On the design of on-chip instruction caches

Cache memories reduce memory latency and traffic in computing systems. Most existing caches are implemented as board-based systems. Advancing VLSI technology will soon permit significant caches to be integrated on chip with the processors they support. In designing on-chip caches, the constraints of VLSI become significant. The primary constraints are economic limitations on circuit area and off-chip communications. The paper explores the design of on-chip instruction-only caches in terms of these constraints. The primary contribution of this work is the development of a unified economic model of on-chip instruction-only cache design which integrates the points of view of the cache designer and of the floorplan architect. With suitable data, this model permits the rational allocation of constrained resources to the achievement of a desired cache performance. Specific conclusions are that random line replacement is superior to LRU replacement, due to an increased flexibility in VLSI floorplan design; that variable set associativity can be an effective tool in regulating a chip's floorplan; and that sectoring permits area efficient caches while avoiding high transfer widths. Results are reported on economic functionality, from chip area and transfer width to miss ratio. These results, or the underlying analysis, can be used by microprocessor architects to make intelligent decisions regarding appropriate cache organizations and resource allocations.     

Three-dimensional structure of lipoamide dehydrogenase from Pseudomonas fluorescens at 2.8 A resolution. Analysis of redox and thermostability properties

The structure of Pseudomonas fluorescens lipoamide dehydrogenase, a dimeric flavoenzyme with a molecular mass of 106,000 daltons, was solved by the molecular replacement method and refined to an R-factor of 19.4% at 2.8 A resolution. The root-mean-square difference from ideal values for bonds and angles is 0.019 A and 3.8 degrees, respectively. The structure is closely related to that of the same flavoprotein from Azotobacter vinelandii. The root-mean-square difference for 932 C alpha atoms is 0.64 A, with 84% sequence identity. The residues in the active site are identical, while 89% of the interface residues are the same in the two enzymes. A few structural variations provide the basis for the differences in thermostability and redox properties between the two homologous proteins. Particularly, in the A. vinelandii molecule a threonine to alanine (T452A) mutation leaves a buried carbonyl oxygen, located at the subunit interface and in proximity of the flavin ring, unpaired to any H-bond donor, probably providing an explanation for the lower stability of the A. vinelandii enzyme with respect to the P. fluorescens enzyme. Six surface loops, which previously could not be accurately positioned in the A. vinelandii structure, are well defined in P. fluorescens lipoamide dehydrogenase. On the basis of the P. fluorescens structure, the six loops could be correctly defined also in the A. vinelandii enzyme. This is an unusual case where similar refinement methodologies applied to two crystal forms of closely related proteins led to electron density maps of substantially different quality. The correct definition of these surface residues is likely to be an essential step for revealing the structural basis of the interactions between lipoamide dehydrogenase and the other members of the pyruvate dehydrogenase multienzyme complex.     

Lipase-catalyzed transformations with unnatural acyl acceptors

Lipases are readily available and extremely versatile enzymes. Besides their natural reaction — hydrolysis of lipids — lipases catalyze a large number of other reactions involving a plethora of acyl acceptors such as alcohols, amines, ammonia and hydrogen peroxide. They are also able to accommodate a variety of different structures in the acyl donor. With chiral substrates many of these reactions proceed with a high degree of enantioselectivity. The scope of these lipase-catalyzed transformations with different acyl acceptors is the subject of this review.     

Flow cytometric quantitation of yeast a novel technique for use in animal model work and in vitro immunologic assays

Animal models of fungal and other infectious diseases often require that the number of organisms in tissue be quantified, traditionally by grinding organs, plating them on agar and counting colony forming units (CFU). This method is labor intensive, slow as some fungi require two weeks of culture and limited in reliability by poor plating efficiency. To circumvent these problems, we developed a flow cytometric method to quantify yeast. In vitro cultured Blastomyces dermatitidis, Cryptococcus neoformans, Candida albicans and Histoplasma capsulatum yeast were labelled with specific monoclonal or polyclonal antibodies to stain surface determinants or with Calcofluor to stain cell-wall chitin. A defined number of fluorescently labelled beads were added prior to acquisition by flow cytometry as a reference standard for quantitation. Beads were readily distinguished from yeast by forward scatter, side scatter and intensity of fluorescence. Cultured yeast were enumerated by both standard CFU determination and flow cytometry in a range of 10(2) to 10(7) cells. Only flow cytometry enabled discrimination of live and dead yeast by using appropriate fluorescent dyes. The flow cytometric method was applied to murine models of histoplasmosis and blastomycosis to quantify the burden of fungi in the lungs of infected mice. Labelling yeast with Calcofluor alone resulted in unacceptably high levels of nonspecific binding to mouse cell debris. In contrast, labelling H. capsulatum with a rabbit polyclonal antiserum and B. dermatitidis with a monoclonal antibody to the surface protein WI-1 permitted accurate quantitation. We conclude that this flow cytometry technique is rapid, efficient and reliable for quantifying the burden of infection in animal models of fungal disease. The technique also should lend itself to performing cytotoxicity assays that require discrimination of live and dead fungi, or phagocytosis assays that require discrimination of intracellular and extracellular organisms.     

Highly-efficient gene transfer with retroviral vectors into human T lymphocytes on fibronectin

Genetically modified lymphocytes have been successfully used for correction of ADA deficiency in children and in controlling graft-versus-host disease (GvHD) after allogeneic bone marrow transplantation. Low transduction efficiencies are, however, limiting for gene therapeutic strategies based on lymphocytes. In this study we compared protocols for highly efficient gene transfer into human T cells using retroviral vector-containing supernatant. We showed that infection of both human primary T cells and CD4+ Jurkat cells is most efficient on the matrix component fibronectin. Transduction was carried out with a retroviral vector encoding both the human intracytoplasmatically truncated low-affinity nerve growth factor receptor (deltaLNGFR) as a gene transfer marker and the Herpes simplex virus thymidine kinase for negative selection. Based on LNGFR expression genetically modified cells were enriched to near purity by magnetic cell sorting (MACS). Enriched cells could be shown to be highly sensitive to ganciclovir.