expression of FOXP2 is regulated and whether its regulatory machinery evolved in a lineage-specific manner in humans. In order to identify FOXP2 regulatory regions containing human-specific changes, we used databases of human-accelerated noncoding sequences or HARs. We found that the topologically associating domain determined using developing human cerebral cortex containing the FOXP2 locus includes two clusters of 12 HARs, placing the locus occupied by FOXP2 among the top regions showing fast acceleration rates in noncoding regions in the human genome. Using in vivo enhancer assays in zebrafish, we found that at least five FOXP2-HARs behave as transcriptional enhancers throughout different developmental stages. In addition, we found that at least two FOXP2-HARs direct the expression of the reporter gene EGFP to foxP2-expressing regions and cells. Moreover, we uncovered two FOXP2-HARs showing reporter expression gain of function in the nervous system when compared with the chimpanzee ortholog sequences. Our results indicate that regulatory sequences in the FOXP2 locus underwent a human-specific evolutionary process suggesting that the transcriptional machinery controlling this gene could have also evolved differentially in the human lineage.
36
Vanderbauwhede, Wim, and Gavin Davidson. "Domain-specific acceleration and auto-parallelization of legacy scientific code in FORTRAN 77 using source-to-source compilation." Computers & Fluids 173 (September 2018): 1–5. http://dx.doi.org/10.1016/j.compfluid.2018.06.005.
Full text
37
Oh, Il-Hoan, and E. Premkumar Reddy. "The C-Terminal Domain of B-Myb Acts As a Positive Regulator of Transcription and Modulates Its Biological Functions." Molecular and Cellular Biology 18, no. 1 (January 1, 1998): 499–511. http://dx.doi.org/10.1128/mcb.18.1.499.
Full textAbstract:
ABSTRACT The myb gene family consists of three members, named A-, B-, and c-myb. All three members of this family encode nuclear proteins that bind DNA in a sequence-specific manner and function as regulators of transcription. In this report, we have examined the biochemical and biological activities of murine B-myb and compared these properties with those of murine c-myb. In transient transactivation assays, murine B-myb exhibited transactivation potential comparable to that of c-myb. An analysis of deletion mutants of B-myb and c-myb showed that while the C-terminal domain of c-Myb acts as a negative regulator of transcriptional transactivation, the C-terminal domain of B-Myb functions as a positive enhancer of transactivation. To compare the biological activities of c-myb and B-myb, the two genes were overexpressed in 32Dcl3 cells, which are known to undergo terminal differentiation into granulocytes in the presence of granulocyte colony-stimulating factor (G-CSF). We observed that c-myb blocked the G-CSF-induced terminal differentiation of 32Dcl3 cells, resulting in their continued proliferation in the presence of G-CSF. In contrast, ectopic overexpression of B-myb blocked the ability of 32D cells to proliferate in the presence of G-CSF and accelerated the G-CSF-induced granulocytic differentiation of these cells. Similar studies with B-myb–c-myb chimeras showed that only chimeras that contained the C-terminal domain of B-Myb were able to accelerate the G-CSF-induced terminal differentiation of 32Dcl3 cells. These studies show that c-myb and B-myb do not exhibit identical biological activities and that the carboxyl-terminal regulatory domain of B-Myb plays a critical role in its biological function.
38
Zhao, Xiaoli, Xiaomin Ma, John H. Dupius, Ruxi Qi, Jenny (Jingxin) Tian, Jiaxin Chen, Xiuyuan Ou, et al. "Negatively charged phospholipids accelerate the membrane fusion activity of the plant-specific insert domain of an aspartic protease." Journal of Biological Chemistry 298, no. 1 (January 2022): 101430. http://dx.doi.org/10.1016/j.jbc.2021.101430.
Full text
39
Vandebon, Jessica, Jose G. F. Coutinho, and Wayne Luk. "Scheduling Hardware-Accelerated Cloud Functions." Journal of Signal Processing Systems 93, no. 12 (October 27, 2021): 1419–31. http://dx.doi.org/10.1007/s11265-021-01695-7.
