Direct Memory Research Articles

Enabling Distributed and Optimal RDMA Resource Sharing in Large-Scale Data Center Networks: Modeling, Analysis, and Implementation

Remote Direct Memory Access (RDMA) suffers from unfairness issues and performance degradation when multiple applications share RDMA network resources. Hence, an efficient resource scheduling mechanism is urged to optimally allocates RDMA resources among applications. However, traditional Network Utility Maximization (NUM) based solutions are inadequate for RDMA due to three challenges: 1) The standard NUM-oriented algorithm cannot deal with coupling variables introduced by multiple dependent RDMA operations; 2) The stringent constraint of RDMA on-board resources complicates the standard NUM by bringing extra optimization dimensions; 3) Naively applying traditional algorithms for NUM suffers from scalability issues in solving a large-scale RDMA resource scheduling problem. In this paper, we present how to optimally share the RDMA resources in large-scale data center networks with a distributed manner. First, we propose Distributed RDMA NUM (DRUM) to model the RDMA resource scheduling problem as a new variation of the NUM problem. Second, we present distributed algorithms to efficiently solve the large-scale, interdependent RDMA resource sharing problem for different RDMA use cases. Through theoretical analysis, the convergence and parallelism of proposed algorithms are guaranteed. Finally, we implement the algorithms as a kernel-level indirection module in the real-world RDMA environment, so as to provide end-to-end resource sharing and performance guarantee. Through extensive evaluations by large-scale simulations and testbed experiments, we show that our method significantly improves applications’ performance under resource contention, achieving <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$1.7-3.1\times$</tex-math> </inline-formula> higher throughput, and in a dynamic context, the largest performance improvement reaches <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$98.1\%$</tex-math> </inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$64.1\%$</tex-math> </inline-formula> in terms of latency and throughput, respectively.

Improving graphics processing unit performance based on neural network direct memory access controller

In this paper proposes the design and implementation of the back propagation algorithm based neural network DMA (Direct Memory Access) Controller for use of multimedia applications. The proposed DMA controller work with the back propagation-training algorithm. The advantages of the back propagation algorithm it will be work on the gradient loss w.r.t the network weights. So this back propagation algorithm is used as training algorithm for the DMA controller. The proposed method is test with the different workload characteristics like heavy workload, medium workload and normal workload. The performance parameters are considered here is like accuracy, precision, recall and F1 score etc. The proposed method is compared with existing methods like CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), LSTM (Long Sort term Memory) and GRU (Gated Recurrent Unit) etc. Finally, the proposed design will give the better performance than existing methods.

DSP-Based Traffic Target Detection for Intelligent Transportation

Internet of Things (IoT)-based intelligent transportation is attracting more and more attention. As a key component of intelligent transportation, traffic video monitoring is very important, in which vehicle and pedestrian detection on the road is a crucial task. Although vehicle and pedestrian detection through deep learning (DL) may achieve high accuracy, it tends to require high computing resources, which hinders its use on IoT devices. As an important class of IoT devices, digital signal processor (DSP) has the characteristics of low energy consumption, small size, and strong performance, which has been widely used in intelligent transportation. In order to use DL on DSP for accurate vehicle and pedestrian detection, we first propose a series of general tactics to optimize the object detection convolutional neural network (CNN) model, including convolution layer optimization, cache optimization, compiler optimization, intrinsics optimization and direct memory access (DMA) acceleration, and then a parallel scheme to extend the model to run on multicore, and further quantize the implementation of the model. We evaluate it on UA-DETRAC and KITTI datasets. Experimental results show that our method achieves a faster speed than running the same CNN model on a mainstream desktop CPU, with only 0.06% accuracy loss.

