<p>With the increasing capacity of today's integrated circuits, a number ofheterogeneous  system-on-chip (SoC)  architectures  in embedded  systemshave been proposed. In order to achieve energy and design cost efficientstreaming applications  on these  systems, new design  space explorationframeworks  and  performance  analysis  approaches are  required.   Thisthesis  considers three state-of-the-art  SoCs architectures,  i.e., themulti-processor SoCs (MPSoCs)  with network-on-chip (NoC) communication,the hybrid CPU/FPGA architectures, and the run-time reconfigurable (RTR)FPGAs.  The main topic of the  author?s research is to model and capturethe  application  scheduling,  architecture  customization,  and  bufferdimensioning  problems, according to  the real-time  requirement.  Sincethese  problems  are NP-complete,  heuristic  algorithms and  constraintprogramming solver are used to compute a solution.For  NoC  communication  based  MPSoCs,  an  approach  to  optimize  thereal-time    streaming    applications    with   customized    processorvoltage-frequency levels and memory  sizes is presented. A multi-clockedsynchronous  model  of  computation   (MoC)  framework  is  proposed  inheterogeneous  timing analysis and  energy estimation.   Using heuristicsearching  (i.e., greedy  and  taboo search),  the  experiments show  anenergy reduction (up to 21%)  without any loss in application throughputcompared with an ad-hoc approach.On hybrid CPU/FPGA architectures,  the buffer minimization scheduling ofreal-time streaming  applications is addressed.  Based  on event models,the  problem  has  been  formalized  decoratively  as  constraint  basescheduling,  and  solved  by  public domain  constraint  solver  Gecode.Compared  with  traditional  PAPS  method,  the  proposed  method  needssignificantly smaller  buffers (2.4%  of PAPS in  the best  case), whilehigh throughput guarantees can still be achieved.Furthermore, a  novel compile-time analysis approach  based on iterativetiming  phases is  proposed  for run-time  reconfigurations in  adaptivereal-time   streaming   applications  on   RTR   FPGAs.   Finally,   thereconfigurations analysis and design trade-offs analysis capabilities ofthe proposed  framework have been  exemplified with experiments  on bothexample and industrial applications.</p> / Andres

With the increasing capacity of today's integrated circuits, a number ofheterogeneous  system-on-chip (SoC)  architectures  in embedded  systemshave been proposed. In order to achieve energy and design cost efficientstreaming applications  on these  systems, new design  space explorationframeworks  and  performance  analysis  approaches are  required.   Thisthesis  considers three state-of-the-art  SoCs architectures,  i.e., themulti-processor SoCs (MPSoCs)  with network-on-chip (NoC) communication,the hybrid CPU/FPGA architectures, and the run-time reconfigurable (RTR)FPGAs.  The main topic of the  author?s research is to model and capturethe  application  scheduling,  architecture  customization,  and  bufferdimensioning  problems, according to  the real-time  requirement.  Sincethese  problems  are NP-complete,  heuristic  algorithms and  constraintprogramming solver are used to compute a solution.For  NoC  communication  based  MPSoCs,  an  approach  to  optimize  thereal-time    streaming    applications    with   customized    processorvoltage-frequency levels and memory  sizes is presented. A multi-clockedsynchronous  model  of  computation   (MoC)  framework  is  proposed  inheterogeneous  timing analysis and  energy estimation.   Using heuristicsearching  (i.e., greedy  and  taboo search),  the  experiments show  anenergy reduction (up to 21%)  without any loss in application throughputcompared with an ad-hoc approach.On hybrid CPU/FPGA architectures,  the buffer minimization scheduling ofreal-time streaming  applications is addressed.  Based  on event models,the  problem  has  been  formalized  decoratively  as  constraint  basescheduling,  and  solved  by  public domain  constraint  solver  Gecode.Compared  with  traditional  PAPS  method,  the  proposed  method  needssignificantly smaller  buffers (2.4%  of PAPS in  the best  case), whilehigh throughput guarantees can still be achieved.Furthermore, a  novel compile-time analysis approach  based on iterativetiming  phases is  proposed  for run-time  reconfigurations in  adaptivereal-time   streaming   applications  on   RTR   FPGAs.   Finally,   thereconfigurations analysis and design trade-offs analysis capabilities ofthe proposed  framework have been  exemplified with experiments  on bothexample and industrial applications. / Andres