Full textAbstract:
AbstractThis paper presents a Function-as-a-Service (FaaS) approach for deploying managed cloud functions onto heterogeneous cloud infrastructures. Current FaaS systems, such as AWS Lambda, allow domain-specific functionality, such as AI, HPC and image processing, to be deployed in the cloud while abstracting users from infrastructure and platform concerns. Existing approaches, however, use a single type of resource configuration to execute all function requests. In this paper, we present a novel FaaS approach that allows cloud functions to be effectively executed across heterogeneous compute resources, including hardware accelerators such as GPUs and FPGAs. We implement heterogeneous scheduling to tailor resource selection to each request, taking into account performance and cost concerns. In this way, our approach makes use of different processor types and quantities (e.g. 2 CPU cores), uniquely suited to handle different types of workload, potentially providing improved performance at a reduced cost. We validate our approach in three application domains: machine learning, bio-informatics, and physics, and target a hardware platform with a combined computational capacity of 24 FPGAs and 12 CPU cores. Compared to traditional FaaS, our approach achieves a cost improvement for non-uniform traffic of up to 8.9 times, while maintaining performance objectives.
40
FAN, Yuyang, Zhi DENG, and Zihang LI. "Verification and reliability analysis of synchronizers in clock domain crossing." Xibei Gongye Daxue Xuebao/Journal of Northwestern Polytechnical University 40, no. 2 (April 2022): 369–76. http://dx.doi.org/10.1051/jnwpu/20224020369.
Full textAbstract:
There are a large number of multi-clock domain circuits in the airborne equipment of aircraft. When data is transmitted across the clock domain, meta-stability may occur, resulting in data transmission errors and reduced circuit reliability. However, due to the occasional and non-reproducible faults caused by metastability, and the high cost of existing cross-clock domain specific verification software, cross-clock domain circuit verification in three-mode redundancy scenarios is not supported. To solve this problem, a method that combines register transfer level (RTL) validation, board-level accelerated testing and computational evaluation based on traditional tools is presented. This method can detect the cross-clock domain transmission problems in three-mode application scenarios or normal scenarios and assess potential cross-clock domain transmission risks using generic simulation tools at an early stage of design. It reduces the cost of economy and time for high safety level airborne complex electronic verification, and improves the reliability of the circuit.
41
Imai, Norihiro, Michitaka Suzuki, Yoji Ishizu, Teiji Kuzuya, Takashi Honda, Kazuhiko Hayashi, Masatoshi Ishigami, et al. "Hepatocyte-specific depletion of ubiquitin regulatory X domain containing protein 8 accelerates fibrosis in a mouse non-alcoholic steatohepatitis model." Histochemistry and Cell Biology 148, no. 3 (April 18, 2017): 219–27. http://dx.doi.org/10.1007/s00418-017-1572-6.
Full text
42
Wang, Min, Li He, Bowei Chen, Yanwei Wang, Lishan Wang, Wei Zhou, Tianxu Zhang, et al. "Transgenerationally Transmitted DNA Demethylation of a Spontaneous Epialleles Using CRISPR/dCas9-TET1cd Targeted Epigenetic Editing in Arabidopsis." International Journal of Molecular Sciences 23, no. 18 (September 10, 2022): 10492. http://dx.doi.org/10.3390/ijms231810492.
