Host Memory Research Articles

Enhancing Linux System Security: A Kernel-Based Approach to Fileless Malware Detection and Mitigation

In the late 20th century, computer viruses emerged as powerful malware that resides permanently in target hosts. For a virus to function, it must load into memory from persistent storage, such as a file on a hard drive. Due to the significant destructive potential of viruses, numerous defense measures have been developed to protect computer systems. Among these, antivirus software is one of the most recognized and widely used. Typically, antivirus solutions rely on static analysis (signature-based) technologies to detect infections in files stored on permanent storage devices, such as hard drives or USB (Universal Serial Bus) flash drives. However, a new breed of malware, fileless malware, has been designed to evade detection and enhance durability. Fileless malware resides solely in the memory of the target hosts, circumventing traditional antivirus software, which cannot access or analyze processes executed directly from memory. This study proposes the Check-on-Execution (CoE) kernel-based approach to detect fileless malware on Linux systems. CoE intervenes by suspending code execution before a program executes code from a process’s writable and executable memory area. To prevent the execution of fileless malware, CoE extracts the code from memory, packages it with an ELF (Executable and Linkable Format) header to create an ELF file, and uses VirusTotal for analysis. Experimental results demonstrate that CoE significantly enhances a Linux system’s ability to defend against fileless malware. Additionally, CoE effectively protects against shell code injection attacks, including buffer and memory overflows, and can handle packed malware. However, it is important to note that this study focuses exclusively on fileless malware, and further research is needed to address other types of malware.

Simulation and reconstruction for 3D elastic wave using multi-GPU and CUDA-aware MPI

3D finite-difference time-domain numerical simulation and reconstruction based on the domain decomposition technique are essential parts of high-performance computation for reverse-time migration and full-waveform inversion. However, the low GPU utilization in computing for small-sized models and the tremendous memory consumption for large-sized models may result in low computational efficiency and high memory costs. This paper proposes a contiguous memory management (CMM) method and a variable-order wavefield reconstruction (VWR) method. The CMM allocates the memory of many small-sized arrays used for MPI communications on a larger-sized contiguous memory block, which aims to reduce the number of MPI communications between subdomains and improve the communication bandwidth, thus reducing the MPI time overhead and improving the GPU utilization. Meanwhile, the VWR can flexibly set the number of layers of boundary wavefield used for source wavefield reconstruction according to the host memory capacity and accuracy requirements. Since one layer of boundary wavefield could be stored using the VWR, the memory consumption of host memory can be significantly alleviated. Numerical experiments show that GPU utilization in computing for the model with a size of 1213 can be improved from 25% to 90% using the CMM method, and the VWR method can reduce memory consumption by about 86% while maintaining good accuracy in wavefield reconstruction. In addition, the issue of how to obtain a domain decomposition scheme with optimal performance is discussed in this paper.

ISP Agent: A Generalized In-storage-processing Workload Offloading Framework by Providing Multiple Optimization Opportunities

As solid-state drives (SSDs) with sufficient computing power have recently become the dominant devices in modern computer systems, in-storage processing (ISP), which processes data within the storage without transferring it to the host memory, is being utilized in various emerging applications. The main challenge of ISP is to deliver storage data to the offloaded workload. This is difficult because of the information gap between the host and storage, the data consistency problem between the host and offloaded workloads, and SSD-specific hardware limitations. Moreover, because the offloaded workloads use internal SSD resources, host I/O performance might be degraded due to resource conflicts. Although several ISP frameworks have been proposed, existing ISP approaches that do not deeply consider the internal SSD behavior are often insufficient to support efficient ISP workload offloading with high programmability. In this article, we propose an ISP agent, a lightweight ISP workload offloading framework for SSD devices. The ISP agent provides I/O and memory interfaces that allow users to run existing function codes on SSDs without major code modifications, and separates the resources for the offloaded workloads from the existing SSD firmware to minimize interference with host I/O processing. The ISP agent also provides further optimization opportunities for the offloaded workload by considering SSD architectures. We have implemented the ISP agent on the OpenSSD Cosmos+ board and evaluated its performance using synthetic benchmarks and a real-world ISP-assisted database checkpointing application. The experimental results demonstrate that the ISP agent enhances host application performance while increasing ISP programmability, and that the optimization opportunities provided by the ISP agent can significantly improve ISP-side performance without compromising host I/O processing.

