Third-hand smoke (THS) is a newly discovered environmental health hazard that results from accumulation and aging of second-hand smoke (SHS) toxins on surfaces where smoking has occurred. Our objective was to determine whether there is a time-dependent effect of THS exposure on health. Using an in vivo exposure mouse system that mimics exposure of humans to THS, we investigated its effects on biomarkers found in serum, and in liver and brain tissues. Mice were exposed to THS for 1, 2, 4, or 6 months and brain, liver, and serum were collected. We found that THS exposure, as early as 1 month, resulted in increased circulating inflammatory cytokines, tumor necrosis factor by an order of magnitude of 2 and granulocyte macrophage colony-stimulating factor by an order of magnitude of 1.5 and in increases in the stress hormone epinephrine and the liver damage biomarker aspartate aminotransferase (AST), increased in magnitude 1.5 and 2.5 times compared with controls, respectively. THS exposure for 2 months resulted in further damage and at 4 and 6 months, many factors related to oxidative stress were altered and caused molecular damage. We also found that the mice became hyperglycemic and hyperinsulinimic suggesting that insulin resistance (IR) may be a significant consequence of long-term exposure to THS. In conclusion, time-dependent THS exposure has a significant effect on health as early as 1 month after initiation of exposure and these alterations progressively worsen with time. Our studies are important because virtually nothing is known about the effects of increased THS exposure time, they can serve to educate the public on the dangers of THS, and the biomarkers we identified can be used in the clinic, once verified in exposed humans.

Third-hand smoke (THS) is a recently discovered environmental health hazard that results from accumulation and aging of second-hand smoke (SHS) toxins on surfaces of environments where smoking has occurred. Our objective was to determine whether there is a dose-dependent effect of THS exposure on biological markers of harm (BMH) using an in vivo exposure system that mimics exposure of humans to THS. THS exposure generated from as low as the 10 cigarettes-smoking regimen, resulted in increased circulating inflammatory cytokines, tumor necrosis factor alpha, interleukin 1 alpha, and granulocyte macrophage colony-stimulating factor. We also found that there was an increase in adrenocorticotropic hormone and superoxide dismutase and a decrease in ATP levels in liver tissue. Many of the altered BMH that are related to oxidative stress and decrease in ATP levels, suggest mitochondrial dysfunction. THS exposure generated from the 20 and 40 cigarettes-smoking regimen resulted in further damage. Our studies are important because virtually nothing is known about the physiological damage caused by different levels of THS exposure. These studies can also serve to educate the public on the dangers of THS and the BMH we identified can potentially be used in the clinic, once verified in exposed humans.

In this paper, we propose using Partitioned Global Address Space (PGAS) GPU one-sided asynchronous small messages to replace the widely used collective communication calls for sparse input multi-GPU embedding retrieval in deep learning recommendation systems. This GPU PGAS communication approach achieves (1) better communication and computation overlap, (2) smoother network usage, and (3) reduced overhead (due to the data unpack and rearrangement steps associated with collective communication calls). We implement a CUDA embedding retrieval backend for PyTorch that supports the proposed PGAS communication scheme and evaluate it on deep learning recommendation inference passes. Our backend outperforms the baseline using NCCL collective calls, achieving 1.97x speedup for the weak scaling test and 2.63x speedup for the strong scaling test in a 4 GPU NVLink-connected system.

We present Atos, a dynamic scheduling framework for multi-node-GPU systems that supports PGAS-style lightweight one-sided memory operations within and between nodes. Atos's lightweight GPU-to-GPU communication enables latency hiding and can smooth the interconnection usage for bisection-limited problems. These benefits are significant for dynamic, irregular applications that often involve fine-grained communication at unpredictable times and without predetermined patterns. Some principles for high performance: (1) do not involve the CPU in the communication control path; (2) allow GPU communication within kernels, addressing memory consistency directly rather than relying on synchronization with the CPU; (3) perform dynamic communication aggregation when interconnections have limited bandwidth. By lowering the overhead of communication and allowing it within GPU kernels, we support large, high-utilization GPU kernels but with more frequent communication. We evaluate Atos on two irregular problems: Breadth-First-Search and PageRank. Atos outperforms the state-of-the-art graph libraries Gunrock, Groute and Galois on both single-node-multi-GPU and multi-node-GPU settings.

