Enric Herrero

Elastic cooperative caching: an autonomous dynamically adaptive memory hierarchy for chip multiprocessors

Next generation tiled microarchitectures are going to be limited by off-chip misses and by on-chip network usage. Furthermore, these platforms will run an heterogeneous mix of applications with very different memory needs, leading to significant optimization opportunities. Existing adaptive memory hierarchies use either centralized structures that limit the scalability or software based resource allocation that increases programming complexity. We propose Elastic Cooperative Caching, a dynamic and scalable memory hierarchy that adapts automatically and autonomously to application behavior for each node. Our configuration uses elastic shared/private caches with fully autonomous and distributed repartitioning units for better scalability. Furthermore, we have extended our elastic configuration with an Adaptive Spilling mechanism to use the shared cache space only when it can produce a performance improvement. Elastic caches allow both the creation of big local private caches for threads with high reuse of private data and the creation of big shared spaces from unused caches. Local data allocation in private regions allows to reduce network usage and efficient cache partitioning allows to reduce off-chip misses. The proposed scheme outperforms previous proposals by a minimum of 12% (on average across the benchmarks) and reduces the number of offchip misses by 16%. Plus, the dynamic and autonomous management of cache resources avoids the reallocation of cache blocks without reuse which results in an increase in energy efficiency of 24%.

Distributed cooperative caching

This paper presents the Distributed Cooperative Caching, a scalable and energy-efficient scheme to manage chip multiprocessor (CMP) cache resources. The proposed configuration is based in the Cooperative Caching framework [3] but it is intended for large scale CMPs. Both centralized and distributed configurations have the advantage of combining the benefits of private and shared caches. In our proposal, the Coherence Engine has been redesigned to allow its partitioning and thus, eliminate the size constraints imposed by the duplication of all tags. At the same time, a global replacement mechanism has been added to improve the usage of cache space. Our framework uses several Distributed Coherence Engines spread across all the nodes to improve scalability. The distribution permits a better balance of the network traffic over the entire chip avoiding bottlenecks and increasing performance for a 32-core CMP by 21% over a traditional shared memory configuration and by 57% over the Cooperative Caching scheme. Furthermore, we have reduced the power consumption of the entire system by using a different tag allocation method and by reducing the number of tags compared on each request. For a 32-core CMP the Distributed Cooperative Caching framework provides an average improvement of the power/performance relation (MIPS3/W) of 3.66× over a traditional shared memory configuration and 4.30× over Cooperative Caching.

Distributed Cooperative Caching: An Energy Efficient Memory Scheme for Chip Multiprocessors

Current trends in CMPs indicate that the core count will increase in the near future. One of the main performance limiters of these forthcoming microarchitectures is the latency and high demand of the on-chip network and the off-chip memory communication. One of the main trade-offs when searching an optimal cache hierarchy is the sharing degree of cache space and its on-die distribution. Several techniques have appeared recently that optimize these parameters to get a better performance. This work provides some insight in the most promising configurations for tiled microarchitectures and shows the advantages and limitations of each of them in terms of performance and energy efficiency. This paper extends previous works by providing a complete study that evaluates different network topologies, single and multithreaded benchmarks, and single and multiprogrammed execution. In all these studies, the Distributed Cooperative Caching shows to be a promising alternative to traditional configurations for chip multiprocessors, providing a scalable and energy efficient solution.

Thread Row Buffers: Improving Memory Performance Isolation and Throughput in Multiprogrammed Environments

The widespread adoption of chip multiprocessors in recent years has increased the number of applications simultaneously accessing DRAM memories. Therefore, memory access patterns have also changed and this has reduced row buffer locality significantly, degrading performance and energy efficiency. Furthermore, concurrent execution of applications also has shown the need of performance isolation among threads in the memory controller to enforce a quality of service in virtualized environments. Existing DRAM memories, however, enforce a tradeoff between throughput and isolation. To solve these problems, this paper proposes the addition of Thread Row Buffers (TRBs) to DRAM memories. TRBs keep an active row per thread, thereby increasing DRAM efficiency by avoiding alternate accesses to a limited number of rows and allowing the implementation of a memory scheduler not bound to the throughput-isolation tradeoff. Thread Row Buffers with Service Partitioning (TRB-SP) increase the row hit-rate by 38 percent with respect to FR-FCFS and by 11 percent with respect to Cache DRAM. This, in turn, increases overall performance by 17 and 7 percent, respectively. TRB-SP is also able to reduce the standard deviation of the memory access time of an application by 40 percent over FR-FCFS, 31 percent over PAR-BS, and 42 percent over Cache DRAM.