Antibiotic-altered gut microbiota explain host memory plasticity and disrupt pace-of-life covariation for an aquatic snail.

There is mounting evidence that intestinal microbiota communities and their genes (the gut microbiome) influence how animals behave and interact with their environment, driving individual variation. Individual covariation in behavioural, physiological, and cognitive traits among individuals along a fast-slow continuum is thought to arise because these traits are linked as part of an adaptive pace-of-life strategy. Yet paradoxically, trait intercorrelation is absent or disrupted in some populations but not others. Here, we provide experimental evidence from aquatic pond snails (Lymnaea stagnalis) that environmental stressors and the gut microbiota explain host phenotypic plasticity and disrupted covariation among traits. Antibiotic exposure at varying levels of ecologically relevant concentrations had multiple effects starting with gut microbiota diversity, differential abundance, and inferred function. Memory declined in line with antibiotic concentrations that caused the most profound gut microbiota disruption, and although pace-of-life traits remained rigid, their covariation did not. Moreover, inferred microbial metabolic pathways with biologically relevant host functions explained individual and treatment variation in phenotypes. Together, our results point to the gut microbiome as a proximate mechanism influencing the emergence and maintenance of phenotypic variation within populations and highlights the need to decipher whether the gut microbiome's sensitivity to environmental pollution facilitates adaptive or maladaptive phenotypic plasticity.

XGNN: Boosting Multi-GPU GNN Training via Global GNN Memory Store

GPUs are commonly utilized to accelerate GNN training, particularly on a multi-GPU server with high-speed interconnects (e.g., NVLink and NVSwitch). However, the rapidly increasing scale of graphs poses a challenge to applying GNN to real-world applications, due to limited GPU memory. This paper presents XGNN, a multi-GPU GNN training system that fully utilizes system memory (e.g., GPU and host memory), as well as high-speed interconnects. The core design of XGNN is the Global GNN Memory Store (GGMS), which abstracts underlying resources to provide a unified memory store for GNN training. It partitions hybrid input data, including graph topological and feature data, across both GPU and host memory. GGMS also provides easy-to-use APIs for GNN applications to access data transparently, forwarding data access requests to the actual physical data partitions automatically. Evaluation on various multi-GPU platforms using three common GNN models with four large-scale datasets shows that XGNN outperforms DGL, Quiver and DGL+C by up to 7.9X (from 2.3X), 15.7X (from 3.3X) and 2.8X (from 1.3X), respectively.

Multi-resource predictive workload consolidation approach in virtualized environments

The revolution of virtualization technologies and Cloud computing solutions has emphasized the need for energy-efficient and Service level agreement (SLA)-aware resource management techniques in cloud data centers. Workload consolidation in Infrastructure-as-a-Service (IaaS) providers allows for efficient utilization of hardware resources and reduced energy consumption by consolidating workloads onto fewer physical servers. To ensure successful workload consolidation, it is crucial for IaaS providers to carefully estimate the host state and identify overloaded and underloaded hosts, thereby avoiding overly aggressive consolidation. Existing proposals determine the host state depending on its current resource utilization or a single anticipated resource utilization value, and often consider only a single resource type of the host, such as CPU. These limitations may lead to unreliable host state estimations, resulting in excessive and needless service migrations between physical machines (PMs). This, in turn, can lead to extra delays in service execution, degraded performance, increased power consumption, and SLA violations. To address these challenges, we propose a workload consolidation approach that leverages a multi-resource and multi-step resource utilization prediction model. Based on this model, our overload and underload decision-making algorithms consider the forecasted future trend (sequence of future value) of each host resource's utilization, including CPU, memory, and bandwidth. Through extensive experimentations conducted with two real-world datasets, we demonstrate that our approach can significantly reduce power consumption, SLA violation rate, and the number of migrations compared to existing benchmarks.

