On the Viability of Microservers for Financial Analytics

Energy consumption and total cost of ownership are daunting challenges for Datacenters, because they scale disproportionately with performance. Datacenters running financial analytics may incur extremely high operational costs in order to meet performance and latency requirements of their hosted applications. Recently, ARM-based microservers have emerged as a viable alternative to high-end servers, promising scalable performance via scale-out approaches and low energy consumption. In this paper, we investigate the viability of ARM-based microservers for option pricing, using the Monte Carlo and Binomial Tree kernels. We compare an ARM-based microserver against a state-of-the-art x86 server. We define application-related but platform-independent energy and performance metrics to compare those platforms fairly in the context of datacenters for financial analytics and give insight on the particular requirements of option pricing. Our experiments show that through scaling out energyefficient compute nodes within a 2U rack-mounted unit, an ARM-based microserver consumes as little as about 60% of the energy per option pricing compared to an x86 server, despite having significantly slower cores. We also find that the ARM microserver scales enough to meet a high fraction of market throughput demand, while consuming up to 30% less energy than an Intel server

Chromosome synapsis and recombination in simple and complex chromosomal heterozygotes of tuco-tuco (Ctenomys talarum: Rodentia: Ctenomyidae):Rodentia: Ctenomyidae)

The chromosomal speciation hypothesis suggests that irregularities in synapsis, recombination, and segregation in heterozygotes for chromosome rearrangements may restrict gene flow between karyotypically distinct populations and promote speciation. Ctenomys talarum is a South American subterranean rodent inhabiting the coastal regions of Argentina, whose populations polymorphic for Robertsonian and tandem translocations seem to have a very restricted gene flow. To test if chromosomal differences are involved in isolation among its populations, we examined chromosome pairing, recombination, and meiotic silencing of unsynapsed chromatin in male meiosis of simple and complex translocation heterozygotes using immunolocalization of the MLH1 marking mature recombination nodules and phosphorylated histone γH2A.X marking unrepaired double-strand breaks. We observed small asynaptic areas labeled by γH2A.X in pericentromeric regions of the chromosomes involved in the trivalents and quadrivalents. We also observed a decrease of recombination frequency and a distalization of the crossover distribution in the heterozygotes and metacentric homozygotes compared to acrocentric homozygotes. We suggest that the asynapsis of the pericentromeric regions are unlikely to induce germ cell death and decrease fertility of the heterozygotes; however, suppressed recombination in pericentromeric areas of the multivalents may reduce gene flow between chromosomally different populations of the Talas tuco-tuco.

'Sorry I meant the patient's left side': impact of distraction on right/left discrimination

Context Medical students can have difficulty in distinguishing left from right. Many infamous medical errors have occurred when a procedure has been performed on the wrong side, such as in the removal of the wrong kidney. Clinicians encounter many distractions during their work. There is limited information on how these affect performance. 
Objectives Using a neuropsychological paradigm, we aim to elucidate the impacts of different types of distraction on left–right (LR) discrimination ability. 
Methods Medical students were recruited to a study with four arms: (i) control arm (no distraction); (ii) auditory distraction arm (continuous ambient ward noise); (iii) cognitive distraction arm (interruptions with clinical cognitive tasks), and (iv) auditory and cognitive distraction arm. Participants’ LR discrimination ability was measured using the validated Bergen Left–Right Discrimination Test (BLRDT). Multivariate analysis of variance was used to analyse the impacts of the different forms of distraction on participants’ performance on the BLRDT. Additional analyses looked at effects of demographics on performance and correlated participants’ self-perceived LR discrimination ability and their actual performance. 
Results A total of 234 students were recruited. Cognitive distraction had a greater negative impact on BLRDT performance than auditory distraction. Combined auditory and cognitive distraction had a negative impact on performance, but only in the most difficult LR task was this negative impact found to be significantly greater than that of cognitive distraction alone. There was a significant medium-sized correlation between perceived LR discrimination ability and actual overall BLRDT performance. 
Conclusions
Distraction has a significant impact on performance and multifaceted approaches are required to reduce LR errors. Educationally, greater emphasis on the linking of theory and clinical application is required to support patient safety and human factor training in medical school curricula. Distraction has the potential to impair an individual's ability to make accurate LR decisions and students should be trained from undergraduate level to be mindful of this.

