Efficient shared memory and RDMA based collectives on multi-rail QsNetII SMP clusters

Clusters of Symmetric Multiprocessors (SMP) are more commonplace than ever in achieving high-performance. Scientific applications running on clusters employ collective communications extensively. Shared memory communication and Remote Direct Memory Access (RDMA) over multi-rail networks are promising approaches in addressing the increasing demand on intra-node and inter-node communications, and thereby in boosting the performance of collectives in emerging multi-core SMP clusters. In this regard, this paper designs and evaluates two classes of collective communication algorithms directly at the Elan user-level over multi-rail Quadrics QsNet^II with message striping: 1) RDMA-based traditional multi-port algorithms for gather, all-gather, and all-to-all collectives for medium to large messages, and 2) RDMA-based and SMP-aware multi-port all-gather algorithms for small to medium size messages. The multi-port RDMA-based Direct algorithm for gather and all-to-all collectives gain an improvement of up to 2.15 for 4 KB messages over elan_gather(), and up to 2.26 for 2 KB messages over elan_alltoall(), respectively. For the all-gather, our SMP-aware Bruck algorithm outperforms all other all-gather algorithms including elan_gather() for 512 B to 8 KB messages, with a 1.96 improvement factor for 4 KB messages. Our multi-port Direct all-gather is the best algorithm for 16 KB to 1 MB, and outperforms elan_gather() by a factor of 1.49 for 32 KB messages. Experimentation with real applications has shown up to 1.47 communication speedup can be achieved using the proposed all-gather algorithms.

DF++ : an adaptive buffer-aware probabilistic delegation forwarding protocol for Delay Tolerant Network

The delegated forwarding (DF) curb transmissions by forwarding the message to a node that holds high quality value seen by the message. However, DF assumes infinite buffer space that is not possible in real time applications. In addition, quality value computation considers the encountering history and does not account for additional network parameters such as aging and transitive connectivity. In this paper, we have proposed a routing protocol called as DF++ that compute quality value based on probabilistic model used in PRoPHET protocol and forwards the message to current node by adaptive computation of available buffer space. We have compared performance of DF++ with DF, Epidemic and PRoPHET routing protocols. The proposed DF++ has higher delivery probability and fewer message drop and transmissions.     

Buffer effects on EcoRV kinetics as measured by fluorescent staining and digital imaging of plasmid cleavage

We have developed a protocol to quantify polymer DNA cleavage which replaces the traditional radiolabeling and scintillation counting with fluorescent staining and digital imaging. This procedure offers high sensitivity, speed, and convenience, while avoiding waste and error associated with traditional 32P radiolabeling. This protocol was used to measure cleavage of pBR322 plasmid DNA by EcoRV, a type II restriction enzyme. EcoRV was found to exhibit an order of magnitude difference in binding in two apparently similar buffers used in previous investigations. To determine the origin of this effect, we measured reaction kinetics in buffers of different chemical nature and concentration: Tris, bis-Tris propane, Tes, Hepes, and cacodylate. We found that buffer concentration and identity had significant effects on EcoRV reaction velocity through large changes in specific binding and nonspecific binding (reflected in the Michaelis constant Km and the dissociation constant for nonspecific binding Kns). There were only small changes in Vmax. The source of the buffer effect is the protonated amines common to many pH buffers. These buffer cations likely act as counterions screening DNA phosphates, where both the protonated buffer structure and concentration affect enzyme binding strength. It appears that by choosing anionic buffers or zwitterionic buffers with a buried positive charge, buffer influence on the protein binding to DNA can be largely eliminated.     

One-Sided Communication on Clusters with Myrinet

This paper describes an efficient implementation of one-sided communication on top of the GM low-level message-passing library for clusters with Myrinet. This approach is compatible with shared memory, exploits pipelining, nonblocking communication, and overlapping memory registration with memory copy to maximize the transfer rate. The paper addresses critical design issues faced on the commodity clusters and then describes possible solutions for matching the low-level network protocol with user-level interfaces. The performance implications of the design decisions are presented and discussed in context of a standalone communication benchmark as well as two applications. Finally, the paper offers some indications on what additional features would be desirable in a communication library like GM to better support one-sided communication.     