FARIAS, Pablo Mayckon Silva. O problema de ordenação de rodadas e problemas de otimização associados. 2013. 127 f. Tese (Doutorado em ciência da computação)- Universidade Federal do Ceará, Fortaleza-CE, 2013. / Submitted by Elineudson Ribeiro (elineudsonr@gmail.com) on 2016-07-20T12:44:36Z No. of bitstreams: 1 2013_tese_pmsfarias.pdf: 1600516 bytes, checksum: afaccc349ce6f2d065494cd2601913c0 (MD5) / Approved for entry into archive by Rocilda Sales (rocilda@ufc.br) on 2016-07-25T12:05:15Z (GMT) No. of bitstreams: 1 2013_tese_pmsfarias.pdf: 1600516 bytes, checksum: afaccc349ce6f2d065494cd2601913c0 (MD5) / Made available in DSpace on 2016-07-25T12:05:15Z (GMT). No. of bitstreams: 1 2013_tese_pmsfarias.pdf: 1600516 bytes, checksum: afaccc349ce6f2d065494cd2601913c0 (MD5) Previous issue date: 2013 / This thesis is composed of three well-delineated parts. In the first part, we introduce the round sorting problem (RSP), which models the minimization of the usage of buffer for the temporary storage of packets to be forwarded in TDMA communications of wireless mesh networks. We present a complete foundation for the definition of the RSP, and show that the problem is NP-hard for two theoretical models of radio interference known in the literature. A mixed integer programming formulation is also presented for a purely combinatorial and applicationindependent generalization of the RSP, the SMSP problem. In the second part of the work, we deal with problems about queries on insertions into sequences of numbers. Our main result in this part of the thesis is to show how, after a preprocessing step which runs in linear time on a sequence A of arbitrary real numbers, it is possible to compute in constant time the greatest sum of a (circular or not) contiguous subsequence of the sequence which results from the insertion of a given real number x into a given position p of A. In the third part of the thesis, we use the query algorithms from the second part to obtain an efficient implementation of the GRASP metaheuristic applied to the SMSP problem. An experimental analysis of this implementation is described, in which the values of the solutions returned by the metaheuristic are compared with those of the solutions obtained through the mixed integer formulation, in the case of small instances, and with the available lower bound, in the case of larger instances. / Esta tese é composta de três partes bem-delineadas. Na primeira parte, nós introduzimos o "problema de ordenação de rodadas" (POR), que modela a minimização do uso de memória ("buffer") para o armazenamento temporário de pacotes a serem repassados em comunicações TDMA de redes de rádio em malha. Nós apresentamos uma fundamentação completa para a definição do POR, e mostramos que o problema é NP-difícil para dois modelos teóricos de interferência de rádio conhecidos na literatura. Uma formulação de programação inteira mista é também apresentada para uma generalização puramente combinatória e independente de aplicação do POR, o problema SMSP. Na segunda parte do trabalho, nós abordamos problemas de consulta sobre inserções em sequências de números. O nosso principal resultado nesta parte da tese é mostrar como, após um passo de pré-processamento que executa em tempo linear sobre uma sequência "A" de números reais quaisquer, é possível computar em tempo constante a maior soma de uma subsequência contígua (circular ou não) da sequência que resulta da inserção de dado um número real "x" numa dada posição "p" de "A". Na terceira parte da tese, nós utilizamos os algoritmos de consulta da segunda parte para obter uma implementação eficiente da meta-heurística GRASP aplicada ao problema SMSP. Uma análise experimental dessa implementação é descrita, onde os valores das soluções retornadas pela meta-heurística são comparados com os das soluções obtidas pela formulação inteira mista, no caso de instâncias pequenas, e com o limite inferior disponível, no caso de instâncias maiores.