Data dependences ( d a t a flow constraants) present a ma7or hurdle to the amount of anstructzon-level parallelasm that can be explozted from a program. Recent work has suggested that the lamats amposed b y data dependences can be overcome to some extent wath the use of data value predactaon. That as, when an anstructaon as fetched, ads result can be predacted so that subsequent anstructzons that depend on the result can use thas predacted value When the correct result becomes avazlable, all znstructzons that are data dependent on that predzctaon can be valzdated Thzs paper anvestagates a varzety of technzques to carry out haghly accurate data value predactzons The first technzqzle anvestagates the potentaal of monatorang the strades b y whach the results produced b y dzfferent anstances of an anstructaon change. The second technaque anvestagates the potentzal of pattern-based two-level predactaon schemes. Szmulatzon results of these two schemes show amprovements over the exzstang method of predactang the last outcome. In partzcular, some benchmarks show amprovement wath the strade-based predactor and others show zmprovement wath the pattern-based predactor. To do unaformly well across benchmarks, we combane these two predzctors to form a hybrad predactor Szmulataon analysts of the hybrad predzctor shows zts overall predactaon accuracy to be better than that of the component predactors across all benchmarks.

No abstract

In a series of five 17-d replicate trials, a total of 54 cannulated and 12 noncannulated pigs were used to determine the effects of weaning age (17 d or 24 d) on pH, dry matter percentage, aerobic and anaerobic microflora, lactate, and volatile fatty acid (VFA) concentrations in the jejunum, ileum, and cecum of weanling pigs. At -14 d of age, cannulated pigs were surgically fitted with T-cannulas in the jejunum (n = 20), ileum (n = 18), or cecum (n = 16). Upon weaning, cannulated pigs were individually caged in an environmentally controlled room with ad libitum access to a phase starter diet and water. Noncannulated pigs were killed at weaning and samples were collected from the jejunum, ileum, and cecum. Digesta and fecal swabs from cannulated pigs were collected twice weekly. The pH of cecal contents was lower (P < 0.05) and dry matter percentage was greater (P < 0.05) than those ofjejunal or ileal contents. Pigs weaned at 24 d of age had increased (P < 0.05) E. coli populations 3 d postweaning compared to preweaning populations, regardless of site of collection, whereas this increase was not observed in pigs weaned at 17 d of age. Unweaned pigs maintained higher (P < 0.05) lactobacilli populations compared to weaned pigs; however, populations declined (P < 0.05) in both groups by 3 d postweaning, with pigs weaned at 24 d of age having lactobacilli populations greater than pigs weaned at 17 d of age. Fecal populations of E. coli and lactobacilli declined (P < 0.05), whereas fecal bifidobacteria populations increased (P < 0.05) postweaning, regardless of weaning age. Concentrations of total fecal anaerobes declined (P < 0.05) in pigs weaned at 17 d of age but were maintained in pigs weaned at 24 d of age. Volatile fatty acid concentrations were greater (P < 0.05) in the cecum than in the jejunum or ileum, and acetic acid concentrations decreased (P < 0.05) postweaning regardless of weaning age. A tendency for L+ lactate concentrations to be greater (P < 0.07) in the ileum and jejunum vs the cecum was observed. Results indicate that weaning and weaning age have significant effects on microbial populations and VFA concentrations.

No abstract

This paper presents a benchmark, CommBench, for use in evaluating and designing telecommunications network processors. The benchmark applications focus on small, computationally intense program kernels typical of the network processor environment. The benchmark is composed of eight programs, four of them oriented towards packet header processing and four oriented towards data stream procesing. The benchmark is defined and various characteristics of the benchmark are presented. These include instruction frequencies, computational complexity, and cache performance. These measured characteristics are compared to the SPEC benchmark which has traditionally been used in evaluating workstation processors. Three examples are presented indicating how CommBench... Read complete Read complete