Balls to the wall: How acoustic information from a ball in motion guides interceptive movement in people with Parkinson’s disease

Previous research has shown that Parkinson's disease (PD) patients can increase the speed of their movement when catching a moving ball compared to when reaching for a static ball (Majsak et al., 1998). A recent model proposed by Redgrave et al. (2010) explains this phenomenon with regard to the dichotomic organization of motor loops in the basal ganglia circuitry and the role of sensory micro-circuitries in the control of goal-directed actions. According to this model, external visual information that is relevant to the required movement can induce a switch from a habitual control of movement toward an externally-paced, goal-directed form of guidance, resulting in augmented motor performance (Bienkiewicz et al., 2013). In the current study, we investigated whether continuous acoustic information generated by an object in motion can enhance motor performance in an arm reaching task in a similar way to that observed in the studies of Majsak et al. (1998, 2008). In addition, we explored whether the kinematic aspects of the movement are regulated in accordance with time to arrival information generated by the ball's motion as it reaches the catching zone. A group of 7 idiopathic PD (6 male, 1 female) patients performed a ball-catching task where the acceleration (and hence ball velocity) was manipulated by adjusting the angle of the ramp. The type of sensory information (visual and/or auditory) specifying the ball's arrival at the catching zone was also manipulated. Our results showed that patients with PD demonstrate improved motor performance when reaching for a ball in motion, compared to when stationary. We observed how PD patients can adjust their movement kinematics in accordance with the speed of a moving target, even if vision of the target is occluded and patients have to rely solely on auditory information. We demonstrate that the availability of dynamic temporal information is crucial for eliciting motor improvements in PD. Furthermore, these effects appear independent from the sensory modality through-which the information is conveyed. 

Precision Teaching in IV Fluid Prescribing and Patient Safety

OBJECTIVES: Precision Teaching (PT) has been shown to be an effective intervention to assess teaching method effectiveness and evaluate learning outcomes. SAFMEDS (Say All Fast Minute Every Day Shuffled) are a practice/assessment procedure within the PT framework to assist learning and fluency. We explored the effects of a brief intervention with PT, to impart high frequency performance in safe intravenous fluid prescription in a group of final year undergraduate medical students.
METHODS: 133 final year undergraduate medical students completed a multiple choice question (MCQ) test on safe IV fluid prescription at the beginning and end of the study. The control group (n= 76) of students were taught using a current standardized teaching method. Students allocated to the intervention arm of the study were additionally instructed on PT and the use of SAFMEDS. The study group (n = 57) received 50 SAFMEDS cards containing information on the principles of IV fluid prescription scenarios. These students were trained/tested twice per day for 1 minute.
RESULTS: Interim analysis showed that the study group displayed an improvement in fluency and accuracy as the study progressed. There was a statistically significant improvement in MCQ performance for the PT group compared with the control group between the beginning and end of the study (35% vs 15%).
CONCLUSION: These results suggest PT employing SAFMEDS is an effective method for improving fluency, accuracy and patient safety in intravenous fluid prescribing amongst undergraduate medical students.

Evaluating Asymmetric Multicore Systems-on-Chip using Iso-Metrics

The end of Dennard scaling has pushed power consumption into a first order concern for current systems, on par with performance. As a result, near-threshold voltage computing (NTVC) has been proposed as a potential means to tackle the limited cooling capacity of CMOS technology. Hardware operating in NTV consumes significantly less power, at the cost of lower frequency, and thus reduced performance, as well as increased error rates. In this paper, we investigate if a low-power systems-on-chip, consisting of ARM's asymmetric big.LITTLE technology, can be an alternative to conventional high performance multicore processors in terms of power/energy in an unreliable scenario. For our study, we use the Conjugate Gradient solver, an algorithm representative of the computations performed by a large range of scientific and engineering codes.