We present Atos, a task-parallel GPU dynamic scheduling framework that is especially targeted at dynamic irregular applications. Compared to the dominant Bulk Synchronous Parallel (BSP) frameworks, Atos exposes additional concurrency by supporting task-parallel formulations of applications with relaxed dependencies, achieving higher GPU utilization, which is particularly significant for problems with concurrency bottlenecks. Atos also offers implicit task-parallel load balancing in addition to data-parallel load balancing, providing users the flexibility to balance between them to achieve optimal performance. Finally, Atos allows users to adapt to different use cases by controlling the kernel strategy and task-parallel granularity. We demonstrate that each of these controls is important in practice. We evaluate and analyze the performance of Atos vs. BSP on three applications: breadth-first search, PageRank, and graph coloring. Atos implementations achieve geomean speedups of 3.44x, 2.1x, and 2.77x and peak speedups of 12.8x, 3.2x, and 9.08x across three case studies, compared to a state-of-the-art BSP GPU implementation. Beyond simply quantifying the speedup, we extensively analyze the reasons behind each speedup. This deeper understanding allows us to derive general guidelines for how to select the optimal Atos configuration for different applications. Finally, our analysis provides insights for future dynamic scheduling framework designs.

Distributed data structures are key to implementing scalable applications for scientific simulations and data analysis. In this paper we look at two implementation styles for distributed data structures: remote direct memory access (RDMA) and remote procedure call (RPC). We focus on operations that require individual accesses to remote portions of a distributed data structure, e.g., accessing a hash table bucket or distributed queue, rather than global operations in which all processors collectively exchange information. We look at the trade-offs between the two styles through microbenchmarks and a performance model that approximates the cost of each. The RDMA operations have direct hardware support in the network and therefore lower latency and overhead, while the RPC operations are more expressive but higher cost and can suffer from lack of attentiveness from the remote side. We also run experiments to compare the real-world performance of RDMA- and RPC-based data structure operations with the predicted performance to evaluate the accuracy of our model, and show that while the model does not always precisely predict running time, it allows us to choose the best implementation in the examples shown. We believe this analysis will assist developers in designing data structures that will perform well on current network architectures, as well as network architects in providing better support for this class of distributed data structures.

The capacity of the colon to absorb microbially produced amino acids (AAs) and the underlying mechanisms of AA transport are incompletely defined. We measured the profile of 16 fecal AAs along the rat ceco-colonic axis and compared unidirectional absorptive AA fluxes across mucosal tissues isolated from the rat jejunum, cecum, and proximal colon using an Ussing chamber approach, in conjunction with ¹H-NMR and ultra-performance liquid chromatography-mass spectrometry chemical analyses. Passage of stool from cecum to midcolon was associated with segment-specific changes in fecal AA composition and a decrease in total AA content. Simultaneous measurement of up to 16 AA fluxes under native luminal conditions, with correction for endogenous AA release, demonstrated absorptive transfer of AAs across the cecum and proximal colon at rates comparable (30-80%) to those across the jejunum, with significant Na⁺-dependent and H⁺-stimulated components. Expression profiling of 30 major AA transporter genes by quantitative PCR revealed comparatively high levels of transcripts for 20 AA transporters in the cecum and/or colon, with the levels of 12 exceeding those in the small intestine. Our results suggest a more detailed model of major apical and basolateral AA transporters in rat colonocytes and provide evidence for a previously unappreciated transfer of AAs across the colonic epithelium that could link the prodigious metabolic capacities of the luminal microbiota, the colonocytes, and the body tissues.NEW & NOTEWORTHY This study provides evidence for a previously unappreciated transfer of microbially generated amino acids across the colonic epithelium under physiological conditions that could link the prodigious metabolic capacities of the luminal microbiota, the colonocytes, and the body tissues. The segment-specific expression of at least 20 amino acid transporter genes along the colon provides a detailed mechanistic basis for uniport, heteroexchange, Na⁺-cotransport, and H⁺-cotransport components of colonic amino acid absorption.