DMA controller design based on SHA-1 dual channel improvement algorithm

In order to make direct memory access (DMA) high-speed transmission while ensuring the security and integrity of data, the traditional Secure Hash Algorithm (SHA) is improved from the algorithm model and hardware architecture level by adopting the methods of data synchronization preprocessing and multi-block multi-output, and a dual-channel design with instant data comparison and data error correction functions adapted to high-speed DMA is achieved. The simulation results show that the maximum clock frequency of the DMA controller is 120 MHz and the throughput rate is up to 3.8 Gbps. The improved comparison channel can output the message summary at an interval of 0.56 μs, whereas the traditional algorithm needs to wait five times as long as the data handling time. The error correction channel also implements multi-block output and error block indexing. The results show that the improved dual-channel design based on the SHA algorithm can be highly adapted to DMA controllers and has good application prospects.

Traumatic narrative of the pemberontakan PKI Madiun 1948 in the Ayat-Ayat yang Disembelih by Anaf Afifi and Thowaf Zuharon: A postmemory study

This paper attempts to dismantle the pattern of trauma narratives developed in Indonesian literature by the second generation of the Pemberontakan PKI Madiun 1948. This second generation carries an Islamic narrative that is opposed to Communism. The narrative position of the Pemberontakan PKI Madiun 1948 is different from the G30S PKI 1965, in which the first PKI was the perpetrator, and the second PKI was the victim. The trauma narrative appears in the Ayat-Ayat yang Disembelih (2015) biography by Anaf Afifi and Thowaf Zuharon. The second generation lacks direct memory, which in Marianne Hirsch’s view, is postmemory. Accordingly, the theory used in this paper is Hirsch's postmemory, which believes that the second generation inherits trauma through investment in imagination, projection, and creation. The analytical method used is the methodological implications of postmemory theory with qualitative data. The pattern of the trauma narrative that emerges from the Islamic narrative to the PKI is; reflexive secondary first-person and mental transmission.

Accelerating micromagnetic and atomistic simulations using multiple GPUs

It is shown that micromagnetic and atomistic spin dynamics simulations can use multiple graphical processing units (GPUs) not only to reduce computation time but also to allow for a larger simulation size than is possible on a single GPU. While interactions that depend on neighboring spins, such as exchange interactions, may be implemented efficiently by transferring data between GPUs using halo regions or direct memory accesses, implementing the long-range demagnetizing interaction is the main difficulty in achieving good performance scaling, where the data transfer rate between GPUs is a significant bottleneck. A multi-GPU convolution algorithm is developed here, which relies on single-GPU FFTs executed in parallel. It is shown that even for micromagnetic simulations where the demagnetizing interaction computation time dominates, good performance scaling may be achieved, with speedup factors up to 1.8, 2.5, and 3.1, for two, three, and four GPUs, respectively. The code developed here can be used for any number of GPUs in parallel, with performance scaling strongly dependent on the inter-GPU data transfer rate and connection topology. This is further improved in micromagnetic simulations, which include a spin transport solver, obtaining speedup factors up to 1.96, 2.8, and 3.7, for two, three, and four GPUs, respectively. The best case scenario is obtained for atomistic simulations, where the demagnetizing interaction is implemented with spin-averaged cells. Using a single workstation with four GPUs, it is shown that atomistic spin dynamics simulations with up to 1 × 109 spins and atomistic Monte Carlo simulations with up to 2 × 109 spins are possible, with near-ideal performance scaling.

Development of SRAM-APB protocol interface and verification

The purpose of this mechanism is to enhance the chip’s internal connections and read/write memory capabilities. The Advanced Microcontroller Bus Architecture (AMBA) is one such shared bus that uses static random-access memory to achieve this goal. As a result, it’s important to weigh a variety of design options before diving into the Verilog description. It’s also important to remember that the system must be designed to accommodate a large number of interoperable modules and memories. The design, on the other hand, starts with fewer modules and a less complicated description and realisation that relies on memory access. ModelSim is used to simulate after the delay has been modelled in Verilog. Since the interface’s static random-access memory uses an APB protocol, its performance may be tested at this stage. In addition, Questasim employs verification modules and System Verilog technologies to guarantee the system’s operation. From the obtained results, the Direct Memory Access (DMA) with SRAM-APB outperforms the alternatives, particularly in frame transmission schemes, with a wire efficiency that is 1.4 times higher and a dynamic energy efficiency that is nearly twice as high as those of conventional configurations.