Full textAbstract:
CRISPR/dCas9 is an important DNA modification tool in which a disarmed Cas9 protein with no nuclease activity is fused with a specific DNA modifying enzyme. A previous study reported that overexpression of the TET1 catalytic domain (TET1cd) reduces genome-wide methylation in Arabidopsis. A spontaneous naturally occurring methylation region (NMR19-4) was identified in the promoter region of the PPH (Pheophytin Pheophorbide Hydrolase) gene, which encodes an enzyme that can degrade chlorophyll and accelerate leaf senescence. The methylation status of NMR19-4 is associated with PPH expression and leaf senescence in Arabidopsis natural accessions. In this study, we show that the CRISPR/dCas9-TET1cd system can be used to target the methylation of hypermethylated NMR19-4 region to reduce the level of methylation, thereby increasing the expression of PPH and accelerating leaf senescence. Furthermore, hybridization between transgenic demethylated plants and hypermethylated ecotypes showed that the demethylation status of edited NMR19-4, along with the enhanced PPH expression and accelerated leaf senescence, showed Mendelian inheritance in F1 and F2 progeny, indicating that spontaneous epialleles are stably transmitted trans-generationally after demethylation editing. Our results provide a rational approach for future editing of spontaneously mutated epialleles and provide insights into the epigenetic mechanisms that control plant leaf senescence.
43
Sun, Zhe, Xupeng Sui, Yanzeng Deng, Li Zou, A. Korobkin, Lixin Xu, and Yichen Jiang. "Characteristics of Slamming Pressure and Force for Trimaran Hull." Journal of Marine Science and Engineering 9, no. 6 (May 24, 2021): 564. http://dx.doi.org/10.3390/jmse9060564.
Full textAbstract:
In this paper, the characteristics of the impact pressure and force of a trimaran section was studied by Computational Fluid Dynamics (CFD). The time domain features of the slamming pressure or force showed a strong correlation with the penetration depth regardless of the specific ways of water entry. The effects of velocity and acceleration on the impact pressure and force were analyzed. It was found that the initial impact of the main hull and the wet-deck slamming were predominantly affected by the entry velocities, whilst the acceleration had almost no effect for initial impact. The impact velocity presented a quadratic relation with slamming pressure/forces, and the relation between acceleration and wet-deck slamming pressure/force was linear. These were consistent with the patterns implied by analytical models such as the Wagner or MLM (Modified Logvinovich model) theories.
44
Xia, Haifeng, and Zhengming Ding. "Cross-Domain Collaborative Normalization via Structural Knowledge." Proceedings of the AAAI Conference on Artificial Intelligence 36, no. 3 (June 28, 2022): 2777–85. http://dx.doi.org/10.1609/aaai.v36i3.20181.
Full textAbstract:
Batch Normalization (BN) as an important component assists Deep Neural Networks in achieving promising performance for extensive learning tasks by scaling distribution of feature representations within mini-batches. However, the application of BN suffers from performance degradation under the scenario of Unsupervised Domain Adaptation (UDA), since the estimated statistics fail to concurrently describe two different domains. In this paper, we develop a novel normalization technique, named Collaborative Normalization (CoN), for eliminating domain discrepancy and accelerating the model training of neural networks for UDA. Unlike typical strategies only exploiting domain-specific statistics during normalization, our CoN excavates cross-domain knowledge and simultaneously scales features from various domains by mimicking the merits of collaborative representation. Our CoN can be easily plugged into popular neural network backbones for cross-domain learning. On the one hand, theoretical analysis guarantees that models with CoN promote discriminability of feature representations and accelerate convergence rate; on the other hand, empirical study verifies that replacing BN with CoN in popular network backbones effectively improves classification accuracy in most learning tasks across three cross-domain visual benchmarks.
45
SHI, Haobin, and Hangyu HU. "Acceleration method of infrared image detail enhancement on FPGA." Xibei Gongye Daxue Xuebao/Journal of Northwestern Polytechnical University 40, no. 3 (June 2022): 524–29. http://dx.doi.org/10.1051/jnwpu/20224030524.