Theorizing Cultural Heritage Informatics as the Intersection of Heritage, Memory, and Information

ABSTRACTThis paper explores the relationship between cultural heritage and collective memory through the lens of information. We explicitly frame the heritage‐memory‐relationship as constituting a theoretical foundation for the field of Cultural Heritage Informatics. In framing heritage and memory as information phenomena, we leverage two information frameworks that appear most suitable to host, translate, and overlay heritage and memory theory: Context, relevance, and labor (Fremery and Buckland, 2022) and Information as thing (Buckland, 1991). Altogether, this preliminary exploration 1) establishes a terminological understanding of cultural heritage, collective memory, and information/informatics; 2) maps these concepts with recourse to the Context, relevance and labor framework (Fremery and Buckland, 2022); 3) articulates cultural heritage informatics as the intersection of heritage and memory focused on processes of selection, transfer, and integration of historic information in the service of identity maintenance; and 4) offers a cultural heritage information framework that highlights the unique potential of Cultural Heritage Informatics to guide future information research in cultural information studies.

An Alternate GPU-Accelerated Algorithm for Very Large Sparse LU Factorization

The LU factorization of very large sparse matrices requires a significant amount of computing resources, including memory and broadband communication. A hybrid MPI + OpenMP + CUDA algorithm named SuperLU3D can efficiently compute the LU factorization with GPU acceleration. However, this algorithm faces difficulties when dealing with very large sparse matrices with limited GPU resources. Factorizing very large matrices involves a vast amount of nonblocking communication between processes, often leading to a break in SuperLU3D calculation due to the overflow of cluster communication buffers. In this paper, we present an improved GPU-accelerated algorithm named SuperLU3D_Alternate for the LU factorization of very large sparse matrices with fewer GPU resources. The basic idea is “divide and conquer”, which means dividing a very large matrix into multiple submatrices, performing LU factorization on each submatrix, and then assembling the factorized results of all submatrices into two complete matrices L and U. In detail, according to the number of available GPUs, a very large matrix is first divided into multiple submatrices using the elimination tree. Then, the LU factorization of each submatrix is alternately computed with limited GPU resources, and its intermediate LU factors from GPUs are saved to the host memory or hard disk. Finally, after finishing the LU factorization of all submatrices, these factorized submatrices are assembled into a complete lower triangular matrix L and a complete upper triangular matrix U, respectively. The SuperLU3D_Alternate algorithm is suitable for hybrid CPU/GPU cluster systems, especially for a subset of nodes without GPUs. To accommodate different hardware resources in various clusters, we designed the algorithm to run in the following three cases: sufficient memory for GPU nodes, insufficient memory for GPU nodes, and insufficient memory for the entire cluster. The results from LU factorization test on different matrices in various cases show that the larger the matrix is, the more efficient this algorithm is under the same GPU memory consumption. In our numerical experiments, SuperLU3D_Alternate achieves speeds of up to 8× that of SuperLU3D (CPU only) and 2.5× that of SuperLU3D (CPU + GPU) on the hybrid cluster with six Tesla V100S GPUs. Furthermore, when the matrix is too big to be handled by SuperLU3D, SuperLU3D_Alternate can still utilize the cluster’s host memory or hard disk to solve it. By reducing the amount of data exchange to prevent exceeding the buffer’s limit of the cluster MPI nonblocking communication, our algorithm enhances the stability of the program.

Schistosoma mansoni infection alters the host pre-vaccination environment resulting in blunted Hepatitis B vaccination immune responses.

Schistosomiasis is a disease caused by parasitic flatworms of the Schistosoma spp., and is increasingly recognized to alter the immune system, and the potential to respond to vaccines. The impact of endemic infections on protective immunity is critical to inform vaccination strategies globally. We assessed the influence of Schistosoma mansoni worm burden on multiple host vaccine-related immune parameters in a Ugandan fishing cohort (n = 75) given three doses of a Hepatitis B (HepB) vaccine at baseline and multiple timepoints post-vaccination. We observed distinct differences in immune responses in instances of higher worm burden, compared to low worm burden or non-infected. Concentrations of pre-vaccination serum schistosome-specific circulating anodic antigen (CAA), linked to worm burden, showed a significant bimodal distribution associated with HepB titers, which was lower in individuals with higher CAA values at month 7 post-vaccination (M7). Comparative chemokine/cytokine responses revealed significant upregulation of CCL19, CXCL9 and CCL17 known to be involved in T cell activation and recruitment, in higher CAA individuals, and CCL17 correlated negatively with HepB titers at month 12 post-vaccination. We show that HepB-specific CD4+ T cell memory responses correlated positively with HepB titers at M7. We further established that those participants with high CAA had significantly lower frequencies of circulating T follicular helper (cTfh) subpopulations pre- and post-vaccination, but higher regulatory T cells (Tregs) post-vaccination, suggesting changes in the immune microenvironment in high CAA could favor Treg recruitment and activation. Additionally, we found that changes in the levels of innate-related cytokines/chemokines CXCL10, IL-1β, and CCL26, involved in driving T helper responses, were associated with increasing CAA concentration. This study provides further insight on pre-vaccination host responses to Schistosoma worm burden which will support our understanding of vaccine responses altered by pathogenic host immune mechanisms and memory function and explain abrogated vaccine responses in communities with endemic infections.