Converting Concurrent Range Index Structure to Range Index Structure for Disaggregated Memory

In this work, we propose the Spread approach, which tailors a concurrent range index structure to a range index structure for disaggregated memory connected via RDMA (Remote Direct Memory Access). The Spread approach leverages the concept of tolerating transient inconsistencies in a concurrent range index structure to reduce the amount of expensive RDMA operations. Based on the Spread approach, we converted Blink-tree, a concurrent range index structure, to a range index structure for disaggregated memory called RF-tree. In our experimental study, RF-tree shows comparable performance to Sherman, a state-of-the-art and carefully crafted range index structure for disaggregated memory.

A Research to Improve Contiguous Memory Allocation in Linux Kernel

The demand for Contiguous Memory Allocation (CMA) has witnessed significant growth in both low-end and high-end devices in recent years. However, the existing practices for utilizing CMA prove insufficient, particularly when catering to the needs of low-end (32-bit) devices. CMA, a Linux program used for memory reservation and allocation, faces limitations in its current implementations. Presently, techniques such as Scatter-Gather Direct Memory Access (DMA), Input Output Memory Management Unit (IOMMU), and Memory Reservation are commonly employed for contiguous memory allocation. Unfortunately, these methods are financially impractical for low-end devices and struggle to efficiently allocate substantial memory chunks, leading to latency concerns. In this paper, we introduce an improved CMA approach that intelligently allocates virtual memory for data mapping as needed. Alternatively, it directly allocates and deallocates physical memory without the necessity of virtual memory mapping, employing the DMA_KERNEL_NO_MAPPING attribute within the DMA Application Programming Interface (API). By adopting this method, latency is reduced, and the facilitation of larger memory allocations is promoted, addressing the limitations of the current techniques.

A High-performance RDMA-oriented Learned Key-value Store for Disaggregated Memory Systems

Disaggregated memory systems separate monolithic servers into different components, including compute and memory nodes, to enjoy the benefits of high resource utilization, flexible hardware scalability, and efficient data sharing. By exploiting the high-performance RDMA (Remote Direct Memory Access), the compute nodes directly access the remote memory pool without involving remote CPUs. Hence, the ordered key-value (KV) stores (e.g., B-trees and learned indexes) keep all data sorted to provide range query services via the high-performance network. However, existing ordered KVs fail to work well on the disaggregated memory systems, due to either consuming multiple network roundtrips to search the remote data or heavily relying on the memory nodes equipped with insufficient computing resources to process data modifications. In this article, we propose a scalable RDMA-oriented KV store with learned indexes, called ROLEX, to coalesce the ordered KV store in the disaggregated systems for efficient data storage and retrieval. ROLEX leverages a retraining-decoupled learned index scheme to dissociate the model retraining from data modification operations via adding a bias and some data movement constraints to learned models. Based on the operation decoupling, data modifications are directly executed in compute nodes via one-sided RDMA verbs with high scalability. The model retraining is hence removed from the critical path of data modification and asynchronously executed in memory nodes by using dedicated computing resources. ROLEX efficiently alleviates the fragmentation and garbage collection issues, due to allocating and reclaiming space via fixed-size leaves that are accessed via the atomic-size leaf numbers. Our experimental results on YCSB and real-world workloads demonstrate that ROLEX achieves competitive performance on the static workloads, as well as significantly improving the performance on dynamic workloads by up to 2.2× over state-of-the-art schemes on the disaggregated memory systems. We have released the open-source codes for public use in GitHub.