Full textAbstract:
Among the many cockpit human-computer interaction systems of airborne, vehicle-mounted, and ship-borne platforms, the image detail enhancement technology can improve the ability of personnel to interpret infrared images, which has very important application requirements. The enhanced algorithm running on the embedded computing platform needs to have a higher processing speed and smaller time delay in meeting the needs of real-time interaction. The current infrared video sensor usually has a lower resolution, and the image detail enhancement algorithm still can achieve real-time processing performance on a homogeneous multi-core CPU processing platform. However, as the resolution of platform sensors continues to increase, image processing speed and time delay are difficult to meet application requirements. We propose a method for enhancing and accelerating infrared image details for FPGA platforms. By considering the FPGA's specific domain software and hardware computing architecture, the traditional bilateral filtering image details algorithm is accelerated using local cache and search table, so that real-time processing performance can still be achieved under the input image of the 4k resolution, while almost no additional memory access delay is added. While ensuring the processing effect, the algorithm greatly improves the processing and delay indicators to meet the application requirements of embedded equipment in specific fields such as the cockpit human-computer interaction system.
46
Carter, Rolinda R. L., Kimberley Talbot, Tyler W. Smith, Agnes Y. Y. Lee, and Edward L. G. Pryzdial. "Enhanced Fibrinolysis by Proteolysed Coagulation Factor Xa Is Inhibited by Rivaroxaban." Blood 120, no. 21 (November 16, 2012): 3352. http://dx.doi.org/10.1182/blood.v120.21.3352.3352.
Full textAbstract:
Abstract Abstract 3352 Introduction: Our lab has previously demonstrated that protease-modulated clotting factor Xa (FXa) acts as an accelerator of tissue plasminogen activator (tPA) to enhance clot lysis in vitro. This activity is facilitated by two specific plasmin cleavages of FXa in the protease domain. The first excises a 3 kDa C-terminal peptide to yield FXa-beta. The second cleavage is at Lys330, yielding non-covalently associated fragments of 33kDa and 13kDa, termed Xa33/13. When we incorporated a chloromethyl ketone (cmk) into the active site of FXa, Xa33/13 was not produced and tPA cofactor function was attenuated. Rivaroxaban (rxbn) is a new generation anticoagulant now approved for treatment of deep vein thrombosis (DVT) in Canada. Similar to the cmk but with higher specificity, rxbn binds directly to the active site of FXa. Here we addressed the possibility that rxbn may comparably affect the proteolytic modulation of FXa and its subsequent participation in fibrinolysis. Methods: Purified FXa with or without rxbn was incubated with plasmin and visualized by Coomassie blue protein staining. The FXa cleavage profile was also evaluated by western blot analysis in normal plasma, plasma from a patient taking rxbn for treatment of DVT, and normal plasma spiked with rxbn (5μM). Innovin was used to activate FX to FXa and initiate clot formation in plasma supplemented with therapeutic tPA (50 nM). The effect of rxbn on FXa-enhanced purified fibrin clot lysis in the presence of tPA (0.1 ρM) and plasminogen (0.6 μM) was monitored by Rayleigh scattering. In the absence of fibrin, chromogenic assays (S-2251) were used to evaluate the effect of rxbn on FXa during the tPA-dependent conversion of plasminogen to plasmin. The enhancement of plasmin generation was furthermore characterized according to the FXa proteolytic profile by western blot. Results: The FXa fragments produced by plasmin-mediated proteolysis were clearly altered in the presence of rxbn, such that generation of the fibrinolytic species, Xa33/13, was inhibited. As expected, western blot analysis of normal plasma samples showed that there was a rapid conversion of FXa-beta to Xa33/13. This differed in rxbn patient plasma in which the conversion to Xa33/13 was attenuated and FXa remained, as supported by the purified FXa cleavage experiment. Due to occupancy of the active site, formation of the FXa-antithrombin (XaAT) complex was reduced. This physiological inhibition complex is generated immediately after FX activation and has been recently shown by our lab to be involved in the generation of Xa33/13 in plasma. Confirming the results observed with patient plasma, normal plasma spiked with excess rxbn (∼5 μM), also showed reduced XaAT generation and subsequent inhibition of FXa conversion to Xa33/13. Both plasmas containing rxbn had a 28 kDa Xa-derived fragment. This fragment was previously reported by us to appear after prolonged treatment with plasmin and correlated to loss of tPA