A review on computational storage devices and near memory computing for high performance applications

The von Neumann bottleneck is imposed due to the explosion of data transfers and emerging data-intensive applications in heterogeneous system architectures. The conventional computation approach of transferring data to CPU is no longer suitable especially with the cost it imposes. Given the increasing storage capacities, moving extensive data volumes between storage and computation cannot scale up. Hence, high-performance data processing mechanisms are needed, which may be achieved by bringing computation closer to data. Gathering insights where data is stored helps deal with energy efficiency, low latency, as well as security. Storage bus bandwidth is also saved when only computation results are delivered to the host memory. Various applications, including database acceleration, machine learning, Artificial Intelligence (AI), offloading (compression/encryption/encoding) and others can perform better and become more scalable if the “move process to data” paradigm is applied. Embedding processing engines inside Solid-State Drives (SSDs), transforming them to Computational Storage Devices (CSDs), provides the needed data processing solution. In this paper, we review the prior art on Near Data Processing (NDP) with focus on In-Storage Computing (ISC), identifying main challenges and potential gaps for future research directions.

Toward a pan-SARS-CoV-2 vaccine targeting conserved epitopes on spike and non-spike proteins for potent, broad and durable immune responses.

The SARS-CoV-2 non-Spike (S) structural protein targets on nucleocapsid (N), membrane (M) and envelope (E), critical in the host cell interferon response and memory T-cell immunity, are grossly overlooked in COVID vaccine development. The current Spike-only vaccines bear an intrinsic shortfall for promotion of a fuller T cell immunity. Vaccines designed to target conserved epitopes could elicit strong cellular immune responses that would synergize with B cell responses and lead to long-term vaccine success. We pursue a universal (pan-SARS-CoV-2) vaccine against Delta, Omicrons and ever-emergent new mutants. We explored booster immunogenicity of UB-612, a multitope-vaccine that contains S1-RBD-sFc protein and sequence-conserved promiscuous Th and CTL epitope peptides on the Sarbecovirus N, M and S2 proteins. To a subpopulation (N = 1,478) of infection-free participants (aged 18-85 years) involved in a two-dose Phase-2 trial, a UB-612 booster (third dose) was administered 6-8 months after the second dose. The immunogenicity was evaluated at 14 days post-booster with overall safety monitored until the end of study. The booster induced high viral-neutralizing antibodies against live Wuhan WT (VNT50, 1,711) and Delta (VNT50, 1,282); and against pseudovirus WT (pVNT50, 11,167) vs. Omicron BA.1/BA.2/BA.5 variants (pVNT50, 2,314/1,890/854), respectively. The lower primary neutralizing antibodies in the elderly were uplifted upon boosting to approximately the same high level in young adults. UB-612 also induced potent, durable Th1-oriented (IFN-γ+-) responses (peak/pre-boost/post-boost SFU/106 PBMCs, 374/261/444) along with robust presence of cytotoxic CD8+ T cells (peak/pre-boost/post-boost CD107a+-Granzyme B+, 3.6%/1.8%/1.8%). This UB-612 booster vaccination is safe and well tolerated without SAEs. By targeting conserved epitopes on viral S2, M and N proteins, UB-612 could provide potent, broad and long-lasting B-cell and T-cell memory immunity and offers the potential as a universal vaccine to fend off Omicrons and new VoCs without resorting to Omicron-specific immunogens. ClinicalTrials.gov ID: NCT04773067; ClinicalTrials.gov ID: NCT05293665; ClinicalTrials.gov ID: NCT05541861.

Increased hepatic interleukin-1, arachidonic acid, and reactive oxygen species mediate the protective potential of peptides shared by gut cysteine peptidases against Schistosoma mansoni infection in mice.