Stable adducts of nerve agents sarin,soman and cyclosarin with TRIS,TES and related buffer compounds—Characterization by LC-ESI-MS/MS and NMR and implications for analytical chemistry

Buffering compounds like TRIS are frequently used in chemical, biochemical and biomedical applications to control pH in solution. One of the prerequisites of a buffer compound, in addition to sufficient buffering capacity and pH stability over time, is its non-reactivity with other constituents of the solution. This is especially important in the field of analytical chemistry where analytes are to be determined quantitatively. Investigating the enzymatic hydrolysis of G-type nerve agents sarin, soman and cyclosarin in buffered solution we have identified stable buffer adducts of TRIS, TES and other buffer compounds with the nerve agents. We identified the molecular structure of these adducts as phosphonic diesters using 1D ¹H-³¹P HSQC NMR and LC-ESI-MS/MS techniques. Reaction rates with TRIS and TES are fast enough to compete with spontaneous hydrolysis in aqueous solution and to yield substantial amounts (up to 20–40%) of buffer adduct over the course of several hours. A reaction mechanism is proposed in which the amino function of the buffer serves as an intramolecular proton acceptor rendering the buffer hydroxyl groups nucleophilic enough for attack on the phosphorus atom of the agents. Results show that similar buffer adducts are formed with a range of hydroxyl and amino function containing buffers including TES, BES, TRIS, BIS-TRIS, BIS-TRIS propane, Tricine, Bicine, HEPES and triethanol amine. It is recommended to use alternative buffers like MOPS, MES and CHES when working with G-type nerve agents especially at higher concentrations and over prolonged times.     

Improved MPI collectives for MPI processes in shared address spaces

As the number of cores per node keeps increasing, it becomes increasingly important for MPI to leverage shared memory for intranode communication. This paper investigates the design and optimization of MPI collectives for clusters of NUMA nodes. We develop performance models for collective communication using shared memory and we demonstrate several algorithms for various collectives. Experiments are conducted on both Xeon X5650 and Opteron 6100 InfiniBand clusters. The measurements agree with the model and indicate that different algorithms dominate for short vectors and long vectors. We compare our shared-memory allreduce with several MPI implementations—Open MPI, MPICH2, and MVAPICH2—that utilize system shared memory to facilitate interprocess communication. On a 16-node Xeon cluster and 8-node Opteron cluster, our implementation achieves on geometric average 2.3X and 2.1X speedup over the best MPI implementation, respectively. Our techniques enable an efficient implementation of collective operations on future multi- and manycore systems.     

The role of vegetation and litter in the nitrogen dynamics of riparian buffer zones in Europe

Plant uptake and denitrification are considered to be the most important processes responsible for N retention and mitigation in riparian buffers. In many riparian buffers, however, nutrients taken up by plants remain in the system only temporarily and may be gradually released by mineralization later. Still, plants increase the residence time of nutrients considerably by reducing their mobility. We investigated the importance of plant N uptake and N immobilization in litter for N retention in riparian buffers. Nitrogen uptake in vegetation and N dynamics in litter were measured over a two-year period in a range of forested and herbaceous riparian buffers along a climatic gradient in Europe, receiving different loadings of N-enriched groundwater. Plant production, nitrogen uptake, and N retention were significantly higher in the forested buffer sites compared to the herbaceous buffer sites. However, in herbaceous buffers, periodic harvesting of herbaceous biomass contributed considerably to the N retention. No relationship between lateral N loading and plant productivity or N uptake was observed; this indicated that plant growth was not N-limited. In the winter period, decaying leaf litter had a small but significant role in N retention in a majority of the riparian ecosystems studied. Moreover, no responses to the climatic gradient were found. Generally, we can state that annual N retention in the vegetation and litter compartment is substantial, making up 13–99% of the total N mitigation.     