cofactor function. As predicted from the altered cleavage pattern caused by rxbn, turbidity assays showed the enhancement of fibrinolysis by FXa was profoundly inhibited in the presence of saturating rxbn. Chromogenic assay results suggest that the enhancement of tPA-dependent plasminogen activation by FXa was significantly delayed in the presence of rxbn causing this reduction of fibrinolytic FXa cofactor function. Western blot analysis of the chromogenic assay samples linked this delay to the inhibition of Xa33/13 generation. These observations suggested that rxbn interferes with the initial phase of plasmin generation and subsequent fibrinolysis, by altering the proteolytic modulation of FXa by plasmin. Conclusions: In our “auxiliary cofactor model” of fibrinolysis, FXa and its plasmin-derived fragment, Xa33/13, act as tPA accelerators in addition to fibrin to enhance plasmin generation. Here we have shown that the incorporation of rxbn into the active site of FXa prevents the conversion of FXa to Xa33/13 and consequently inhibits enhancement of tPA-dependent plasmin generation and clot lysis. Molecular modeling suggests that the rxbn binding site is very near Lys330, which may explain why plasmin-mediated proteolysis to Xa33/13 is altered. While these data do not have bearing on the favourable anticoagulant properties of rxbn, they may highlight an unforeseen inhibitory effect on fibrinolysis. Disclosures: Lee: Bayer Inc.: Honoraria.
47
Focosi, Daniele, Fabrizio Maggi, Massimo Franchini, Scott McConnell, and Arturo Casadevall. "Analysis of Immune Escape Variants from Antibody-Based Therapeutics against COVID-19: A Systematic Review." International Journal of Molecular Sciences 23, no. 1 (December 21, 2021): 29. http://dx.doi.org/10.3390/ijms23010029.
Full textAbstract:
The accelerated SARS-CoV-2 evolution under selective pressure by massive deployment of neutralizing antibody-based therapeutics is a concern with potentially severe implications for public health. We review here reports of documented immune escape after treatment with monoclonal antibodies and COVID-19-convalescent plasma (CCP). While the former is mainly associated with specific single amino acid mutations at residues within the receptor-binding domain (e.g., E484K/Q, Q493R, and S494P), a few cases of immune evasion after CCP were associated with recurrent deletions within the N-terminal domain of the spike protein (e.g., ΔHV69-70, ΔLGVY141-144 and ΔAL243-244). The continuous genomic monitoring of non-responders is needed to better understand immune escape frequencies and the fitness of emerging variants.
48
Pourjafar-Dehkordi, Danial, and Martin Zacharias. "Binding-induced functional-domain motions in the Argonaute characterized by adaptive advanced sampling." PLOS Computational Biology 17, no. 11 (November 29, 2021): e1009625. http://dx.doi.org/10.1371/journal.pcbi.1009625.
Full textAbstract:
Argonaute proteins in combination with short microRNA (miRNAs) can target mRNA molecules for translation inhibition or degradation and play a key role in many regulatory processes. The miRNAs act as guide RNAs that associate with Argonaute and the complementary mRNA target region. The complex formation results in activation of Argonaute and specific cleavage of the target mRNA. Both the binding and activation processes involve essential domain rearrangements of functional importance. For the Thermus Thermophilus Argonaute (TtAgo) system guide-bound (binary) and guide/target-bound (ternary) complexes are known but how the binding of guide and target mediate domain movements is still not understood. We have studied the Argonaute domain motion in apo and guide/target bound states using Molecular Dynamics simulations and a Hamiltonian replica exchange (H-REMD) method that employs a specific biasing potential to accelerate domain motions. The H-REMD technique indicates sampling of a much broader distribution of domain arrangements both in the apo as well as binary and ternary complexes compared to regular MD simulations. In the apo state domain arrangements corresponding to more compact (closed) states are mainly sampled which undergo an opening upon guide and guide/target binding. Whereas only limited overlap in domain geometry between apo and bound states was found, a larger similarity in the domain distribution is observed for the simulations of binary and ternary complexes. Comparative simulations on ternary complexes with 15 or 16 base pairs (bp) formed between guide and target strands (instead of 14) resulted in dissociation of the 3’-guide strand from the PAZ domain and domain rearrangement. This agrees with the experimental observation that guide-target pairing beyond 14 bps is required for activation and gives a mechanistic explanation for the experimentally observed activation process.