Multiple antigen peptide (MAP) construct of peptide with high homology to Schistosoma mansoni cathepsin B1, MAP-1, and to cathepsins of the L family, MAP-2, consistently induced significant (P < 0.05) reduction in challenge S. mansoni worm burden. It was, however, necessary to modify the vaccine formula to counteract the MAP impact on the parasite egg counts and vitality, and discover the mechanisms underlying the vaccine protective potential. Outbred mice were immunized with MAP-2 in combination with alum and/or MAP-1. Challenge infection was performed three weeks (wks) after the second injection. Blood and liver pieces were obtained on an individual mouse basis, 23 days post-infection (PI), a time of S. mansoni development and feeding in the liver before mating. Serum samples were examined for the levels of circulating antibodies and cytokines. Liver homogenates were used for assessment of liver cytokines, uric acid, arachidonic acid (ARA), and reactive oxygen species (ROS) content. Parasitological parameters were evaluated 7 wks PI. Immunization of outbred mice with MAP-2 in combination with alum and/or MAP-1 elicited highly significant (P < 0.005) reduction of around 60% in challenge S. mansoni worm burden and no increase in worm eggs' loads or vitality, compared to unimmunized or alum pre-treated control mice. Host memory responses to the immunogens are expected to be expressed in the liver stage when worm feeding and cysteine peptidases release start to be active. Serum antibody and cytokine levels were not significantly different between control and vaccinated mouse groups. Highly significant (P < 0.05 - <0.0001) increase in liver interleukin-1, ARA, and ROS content was recorded in MAP-immunized compared to control mice. The findings provided an explanation for the gut cysteine peptidases vaccine-mediated reduction in challenge worm burden and increase in egg counts.

CXL-Enabled Enhanced Memory Functions

The arrival of the Compute Express Link (CXL) protocol is a significant milestone for the systems community. CXL provides a standardized, cache-coherent memory protocol that can be used to attach devices and memory to a system, while maintaining memory coherency with the host processor. CXL enables accelerators (e.g., graphics processing units and data processing units) to both have direct load/store access to the host memory and the ability to make their own on-device memory likewise accessible to the host central processing unit. Because CXL allows technology interposition on the memory data plane, it opens up the possibility of “pushing down” functions into the memory subsystem. In this article, we introduce the concept of enhanced memory functions (EMFs). We then describe two use cases, one prototyped using a field-programmable gate array-based intelligent memory controller platform. Finally, we show initial experimental results indicating that EMFs could present valuable solutions to problems that are difficult to solve within existing computer architectures.

Early transcriptional and epigenetic divergence of CD8+ T cells responding to acute versus chronic infection.

During a microbial infection, responding CD8+ T cells give rise to effector cells that provide acute host defense and memory cells that provide sustained protection. An alternative outcome is exhaustion, a state of T cell dysfunction that occurs in the context of chronic infections and cancer. Although it is evident that exhausted CD8+ T (TEX) cells are phenotypically and molecularly distinct from effector and memory CD8+ T cells, the factors regulating the earliest events in the differentiation process of TEX cells remain incompletely understood. Here, we performed single-cell RNA-sequencing and single-cell ATAC-sequencing of CD8+ T cells responding to LCMV-Armstrong (LCMV-Arm) or LCMV-Clone 13 (LCMV-Cl13), which result in acute or chronic infections, respectively. Compared to CD8+ T cells that had undergone their first division in response to LCMV-Arm (Div1ARM) cells, CD8+ T cells that had undergone their first division in response to LCMV-Cl13 (Div1CL13) expressed higher levels of genes encoding transcription factors previously associated with exhaustion, along with higher levels of Ezh2, the catalytic component of the Polycomb Repressive Complex 2 (PRC2) complex, which mediates epigenetic silencing. Modulation of Ezh2 resulted in altered expression of exhaustion-associated molecules by CD8+ T cells responding to LCMV-Cl13, though the specific cellular and infectious contexts, rather than simply the level of Ezh2 expression, likely determine the eventual outcome. Taken together, these findings suggest that the differentiation paths of CD8+ T cells responding to acute versus chronic infections may diverge earlier than previously appreciated.