Narrating History and Memory: A Comparative Reading of Shujuan and Boku in Geling Yans The Flowers of War and Murakami Harukis Killing Commendatore

According to Pierre Noras argument on the relationship between memory and history, these two concepts are not synonyms but antonyms. In the modern world, critical history begins from rational reflection, and it represses memory, which correlates more to the personal narrative field. The memory theory gives us the insight to reexamine the narration in literary works reflecting the war memory. This paper demonstrates different voices in narrating the individualized traumatic memory of the Nanking Massacre through the comparative reading between fictional works of Murakami Haruki and Geling Yan, Killing Commendatore and The Flowers of War. Shujuan, the protagonist in the novella of Yan, is both the observer and survivor of the massacre. Her storyline obtains two heterogeneous narratives that respectively belong to the field of personal memory and history; while her responsibility of memorizing the unrecorded experience makes her to pick a historical view to narrate her past. Boku, the first-person protagonist in Murakamis fiction, on the other hand, is a contemporary character with no experience or direct memory of the war. However, he gets connected with the war memory through the Murakami-style surreal experience. His experience demonstrates how personal memory can console the trauma after the war, usually neglected by the grand history. In the paper, I argue that although the two authors are from different cultural backgrounds and apply different styles, their narrations that focus on the personal memory field show the consoling power that connects the past and present and heals the trauma left by the war. Further, their fictions still contribute to the political and historical discussions on Nanking, presenting the power against revisionists and bridging the conflicts inside the discussions around such a traumatic historical issue.

UrsaX: Integrating Block I/O and Message Transfer for Ultrafast Block Storage on Supercomputers

It is increasingly important for the next-generation exascale supercomputers to extend its applications beyond traditional HPC (high-performance computing) scenarios, so as to achieve high social and economic benefit. Similar to AWS (Amazon Web Services) and Alibaba Cloud, cloud-style virtual HPC service is a promising application scenario on supercomputers, for which remote block storage is the key to provide tenants with supercomputers extremely-high storage performance. Unfortunately, the state-of-the-art block storage software systems (such as URSA and Ceph) cannot adapt to the advanced hardware features of supercomputers. This paper presents UrsaX, an efficient block storage service for our next-generation Tianhe exascale supercomputer that is equipped with the high-performance GLEX network and NVMe (Non-Volatile Memory Express) SSDs. UrsaXs virtual disks, which can be mounted like normal physical ones, enable not only traditional HPC applications but also supercomputeroblivious POSIX applications to enjoy the high performance of supercomputers. At the core of UrsaX is with a novel design of the efficient integration of on-disk block I/O and in-network message transfer on supercomputers. UrsaX utilizes the NVMe Fabrics kernel module to expand the NVMe standard on the supercomputer network, and separates metadata I/O and data I/O of blocks respectively being handled over the MP (Mini Packet) and RDMA (Remote Direct Memory Access) protocols. We thoroughly explore the design space for remote block storage on supercomputers including parallelism, scalability, fault tolerance, and consistency. We conduct extensive evaluation on a subset of our exascale supercomputer consisting of 44 storage machines (each with four NVMe SSDs). The result shows that UrsaX achieves local-storage-level I/O latency (tens of microseconds) while being able to linearly increase the aggregate performance (IOPS and throughput) as the system scale increases, an order of magnitude higher than the state-of-the-art block storage systems.

FG-PFC: A Fine-Grained PFC Mechanism for Lossless RDMA

Remote Direct Memory Access (RDMA) is widely deployed in data centers to improve the performance, efficiency, and reliability of data centers. Priority-based Flow Control (PFC) can ensure no packet loss as a necessary approach to achieve lossless RDMA. However, since PFC is a coarse-grained flow control mechanism based on port and priority, it can cause the head-of-line blocking problem and affect other flows, resulting in high latency. In this paper, we propose a fine-grained PFC mechanism (FG-PFC), which refines port and priority into flow granularity and further determines and pauses the flow most likely to cause congestion. The benefit is reducing the persistent high occupancy of the buffer and protecting the innocent flow. Simulations and experiments demonstrate that FG-PFC outperforms PFC in terms of flow completion time (FCT).