49
Wright, Jesse S., and Robert J. Kadner. "The Phosphoryl Transfer Domain of UhpB Interacts with the Response Regulator UhpA." Journal of Bacteriology 183, no. 10 (May 15, 2001): 3149–59. http://dx.doi.org/10.1128/jb.183.10.3149-3159.2001.
Full textAbstract:
ABSTRACT Bacterial two-component regulatory systems control the expression of target genes through regulated changes in protein phosphorylation. Signal reception alters the ability of a membrane-bound histidine kinase (HK) protein to transfer phosphate from ATP to a highly conserved histidine residue. The transfer of phosphate from the histidine to an aspartate residue on the cognate response regulator (RR) changes the ability of the latter protein to bind to target DNA sequences and to alter gene transcription. UhpB is the HK protein which controls production of the sugar phosphate transporter UhpT. Elevated expression of full-length UhpB or of a soluble hybrid protein, GST-Bc, which is glutathione S-transferase (GST) fused to the cytoplasmic C-terminal portion of UhpB, results in complete blockage ofuhpT expression in a uhp + strain. This dominant-negative interference could result from the ability of GST-Bc to bind and sequester the RR UhpA and to accelerate its dephosphorylation. The portion of GST-Bc responsible for the interference phenotype was localized using truncation, linker insertion, and point mutations to the region between residues 293 and 366 flanking His-313, the putative site of autophosphorylation. Point mutations which allow GST-Bc to activate uhpT expression or which relieve the interference phenotype were obtained at numerous sites throughout this region. This region of UhpB is related to the phosphoryl transfer domain of EnvZ, which forms half of an interdimer four-helix bundle and is responsible for dimerization of its cytoplasmic domain. The expression of GST fusion proteins carrying the corresponding portions of EnvZ strongly interfered with the activation of porin gene expression by OmpR. The GST-Bc protein accelerated dephosphorylation of P-UhpA. Reverse transfer of phosphate from P-UhpA to GST-Bc was observed in the presence of the metal chelator EDTA and depended on the presence of His-313. Phosphate transfer from P-UhpA to the liberated phosphoryl transfer domain also occurred. Taken together, these results indicate that the phosphoryl transfer-dimerization domain of UhpB participates in the specific binding of UhpA, in the control of autokinase activity, and in the dephosphorylation of P-UhpA.
50
Zhang, Jianzhi, David M. Webb, and Ondrej Podlaha. "Accelerated Protein Evolution and Origins of Human-Specific Features: FOXP2 as an Example." Genetics 162, no. 4 (December 1, 2002): 1825–35. http://dx.doi.org/10.1093/genetics/162.4.1825.
Full textAbstract:
Abstract Genes responsible for human-specific phenotypes may have been under altered selective pressures in human evolution and thus exhibit changes in substitution rate and pattern at the protein sequence level. Using comparative analysis of human, chimpanzee, and mouse protein sequences, we identified two genes (PRM2 and FOXP2) with significantly enhanced evolutionary rates in the hominid lineage. PRM2 is a histone-like protein essential to spermatogenesis and was previously reported to be a likely target of sexual selection in humans and chimpanzees. FOXP2 is a transcription factor involved in speech and language development. Human FOXP2 experienced a >60-fold increase in substitution rate and incorporated two fixed amino acid changes in a broadly defined transcription suppression domain. A survey of a diverse group of placental mammals reveals the uniqueness of the human FOXP2 sequence and a population genetic analysis indicates possible adaptive selection behind the accelerated evolution. Taken together, our results suggest an important role that FOXP2 may have played in the origin of human speech and demonstrate a strategy for identifying candidate genes underlying the emergences of human-specific features.