Accelerating turbulent reacting flow simulations on many-core/GPUs using matrix-based kinetics

The present work assesses the impact - in terms of time to solution, throughput analysis, and hardware scalability - of transferring computationally intensive tasks, found in compressible reacting flow solvers, to the GPU. Attention is focused on outlining the workflow and data transfer penalties associated with “plugging in” a recently developed GPU-based chemistry library into (a) a purely CPU-based solver and (b) a GPU-based solver, where, except for the chemistry, all other variables are computed on the GPU. This comparison allows quantification of host-to-device (and vice versa) data transfer penalties on the overall solver speedup as a function of mesh and reaction mechanism size. To this end, a recently developed GPU-based chemistry library known as UMChemGPU is employed to treat the kinetics in the flow solver KARFS. UMChemGPU replaces conventional CPU-based Cantera routines using a matrix-based formulation. The impact of i) data transfer times, ii) chemistry acceleration, and iii) the hardware architecture is studied in detail in the context of GPU saturation limits. Hydrogen and dimethyl ether (DME) reaction mechanisms are used to assess the impact of the number of species/reactions on overall/chemistry-only speedup. It was found that offloading the source term computation to UMChemGPU results in up to 7X reduction in overall time to solution and four orders of magnitude faster source term computation compared to conventional CPU-based methods. Furthermore, the metrics for achieving maximum performance gain using GPU chemistry with an MPI + CUDA solver are explained using the Roofline model. Integrating the UMChemGPU with an MPI + OpenMP solver does not improve the overall performance due to the associated data copy time between the device (GPU) and host (CPU) memory spaces. The performance portability was demonstrated using three different GPU architectures, and the findings are expected to translate to a wide variety of high-performance codes in the combustion community.

Wukong+G: Fast and Concurrent RDF Query Processing Using RDMA-Assisted GPU Graph Exploration

RDF graph has been increasingly used to store and represent information shared over the Web, including social graphs and knowledge bases. With the increasing scale of RDF graphs and the concurrency level of SPARQL queries, current RDF systems are confronted with inefficient concurrent query processing on massive data parallelism. The situation becomes more severe in the face of data-intensive queries (aka heavy query), which usually lead to suboptimal response time (latency) as well as throughput collapse. In this article, we present Wukong+G, the first graph-based distributed RDF query processing system that efficiently exploits the hybrid parallelism of CPU and GPU. Wukong+G is made fast and concurrent with four key designs. First, Wukong+G tames massive random memory accesses in graph exploration by efficiently mapping data between CPU and GPU for latency hiding, including a set of techniques like query-aware prefetching, pattern-aware pipelining and fine-grained swapping. Second, Wukong+G scales up by introducing a GPU-friendly RDF store to support RDF graphs exceeding GPU memory size, by using techniques like predicate-based grouping, pairwise caching and look-ahead replacing to narrow the gap between host and device memory scale. Third, Wukong+G scales out through a communication layer that decouples the transferring process for query metadata and intermediate results, and further leverages both native and GPUDirect RDMA to enable efficient communication on a CPU/GPU cluster. Finally, Wukong+G simultaneously runs multiple queries on a single GPU to improve overall throughput and fully exploits hardware heterogeneity (CPU/GPU) by scheduling a single query on CPU and GPU adaptively. We have implemented Wukong+G by extending a state-of-the-art distributed RDF store (i.e., Wukong) with distributed GPU support. Evaluation on a heterogeneous CPU/GPU cluster with RDMA-capable network shows that Wukong+G outperforms Wukong by up to 9.0× (from 2.3×) and scales well on 10 GPU cards for heavy queries. Wukong+G can also improve both latency and throughput by more than one order of magnitude when facing hybrid workloads.

G-NMP: Accelerating Graph Neural Networks with DIMM-based Near-Memory Processing

Graph Neural Networks (GNNs) are of great value in numerous applications and promote the development of cognitive intelligence, due to the capability of modeling non-euclidean data structures. However, the inherent irregularity makes GNNs memory-bound, and the hybrid computing paradigm of GNNs poses significant challenges for efficient deployment on existing hardware architectures. Near-Memory Processing (NMP) is a promising solution for alleviating the memory wall problem. In this paper, we present G-NMP, a practical and efficient DIMM-based NMP solution for accelerating GNNs, which accelerates both sparse Aggregation and dense Combination computations on DIMM for the first time. We propose a novel G-NMP hardware architecture to exploit rank-level memory parallelism efficiently, and the G-ISA instructions to reduce host memory requests significantly. We conduct several data flow optimizations on the G-NMP to improve memory-compute overlap and to realize efficient matrix computation. Then we develop an adaptive data allocation strategy for diverse vector sizes to further exploit feature-level parallelism. We also propose a novel memory request scheduling method to achieve flexible and low-overhead DRAM ownership transition between host and G-NMP. Overall, G-NMP achieves consistent performance advantages across diverse GNN models and datasets, and offers 1.46× overall performance and 1.29× energy efficiency on average compared with the state-of-the-art work.