A randomised comparison of daunorubicin 90mg/m2 vs 60mg/m2 in AML induction: results from the UK NCRI AML17 trial in 1206 patients

Modifying induction therapy in AML may improve the remission rate and reduce the risk of relapse thereby improving survival. Escalation of the daunorubicin dose to 90mg/m(2) has shown benefit for some patient subgroups when compared with a dose of 45mg/m(2) and has been recommended as a standard of care. However 60mg/m(2) is widely used and has never been directly compared to 90mg/m(2). As part of the UK NCRI AML17 trial 1206 adults with untreated AML or high risk MDS, mostly under 60 years of age, were randomised to a first induction course of chemotherapy which delivered either 90mg/m(2) or 60mg/m(2) on days 1,3 and 5 combined with cytosine arabinoside. All patients then received a second course which included daunorubicin 50mg/m(2) on days 1,3 and 5. There was no overall difference in complete remission rate (CR) (73% vs 75%, OR1.07 (0.83-1.39), p=0.6) or in any recognised subgroup. The 60 day mortality was increased in the 90mg/m2 arm (10% vs 5% (HR 1.98(1.30-3.02) p=0.001)), which resulted in no difference in overall 2 year survival (59% vs 60%, HR 1.16(0.95-1.43), p=0.15). In exploratory subgroup analysis there was no subgroup which showed significant benefit, although there was a significant interaction by FLT3 ITD mutation. The trial is registered to www.isrctn.com as ISRCTN55675535.

Vosaroxin and vosaroxin plus low-dose Ara-C (LDAC) vs low-dose Ara-C alone in older patients with acute myeloid leukemia

The development of new treatments for older patients with acute myeloid leukemia is an active area, but has met with limited success. Vosaroxin, a quinolone-derived intercalating agent has several properties that could prove beneficial. Initial clinical studies showed it to be well-tolerated in older patients with relapsed/refractory disease. In vitro data suggested synergy with cytarabine (Ara-C). To evaluate vosaroxin, we performed 2 randomized comparisons within the "Pick a Winner" program. A total of 104 patients were randomized to vosaroxin vs low-dose Ara-C (LDAC) and 104 to vosaroxin + LDAC vs LDAC. When comparing vosaroxin with LDAC, neither response rate (complete recovery [CR]/complete recovery with incomplete count recovery [CRi], 26% vs 30%; odds ratio [OR], 1.16 (0.49-2.72); P = .7) nor 12-month survival (12% vs 31%; hazard ratio [HR], 1.94 [1.26-3.00]; P = .003) showed benefit for vosaroxin. Likewise, in the vosaroxin + LDAC vs LDAC comparison, neither response rate (CR/CRi, 38% vs 34%; OR, 0.83 [0.37-1.84]; P = .6) nor survival (33% vs 37%; HR, 1.30 [0.81-2.07]; P = .3) was improved. A major reason for this lack of benefit was excess early mortality in the vosaroxin + LDAC arm, most obviously in the second month following randomization. At its first interim analysis, the Data Monitoring and Ethics Committee recommended closure of the vosaroxin-containing trial arms because a clinically relevant benefit was unlikely.

THE SAAF STUDY: evaluation of the Safeguarding Children Assessment and Analysis Framework (SAAF), compared with management as usual, for improving outcomes for children and young people who have experienced, or are at risk of, maltreatment: study protocol for a randomised controlled trial

Background: Serious case reviews and research studies have indicated weaknesses in risk assessments conducted by child protection social workers. Social workers are adept at gathering information but struggle with analysis and assessment of risk. The Department for Education wants to know if the use of a structured decision-making tool can improve child protection assessments of risk.

Methods/design: This multi-site, cluster-randomised trial will assess the effectiveness of the Safeguarding Children Assessment and Analysis Framework (SAAF). This structured decision-making tool aims to improve social workers' assessments of harm, of future risk and parents' capacity to change. The comparison is management as usual.

Inclusion criteria: Children's Services Departments (CSDs) in England willing to make relevant teams available to be randomised, and willing to meet the trial's training and data collection requirements.

Exclusion criteria: CSDs where there were concerns about performance; where a major organisational restructuring was planned or under way; or where other risk assessment tools were in use.

Six CSDs are participating in this study. Social workers in the experimental arm will receive 2 days training in SAAF together with a range of support materials, and access to limited telephone consultation post-training. The primary outcome is child maltreatment. This will be assessed using data collected nationally on two key performance indicators: the first is the number of children in a year who have been subject to a second Child Protection Plan (CPP); the second is the number of re-referrals of children because of related concerns about maltreatment. Secondary outcomes are: i) the quality of assessments judged against a schedule of quality criteria and ii) the relationship between the three assessments required by the structured decision-making tool (level of harm, risk of (re) abuse and prospects for successful intervention).