Protocol for laboratory testing of crude-oil bioremediation products in freshwater conditions

In 1993, the Environmental Protection Agency, National Risk Management Research Laboratory (EPA, NRMRL), with the National Environmental Technology Application Center (NETAC), developed a protocol for evaluation of bioremediation products in marine environments [18]. The marine protocol was adapted for application in freshwater environments by using a chemically defined medium and an oil-degrading consortium as a positive control. Four products were tested using the modified protocol: two with nutrients and an oleophilic component; one with nutrients, sorbent, and organisms; and one microbial stimulant. A separate experiment evaluated the use of HEPES and MOPSO buffers as replacements for phosphate buffer. The oleophilic nutrient products yielded oil degradation similar to the positive control, with an average alkane removal of 97.1±2.3% and an aromatic hydrocarbon removal of 64.8±1.2%. The positive control, which received inoculum plus nutrients, demonstrated alkane degradation of 98.9±0.1% and aromatic degradation of 52.9±0.1%. The sorbent-based product with inoculum failed to demonstrate oil degradation, while the microbial stimulant showed less oil degradation than the positive control. Replacement of phosphate buffer with other buffers had no significant effect on one product's performance. Differences in product performance were easily distinguishable using the protocol, and performance targets for alkane and aromatic hydrocarbon degradation are suggested. Electronic Publication     

Improved yield of a ligand-binding GPCR expressed in E. coli for structural studies

The G-protein coupled receptor (GPCR), rat brain neurotensin receptor type I (NTS1) is one of a small number of GPCRs that have been successfully expressed in Escherichia coli as a functional, ligand-binding receptor, but yields of purified receptor are still low for comprehensive structural studies. Here, several approaches have been examined to optimize the yields of active, ligand-binding receptor. Optimisation of E. coli strain and induction protocol yielded a significant improvement in expression of active receptor. Expression of the receptor in BL21(DE3) cells, in combination with autoinduction improved expression 10-fold when compared with previously reported expression protocols using IPTG-mediated induction in DH5alpha cells. Optimization of the purification protocol revealed that supplementation of buffers with phospholipids enhanced recovery of active receptor. The methods examined are potentially applicable to other GPCRs expressed in E. coli.     

Solubilization of proteins from human lymph node tissue and two-dimensional gel storage

In the present study, we compared six different solubilization buffers and optimized two-dimensional electrophoresis (2-DE) conditions for human lymph node proteins. In addition, we developed a simple protocol for 2-D gel storage. Efficient solubilization was obtained with lysis buffers containing (a) 8 M urea, 4% CHAPS (3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate), 40 mM Tris base, 65 mM DTT (dithiothreitol) and 0.2% carrier ampholytes; (b) 5 M urea, 2 M thiourea, 2% CHAPS, 2% SB 3-10 (N-decyl-N,N-dimethyl-3-ammonio-1-propanesulfonate), 40 mM Tris base, 65 mM DTT and 0.2% carrier ampholytes or (c) 7 M urea, 2 M thiourea, 4% CHAPS, 65 mM DTT and 0.2% carrier ampholytes. The optimal protocol for isoelectric focusing (IEF) was accumulated voltage of 16,500 Vh and 0.6% DTT in the rehydration solution. In the experiments conducted for the sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), best results were obtained with a doubled concentration (50 mM Tris, 384 mM glycine, 0.2% SDS) of the SDS electrophoresis buffer in the cathodic reservoir as compared to the concentration in the anodic reservoir (25 mM Tris, 192 mM glycine, 0.1% SDS). Among the five protocols tested for gel storing, success was attained when the gels were stored in plastic bags with 50% glycerol. This is the first report describing the successful solubilization and 2D-electrophoresis of proteins from human lymph node tissue and a 2-D gel storage protocol for easy gel handling before mass spectrometry (MS) analysis.     

In vitro packaging and replication of individual genomic segments of bacteriophage phi 6 RNA.

The genome of bacteriophage phi 6 contains three segments of double-stranded RNA. Procapsid structures whose formation was directed by cDNA copies of the large genomic segment are capable of packaging the three viral message sense RNAs in the presence of ATP. Addition of UTP, CTP, and GTP results in the synthesis of minus strands to form double-stranded RNA. In this report, we show that procapsids are capable of taking up any of the three plus-strand single-stranded RNA segments independently of the others. In manganese-containing buffers, synthesis of the corresponding minus strand takes place. In magnesium-containing buffers, individual message sense viral RNA segments were packaged, but minus-strand replication did not take place unless all three viral single-stranded RNA segments were packaged. Since the conditions of packaging in magnesium buffer more closely resemble those in vivo, these results indicated that there is no specific order or dependence in packaging and that replication is regulated so that it does not begin until all segments are in place.     