Power-efficient spilling techniques for chip multiprocessors

Current trends in CMPs indicate that the core count will increase in the near future. One of the main performance limiters of these forthcoming microarchitectures is the latency and high-demand of the on-chip network and the off-chip memory communication. To optimize the usage of on-chip memory space and reduce off-chip traffic several techniques have proposed to use the N-chance forwarding mechanism, a solution for distributing unused cache space in chip multiprocessors. This technique, however, can lead in some cases to extra unnecessary network traffic or inefficient cache allocation. This paper presents two alternative power-efficient spilling methods to improve the efficiency of the N-chance forwarding mechanism. Compared to traditional Spilling, our Distance-Aware Spilling technique provides an energy efficiency improvement (MIPS3/W) of 16% on average, and a reduction of the network usage of 14% in a ring configuration while increasing performance 6%. Our Selective Spilling technique is able to avoid most of the unnecessary reallocations and it doubles the reuse of spilled blocks, reducing network traffic by an average of 22%. A combination of both techniques allows to reduce the network usage by 30% on average without degrading performance, allowing a 9% increase of the energy efficiency.

Development and validation of hydrophobic molecular fields derived from the quantum mechanical IEF/PCM-MST solvation models in 3D-QSAR.

Since the development of structure–activity relationships about 50 years ago, 3D‐QSAR methods belong to the most refined ligand‐based in silico techniques for prediction of biological data using physicochemical molecular fields. In this scenario, this study reports the development and validation of quantum mechanical (QM)‐based hydrophobic descriptors derived from the parametrized MST continuum solvation model to be used in 3D‐QSAR studies within the framework of the Hydrophobic Pharmacophore (HyPhar) method. To this end, five sets of compounds reported in the literature (dopamine D2/D4 antagonists, antifungal 2‐aryl‐4‐chromanones, and inhibitors of GSK‐3, cruzain and thermolysin) have been revisited. The results derived from the QM/MST‐based hydrophobic descriptors have been compared with previous CoMFA and CoMSIA studies, and examined in light of the available X‐ray crystallographic structures of the targets. The analysis reveals that the combination of electrostatic and nonelectrostatic components of the octanol/water partition coefficient yields pharmacophoric models fully comparable with the predictive potential of standard 3D‐QSAR techniques. Moreover, the graphical representation of the hydrophobic maps provides a direct linkage with the pattern of interactions found in crystallographic structures. Overall, the introduction of the QM/MST‐based descriptors, which could be easily adapted to other continuum solvation formalisms, paves the way to novel computational strategies for disclosing structure–activity relationships in drug design.

Development and Validation of Molecular Overlays Derived From 3D Hydrophobic Similarity with PharmScreen

Molecular alignment is a standard procedure for three-dimensional (3D) similarity measurements and pharmacophore elucidation. This process is influenced by several factors, such as the physicochemical descriptors utilized to account for the molecular determinants of biological activity and the reference templates. Relying on the hypothesis that the maximal achievable binding affinity for a drug-like molecule is largely due to desolvation, we explore a novel strategy for 3D molecular overlays that exploits the partitioning of molecular hydrophobicity into atomic contributions in conjunction with information about the distribution of hydrogen-bond (HB) donor/acceptor groups. A brief description of the method, as implemented in the software package PharmScreen, including the derivation of the fractional hydrophobic contributions within the quantum mechanical version of the Miertus-Scrocco-Tomasi (MST) continuum model, and the procedure utilized for the optimal superposition between molecules, is presented. The computational procedure is calibrated by using a data set of 402 molecules pertaining to 14 distinct targets taken from the literature and validated against the AstraZeneca test, which comprises 121 experimentally derived sets of molecular overlays. The results point out the suitability of the MST-based hydrophobic parameters for generating molecular overlays, as correct predictions were obtained for 94%, 79%, and 54% of the molecules classified into easy, moderate, and hard sets, respectively. Moreover, the results point out that this accuracy is attained at a much lower degree of identity between the templates used by hydrophobic/HB fields and electrostatic/steric ones. These findings support the usefulness of the hydrophobic/HB descriptors to generate complementary overlays that may be valuable to rationalize structure-activity relationships and for virtual screening campaigns.