Data Orchestration for Accelerating GPU-Based Light Field Rendering Aiming at a Wide Virtual Space

Recently, research on six-degree-of-freedom virtual reality (VR) systems based on a light field (LF) has been actively conducted. The LF-based approach is photorealistic and less prone to errors than the existing three-dimensional (3D) modeling-based approach. On the other hand, because the amount of data is very large for LF-based immersive virtual reality, it is important to manage the data efficiently. This paper makes two major contributions to data management for rendering acceleration in this context. First, the GPUs commonly used for high-speed rendering require optimization of the data transfer process between the host memory and the device memory. This paper proposes a simple but effective way to maximize data reuse by finding the proper size of a transfer unit depending on the target applications and the system capability constraints. Also, view synthesis and data transfer are performed in a pipelined manner to hide the data movement time. Second, for the LF rendering of a wide space, data management in the storage-host-GPU hierarchy is attempted for the first time. As the size of the virtual space increases, the required amount of LF data grows rapidly and exceeds the allocatable host memory capacity. Because the loading of data stored in storage is a very slow process, a system-level LF rendering structure that carefully considers the characteristics of the storage-host-GPU memory hierarchy should be designed. In this paper, progressive LF updating is proposed, and it prevents rendering stalls caused by slow storage data loading. Experimental results show that a 360-degree view with a <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$9000 \times 2048$ </tex-math></inline-formula> resolution can be rendered in an average of 2.57 ms from LF data with a <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$4096 \times 2048$ </tex-math></inline-formula> resolution, achieving performance close to 400 frames per second.

Sim[formula omitted]PIM: A complete simulation framework for Processing-in-Memory

With the help of modern memory integration technologies, Processing-in-Memory (PIM) has emerged as a practical approach to mitigate the memory wall while improving performance and energy efficiency in contemporary applications. Since these designs encompass accelerating and increasing the efficiency of critical specific and general-purposed applications, it is expected that these accelerators will be coupled to existing systems and consequently with systems capable of multi-thread computing. However, there is a lack of tools capable of quickly simulating different PIMs designs and their suitable integration with other hosts. This gap is even worse when considering simulations of multi-core systems. This work presents Sim2PIM, a Simple Simulator for PIM devices that seamlessly integrates any PIM architecture with the host processor and memory hierarchy. The framework simulation achieves execution speeds and accuracy on par with the perf tool on host code, less than 10% run-time overhead, and around 2% difference in metrics. Additionally, by exploring the thread parallelism in the application and utilizing the host hardware, Sim2PIM can achieve more than 8× simulation speedup compared to a sequential simulation and orders of magnitude compared to other simulators. Sim2PIM is available to download at https://pim.computer/.

Honeybee gut Lactobacillus modulates host learning and memory behaviors via regulating tryptophan metabolism

Honeybees are highly social insects with a rich behavioral repertoire and are a versatile model for neurobiological research. Their gut microbiota comprises a limited number of host-restricted bacterial phylotypes that are important for honeybee health. However, it remains unclear how specific gut members affect honeybee behaviors. Here, we find that antibiotic exposure disturbs the gut community and influences honeybee phenotypes under field conditions. Using laboratory-generated gnotobiotic bees, we show that a normal gut microbiota is required for olfactory learning and memory abilities. Brain transcriptomic profiling reveals distinct brain gene expression patterns between microbiota-free and conventional bees. Subsequent metabolomic analyses of both hemolymph and gut samples show that the microbiota mainly regulates tryptophan metabolism. Our results indicate that host-specific Lactobacillus strains promote memory behavior by transforming tryptophan to indole derivatives that activate the host aryl hydrocarbon receptor. Our findings highlight the contributions of specific gut members to honeybee neurological processes, thus providing a promising model to understand host-microbe interactions.