Scalable PGAS collective operations in NUMA clusters

The increasing number of cores per processor is turning manycore-based systems in pervasive. This involves dealing with multiple levels of memory in non uniform memory access (NUMA) systems and processor cores hierarchies, accessible via complex interconnects in order to dispatch the increasing amount of data required by the processing elements. The key for efficient and scalable provision of data is the use of collective communication operations that minimize the impact of bottlenecks. Leveraging one sided communications becomes more important in these systems, to avoid unnecessary synchronization between pairs of processes in collective operations implemented in terms of two sided point to point functions. This work proposes a series of algorithms that provide a good performance and scalability in collective operations, based on the use of hierarchical trees, overlapping one-sided communications, message pipelining and the available NUMA binding features. An implementation has been developed for Unified Parallel C, a Partitioned Global Address Space language, which presents a shared memory view across the nodes for programmability, while keeping private memory regions for performance. The performance evaluation of the proposed implementation, conducted on five representative systems (JuRoPA, JUDGE, Finis Terrae, SVG and Superdome), has shown generally good performance and scalability, even outperforming MPI in some cases, which confirms the suitability of the developed algorithms for manycore architectures.     

Extraction of label during processing of brain tissue for electron microscopic autoradiographic investigation of glycoprotein metabolism

Rats were injected intracerebrally with (3H) fucose and survived for 30 min and 24 h. Following perfusion fixation with aldehydes, Vibratome sections of the brains were processed for embedding in resin and the radioactivity in the various solutions measured by scintillation counting. The decline of radioactivity was monitored in up to 18 successive buffer washes. Whereas 85% of the initial radioactivity could be eluted after 30 min, only 7% were extractable after 24 h survival time. Addition of unlabelled L-fucose to the fixative and buffers caused a small insignificant increase of the total radioactivity extracted. Only small amounts of radioactivity could be removed by treatment with trichloroacetic acid after extensive rinsing in buffer. It is concluded that thorough rinsing in buffer (which is more important with short-time experiments) is effective in removing the acid-soluble radioactivity. An additional safe-guard lies in dehydration by ethanol. The ethanol series also removes acid-soluble radioactivity in addition to small amounts of what is probably higher molecular weight material.     

Desiccation tolerance of spermatozoa dried at ambient temperature: production of fetal mice

Long-term preservation of mouse sperm by desiccation is economically and logistically attractive. The current investigation is a feasibility study of the preservation of mouse sperm by convective drying in an inert gas (nitrogen). Mouse sperm from the B6D2F1 strain isolated in an EGTA-supplemented Tris-HCl buffer were dried using three different drying rates and were stored for 18-24 h at 4 degrees C. The mean final moisture content was <5% for all the protocols. After intracytoplasmic sperm injection (ICSI), the mean blastocyst formation rates were 64%, 58%, and 35% using the rapid-, moderate-, and slow-drying protocols, respectively. The slow-drying protocol resulted in a rate of development significantly lower than that observed using rapid- and moderate-drying protocols and indicated that a slower drying rate may be detrimental to the DNA integrity of mouse sperm. The transfer of 85 two- or four-cell embryos that were produced using rapidly desiccated sperm resulted in 11 fetuses (13%) on Day 15 compared with the production of 34 fetuses (40%) produced using the transfer of 86 two- or four-cell embryos that were produced using fresh sperm (P < 0.05). The results demonstrate the feasibility of using a convective drying protocol for the successful desiccation of mouse sperm and identifies some of the important parameters required for optimization of the procedure.     

A quick method to isolate RNA from wheat and other carbohydrate-rich seeds

No reports on isolating RNA from carbohydrate-rich wheat seeds have been published. Because of the presence of carbohydrates, published protocols yield small amounts of poor quality RNA. Extracting seeds in a buffer (pH 9, 150 mM NaCl, 1% sarcosyl) ensured maximum RNA solubility and the removal of most interfering substances. Extracted RNA was purified using a guanidine hydrochloride-based buffer system. This protocol yields up to 148 μg of RNA from 100 mg of tissue in 3.5 h. An A₂₆₀/A₂₈₀ ratio of 1.85 indicates RNA purity. Isolated RNA was amenable to downstream applications such as differential display. The developed method was extended to other carbohydrate-rich seeds, such as barley and maize, with success.     