Application of the quantum mechanical IEF/PCM-MST hydrophobic descriptors to selectivity in ligand binding

AbstractWe have recently reported the development and validation of quantum mechanical (QM)-based hydrophobic descriptors derived from the parametrized IEF/PCM-MST continuum solvation model for 3D-QSAR studies within the framework of the Hydrophobic Pharmacophore (HyPhar) method. In this study we explore the applicability of these descriptors to the analysis of selectivity fields. To this end, we have examined a series of 88 compounds with inhibitory activities against thrombin, trypsin and factor Xa, and the HyPhar results have been compared with 3D-QSAR models reported in the literature. The quantitative models obtained by combining the electrostatic and non-electrostatic components of the octanol/water partition coefficient yield results that compare well with the predictive potential of standard CoMFA and CoMSIA techniques. The results also highlight the potential of HyPhar descriptors to discriminate the selectivity of the compounds against thrombin, trypsin, and factor Xa. Moreover, the graphical representation of the hydrophobic maps provides a direct linkage with the pattern of interactions found in crystallographic structures. Overall, the results support the usefulness of the QM/MST-based hydrophobic descriptors as a complementary approach for disclosing structure-activity relationships in drug design and for gaining insight into the molecular determinants of ligand selectivity. Graphical AbstractQuantum Mechanical continuum solvation calculations performed with the IEF/PCM-MST method are used to derived atomic hydrophobic descriptors, which are then used to discriminate the selectivity of ligands against thrombin, trypsin and factor Xa. The descriptors provide complementary view to standard 3D-QSAR analysis, leading to a more comprehensive understanding of ligand recognition.

Variaciones en la concentracion serica de oligoelementos en el curso de neumonias bacterianas

En 20 pacientes con neumonia bacteriana, determinamos secuencialmente, los dias 2.°, 5.° y 8.° de enfermedad y un mes tras la curacion del proceso, la concentracion serica de 3 oligoelementos, hierro (Fe), zinc (Zn) y cobre (Cu), asi como de zinc intraeritrocitario (Zn-H) y de las proteinas transportadoras de dos de ellos, la trans-ferrina (TF) y la cerulaplasmina (Ce). En nuestros pacientes, objetivamos una muy significativa hiposideremia coincidiendo con las fases de maxima leucocitosis, que ademas se correlaciono directamente con un descenso paralelo de la TF. El Zn serico, sufrio un precoz y significativo descenso al inicio de la neumonia, que se correlaciono inversamente con la cifra de polimorfonucleares, como expresion de una probable relacion fisiopatologica entre ambos. Por el contrario, el Zn-H, no se modifico a lo largo del proceso, lo cual apoya el que la hipozincemia obedezca a un mecanismo de redistribucion y no a una autentica depleccion corporal. El Cu serico, a diferencia del Zn y del Fe, experimento un moderado ascenso que se correlaciono estrechamente con el de la Ce, y que persistio hasta la curacion de la enfermedad. Esto, hace pensar que la hipercupremia en las neumonias, obedezca al aumento de su proteina transportadora.

Renal Handling of Uric Acid in Normal Subjects: Its Behaviour with Respect to Different Filtered Loads

Serum uric acid concentration in man is the result of complex biosynthetic, catabolic, and excretory processes. Recent investigations of the urate transport system in the human kidney have provided evidence for a four-component model: glomerular filtration, presecretory reabsorption, tubular secretion, and postsecretory reabsorption1.