Direct emotional interaction in prisoner's dilemma game

In this paper, we develop a prisoner's dilemma game model with three types of strategies and six types of emotions. We propose a direct emotional interaction mechanism considering memory effect in which a player's favorability serves as an index of emotions and directly impacts payoffs and emotions. Our model enables that individuals gradually increase the tendency to loneliness or defection in the face of unfriendly people, and increase the tendency to cooperation in the face of friendly people. By simulations, we further show that both direct emotional interaction and memory have the effect of enhancing the competitive advantage of cooperators and individuals with friendly emotions. Besides, the proportion of friendly groups increases if the degree of direct emotional interaction increases. Also, the impact of memory on cooperation has a non-monotonic characteristic. As a result, direct emotional interaction with memory effect leads to the improvement of social average favorability, increases the number of friendly populations, improves the overall level of social cooperation, and enables friendly people to gain advantages in evolution. Therefore, our work contributes to understanding the role of emotional factors in the evolution of social cooperation.

Nonequilibrium diffusion of active particles bound to a semiflexible polymer network: Simulations and fractional Langevin equation.

In a viscoelastic environment, the diffusion of a particle becomes non-Markovian due to the memory effect. An open question concerns quantitatively explaining how self-propulsion particles with directional memory diffuse in such a medium. Based on simulations and analytic theory, we address this issue with active viscoelastic systems where an active particle is connected with multiple semiflexible filaments. Our Langevin dynamics simulations show that the active cross-linker displays superdiffusive and subdiffusive athermal motion with a time-dependent anomalous exponent α. In such viscoelastic feedback, the active particle always exhibits superdiffusion with α = 3/2 at times shorter than the self-propulsion time (τA). At times greater than τA, the subdiffusive motion emerges with α bounded between 1/2 and 3/4. Remarkably, active subdiffusion is reinforced as the active propulsion (Pe) is more vigorous. In the high Pe limit, athermal fluctuation in the stiff filament eventually leads to α = 1/2, which can be misinterpreted with the thermal Rouse motion in a flexible chain. We demonstrate that the motion of active particles cross-linking a network of semiflexible filaments can be governed by a fractional Langevin equation combined with fractional Gaussian noise and an Ornstein-Uhlenbeck noise. We analytically derive the velocity autocorrelation function and mean-squared displacement of the model, explaining their scaling relations as well as the prefactors. We find that there exist the threshold Pe (Pe∗) and crossover times (τ∗ and τ†) above which active viscoelastic dynamics emerge on timescales of τ∗≲ t ≲ τ†. Our study may provide theoretical insight into various nonequilibrium active dynamics in intracellular viscoelastic environments.

Thermal mechanical and dielectric properties investigation of directional shape memory MCNTs/EP composites

The anisotropy of two-dimensional materials seriously hinders the improvement of the thermal mechanical and dielectric property of modified composites. Herein, the directional shape memory multi-walled carbon nanotubes/epoxy resin (MWCNTs/EP) composites were prepared by chemical thermal curing, and the directional alignment of magnetic CNTs (MCNTs) was conducted by employing a magnetic field of 0.33 T. The effect of directional aligned MCNTs on shape memory epoxy resins (SME) was investigated. The dispersion and orientation effects on heat transfer and shape memory performance of the directional shape memory MCNTs/EP composites were expounded. The results showed that when the mass fraction of MCNTs was 1.5 wt%, the dispersion was 77.5% and the orientation rate was 80%, which the thermal mechanical properties were the significantly improved. The temperature rise of the MCNTs/EP composites was 38.7% higher, the shape recovery rate was 1.01% higher and the shape recovery rate is increased to 360% of the raw material. Meanwhile, the dielectric property of the directional aligned MCNTs/EP composites was enhanced, which proving that directional aligned CNTs can be used to promote shape memory and electromagnetic performances.