Gossip-Style Failure Detection and Distributed Consensus for Scalable Heterogeneous Clusters

Gossip protocols provide a means by which failures can be detected in large, distributed systems in an asynchronous manner without the limits associated with reliable multicasting for group communications. However, in order to be effective with application recovery and reconfiguration, these protocols require mechanisms by which failures can be detected with system-wide consensus in a scalable fashion. This paper presents three new gossip-style protocols supported by a novel algorithm to achieve consensus in scalable, heterogeneous clusters. The round-robin protocol improves on basic randomized gossiping by distributing gossip messages in a deterministic order that optimizes bandwidth consumption. Redundant gossiping is completely eliminated in the binary round-robin protocol, and the round-robin with sequence check protocol is a useful extension that yields efficient detection times without the need for system-specific optimization. The distributed consensus algorithm works with these gossip protocols to achieve agreement among the operable nodes in the cluster on the state of the system featuring either a flat or a layered design. The various protocols are simulated and evaluated in terms of consensus time and scalability using a high-fidelity, fault-injection model for distributed systems comprised of clusters of workstations connected by high-performance networks.     

Extraction of label during processing of brain tissue for electron microscopic autoradiographic investigation of glycoprotein metabolism

Summary Rats were injected intracerebrally with (³H) fucose and survived for 30 min and 24 h. Following perfusion fixation with aldehydes, Vibratome sections of the brains were processed for embedding in resin and the radioactivity in the various solutions measured by scintillation counting. The decline of radioactivity was monitored in up to 18 successive buffer washes. Whereas 85% of the initial radioactivity could be eluted after 30 min, only 7% were extractable after 24 h survival time. Addition of unlabelled L-fucose to the fixative and buffers caused a small insignificant increase of the total radioactivity extracted. Only small amounts of radioactivity could be removed by treatment with trichloroacetic acid after extensive rinsing in buffer. It is concluded that thorough rinsing in buffer (which is more important with shorttime experiments) is effective in removing the acid-soluble radioactivity. An additional safe-guard lies in dehydration by ethanol. The ethanol series also removes acid-soluble radioactivity in addition to small amounts of what is probably higher molecular weight material.     

Integration of bacteria capture via filtration and in situ lysis for recovery of plasmid DNA under industry-compatible conditions

Combining capture and lysis of the bacteria with partial purification of the plasmid DNA is beneficial for the design of efficient plasmid production processes at larger scale. Such an approach is possible when the bacteria are captured by filtration. Taking industrial requirements into account, however, such a capture requires complex filtration mixtures containing retentive additives such as bentonite and polycations. This makes the straightforward transfer of established lysis protocols to in situ lysis difficult. In this contribution, the different steps of such a protocol are designed for complex filter cakes, including fragilization (by lysozyme), lysis (alkaline pH/acidic pH, 70/37 degrees C, urea/NaCl/Triton), and specific elution (pH, NaCl, CaCl2, guanidinium hydrochloride). Results are compared in regard to plasmid quality (topoisomeric form) and quantity (compared to the yield obtained by a commercial miniprep of a small aliquot of the bacteria suspension from the bioreactor). Best results in these terms were obtained by the Triton lysis protocol performed at 37 degrees C (30 min of contact with a lysis buffer composed of 50 mM Tris pH 8, 1% Triton, 1 g/L lysozyme, and 6 M guanidinium hydrochloride) followed by the specific elution of the plasmid DNA in 50 mM Tris buffer pH 8.     

Sample preparation for two-dimensional gel electrophoresis

The choice of sample preparation protocol is a critical influential factor for isoelectric focusing which in turn affects the two-dimensional gel result in terms of quality and protein species distribution. The optimal protocol varies depending on the nature of the sample for analysis and the properties of the constituent protein species (hydrophobicity, tendency to form aggregates, copy number) intended for resolution. This review explains the standard sample buffer constituents and illustrates a series of protocols for processing diverse samples for two-dimensional gel electrophoresis, including hydrophobic membrane proteins. Current methods for concentrating lower abundance proteins, by removal of high abundance proteins, are also outlined. Finally, since protein staining is becoming increasingly incorporated into the sample preparation procedure, we describe the principles and applications of current (and future) pre-electrophoretic labelling methods.