Discussion: This is the first study to examine the effectiveness of SAAF. It will contribute to a very limited literature on the contribution that structured decision-making tools can make to improving risk assessment and case planning in child protection and on what is involved in their effective implementation.

Long-Term Follow-Up of a Large Active Surveillance Cohort of Patients With Prostate Cancer

PURPOSE: Active surveillance is increasingly accepted as a treatment option for favorable-risk prostate cancer. Long-term follow-up has been lacking. In this study, we report the long-term outcome of a large active surveillance protocol in men with favorable-risk prostate cancer.

PATIENTS AND METHODS: In a prospective single-arm cohort study carried out at a single academic health sciences center, 993 men with favorable- or intermediate-risk prostate cancer were managed with an initial expectant approach. Intervention was offered for a prostate-specific antigen (PSA) doubling time of less than 3 years, Gleason score progression, or unequivocal clinical progression. Main outcome measures were overall and disease-specific survival, rate of treatment, and PSA failure rate in the treated patients.

RESULTS: Among the 819 survivors, the median follow-up time from the first biopsy is 6.4 years (range, 0.2 to 19.8 years). One hundred forty-nine (15%) of 993 patients died, and 844 patients are alive (censored rate, 85.0%). There were 15 deaths (1.5%) from prostate cancer. The 10- and 15-year actuarial cause-specific survival rates were 98.1% and 94.3%, respectively. An additional 13 patients (1.3%) developed metastatic disease and are alive with confirmed metastases (n = 9) or have died of other causes (n = 4). At 5, 10, and 15 years, 75.7%, 63.5%, and 55.0% of patients remained untreated and on surveillance. The cumulative hazard ratio for nonprostate-to-prostate cancer mortality was 9.2:1.

CONCLUSION: Active surveillance for favorable-risk prostate cancer is feasible and seems safe in the 15-year time frame. In our cohort, 2.8% of patients have developed metastatic disease, and 1.5% have died of prostate cancer. This mortality rate is consistent with expected mortality in favorable-risk patients managed with initial definitive intervention.

Methods and Metrics for Fair Server Assessment under Real-Time Financial Workloads

We present a rigorous methodology and new metrics for fair comparison of server and microserver platforms. Deploying our methodology and metrics, we compare a microserver with ARM cores against two servers with ×86 cores running the same real-time financial analytics workload. We define workload-specific but platform-independent performance metrics for platform comparison, targeting both datacenter operators and end users. Our methodology establishes that a server based on the Xeon Phi co-processor delivers the highest performance and energy efficiency. However, by scaling out energy-efficient microservers, we achieve competitive or better energy efficiency than a power-equivalent server with two Sandy Bridge sockets, despite the microserver's slower cores. Using a new iso-QoS metric, we find that the ARM microserver scales enough to meet market throughput demand, that is, a 100% QoS in terms of timely option pricing, with as little as 55% of the energy consumed by the Sandy Bridge server.

ALEA: Fine-Grain Energy Profiling with Basic Block Sampling

Energy efficiency is an essential requirement for all contemporary computing systems. We thus need tools to measure the energy consumption of computing systems and to understand how workloads affect it. Significant recent research effort has targeted direct power measurements on production computing systems using on-board sensors or external instruments. These direct methods have in turn guided studies of software techniques to reduce energy consumption via workload allocation and scaling. Unfortunately, direct energy measurements are hampered by the low power sampling frequency of power sensors. The coarse granularity of power sensing limits our understanding of how power is allocated in systems and our ability to optimize energy efficiency via workload allocation.
We present ALEA, a tool to measure power and energy consumption at the granularity of basic blocks, using a probabilistic approach. ALEA provides fine-grained energy profiling via sta- tistical sampling, which overcomes the limitations of power sens- ing instruments. Compared to state-of-the-art energy measurement tools, ALEA provides finer granularity without sacrificing accuracy. ALEA achieves low overhead energy measurements with mean error rates between 1.4% and 3.5% in 14 sequential and paral- lel benchmarks tested on both Intel and ARM platforms. The sampling method caps execution time overhead at approximately 1%. ALEA is thus suitable for online energy monitoring and optimization. Finally, ALEA is a user-space tool with a portable, machine-independent sampling method. We demonstrate two use cases of ALEA, where we reduce the energy consumption of a k-means computational kernel by 37% and an ocean modelling code by 33%, compared to high-performance execution baselines, by varying the power optimization strategy between basic blocks.