Data Center Ethernet and Remote Direct Memory Access: Issues at Hyperscale

Data Center Ethernet and Remote Direct Memory Access: Issues at Hyperscale

Anisotropic mass transport enables distinct synaptic behaviors on 2D material surface

Anisotropy is an intrinsic property in crystals with low structural symmetry, and such well-textured materials usually show distinct electronic transport and optical properties along different lattice orientations, offering wide applications in electronic and photonic devices. As a typical low-symmetry materials, crystalline GeSe with orthorhombic structure shows large electric and optical anisotropies. In this work, we take advantage of the anisotropic mass transport and filamentary growth of Ag ions on the GeSe surface to fabricate planar memristive devices which show directional memory and transient switching phenomena. The anisotropic switching behaviors stem from the distinct morphology of metallic filaments that are directionally dependent on the mobility of ions, e.g., ions diffusing along the low-barrier direction tend to form stark conductive channels while those with low mobility only entail slim and weak dendrites, which have been clearly observed under electronic microscopy. The functionality could be utilized to mimic various synaptic events, such as long-term memory enabled by stable conductive channels and short-term memory by the spontaneous rupture of weak filaments, all implemented in one physical device. Two integration schemes based on the anisotropic devices are designed and demonstrated for different application scenarios, paving the way for its applications in multifunctional brain-inspired computing systems.

Narrative Processes, the Power of the Image,and Postmemory in a Graphic Novel: The Example of Petit-fils d’Algérie1by Joël Alessandras Petit-fils d'Algérie (2015).

How do narrative processes serve the post-memorial project of the actorialized narrator? This is what we propose to examine here, based on a study of Joël Alessandra's Petit-fils d'Algérie, in which the author, who set out in the footsteps of his ancestors, recounts his trip to Algeria in 2014. This graphic novel has a complex narrative structure, with several narrators, temporal shifts and a great importance attached to the story’s setting. Drawing on the definition of post-memory as an indirect memory of the facts (Howell, 2015) and on the narratology of comics (Groensteen 1999, 2011, 2021), this article examines the close links between narrative processes, their visualisation and the construction of a personal memory which thereby becomes transmissible, before leading to a rapprochement with the notion of the "post-memorial subject" (Fevry, 2018).

Accelerated FPGA-Based Vector Directional Filter for Real-Time Color Image Denoising with Enhanced Performance

This paper presents an accelerated implementation of the Vector Directional Filter (VDF) on a Field Programmable Gate Array (FPGA) for real-time denoising of color images. The VDF effectively suppresses noise while preserving edges and fine details, making it ideal for a range of applications such as satellite and multispectral biomedical imaging. However, the filter's high computational complexity poses challenges for real-time processing. Existing solutions either fail to meet real-time execution requirements or compromise image quality through hardware implementation approximations. To overcome these challenges, we first model the VDF using C/C++ programming, and subsequently design an efficient floating-point hardware architecture employing the High-Level Synthesis (HLS) flow. Optimal directives are selected using the Xilinx Vivado HLS tool. The VDF architecture is then integrated as a coprocessor with the Cortex-A53 hardcore processor in the XCZU9EG FPGA. To enhance data bandwidth between software and hardware components, three Direct Memory Access (DMAs) units are utilized to transfer three image lines in parallel. Furthermore, internal memory is implemented on the XCZU9EG FPGA, providing increased flexibility for managing the restored image. The VDF Software/Hardware (SW/HW) design's robustness and accuracy are validated through experimental studies on the ZCU102 board. Our design accelerates the filtering process by 21 times, maintaining visual quality and effectively removing noise from color images compared to the VDF SW design. Additionally, our solution outperforms existing approaches in terms of filtered image quality and processing time, showing a 24% improvement in the worst-case scenario.