Haskell in Green Land: Analyzing the Energy Behavior of a Purely Functional Language
•Transferir como PPTX, PDF•
1 gostou•426 visualizações
Presentation @ SANER'16
Recomendados
Recomendados
Mais conteúdo relacionado
Mais de GreenLabAtDI
Mais de GreenLabAtDI (12)
Último
Último (20)
Haskell in Green Land: Analyzing the Energy Behavior of a Purely Functional Language
- 1. Haskell in Green Land: Analyzing the Energy Behavior of a Purely Functional Language Luís Gabriel Lima1, Gilberto Melfe2, Francisco Soares-Neto1, Paulo Lieuthier1, João Paulo Fernandes2, and Fernando Castor1 1 {lgnfl, fmssn, pvjl, castor}@cin.ufpe.br 2 gilbertomelfe@gmail.com, jpf@di.ubi.pt 1
- 2. Global energy system is unsustainable 2
- 3. 3
- 4. 4
- 5. Haskell has feelings too! 5
- 6. rq 6 Can we save energy by refactoring Haskell programs to use different data structure implementations or concurrent programming constructs?
- 7. Experimental Setup 2x10-core Intel Xeon E5-2660 v2 processors(Ivy Bridge) 256GB of DDR3 1600MHz RAM Criterion RAPL 7
- 8. Study 1: Purely functional data structures 8
- 9. RQ1. How do different implementations of the same abstractions compare in terms of run time and energy efficiency? 9
- 10. RQ2. For concrete operations, what is the relationship between their performance and energy consumption? 10
- 11. Study 1: Edison Library Collections Associative Collections Sequences EnumSet StandardSet UnbalancedSet LazyPairingHeap LeftistHeap MinHeap SkewHeap SplayHeap AssocList PatriciaLoMap StandardMap TernaryTrie BankersQueue SimpleQueue BinaryRandList JoinList RandList BraunSeq FingerSeq ListSeq RevSeq SizedSeq MyersStack 11
- 12. Study 1: Benchmark iters operation base aux 1 1000 1 1000 5000 1 10000 10 5000 10 add addAll clear contains containsAll iterator remove removeAll toArray retainAll 100000 100000 100000 100000 100000 100000 100000 100000 100000 100000 100000 1000 n.a. 1 1000 n.a. 1 1000 n.a. 1000 12
- 13. 13 iters = 0; while iters < 10 retainAll base aux; iters++;
- 17. Study 1: Findings 17 RQ1. How do different implementations of the same abstractions compare in terms of runtime and energy efficiency? RQ2. For concrete operations, what is the relationship between their performance and their energy consumption? Energy is proportional to execution time. Full details on green-haskell.github.io.
- 18. Study 2: Concurrent programming constructs 18
- 19. 19 RQ1. Do alternative thread management constructs have different impacts on energy consumption?
- 20. 20 RQ2. Do alternative data-sharing primitives have different impacts on energy consumption?
- 21. Study 2: Concurrency Primitives Thread management: forkIO, forkOn, forkOS 22
- 22. Study 2: Concurrency Primitives Data sharing: MVar, TVar, TMVar 23
- 23. 9 benchmarks: IO, memory, synchronization bound Up to 9 variants per benchmark 9 configurations for # of processors 24
- 25. 26 Small changes can produce big savings
- 26. Faster is not always greener 27
- 27. 28 There is no overall winner
- 28. 29
- 29. Made two tools energy-aware: Haskell profiler Criterion 31
- 31. fsklsdfnjdsfjsadfhjksdhflkjsdhflkjsah 33 Population: 56483 (2013) Haskell programmers: 0 (est.)
Notas do Editor
- Dispositivos móveis estão cada vez mais presentes baterias desses dispositivos durem o máximo possível Servidores são o coração da internet Devido a grande escala, pequenas diferenças em consumo de energia podem resultar em muita economia de dinheiro Gastar menos energia é tão importante que muitas empresas estão movendo seus data centers para regiões frias com intenção reduzir o consumo de energia com refrigeração.
- A large body of work in hardware/architecture, OS, runtime systems Thesedays, there’s a growing concern with tools/methods to analyse/improve energy consumption at the application level Developers are interested in saving energy 1 Pinto et al. “Mining Questions About Software Energy Consumption”. MSR'14.
- Functional programming (FP) is on the rise Mainstream languages are incorporating FP features Haskell influences a lot of functional PLs Lack of studies on energy efficiency
- Haskell está sendo utilizado tanto no contexto móvel quanto em servidores.
- Estas operações são *abstractas* cujas implementações concretas se conseguem com funções nas interfaces
- |base| = 100 000 |aux| = 1 000
- We can duplicate the number of configurations, since we ran experiments in more than one machine. At least two!
- Results of remove operation for Sequences
- Capabilities: virtual processors that the Haskell runtime system uses to manage parallel execution.
- forkIO - creates a new thread that is managed by the language’s scheduler forkOn - creates a new thread to be executed on a specific processor forkOS — creates a thread that is bound to a specific operating system thread. This is important for foreign functions that make use of thread-local state.
- MVar - em Haskell, é a primitiva básica para partilha de informação Pode ser entendida como uma caixa que ou está vazia ou está cheia. takeMVar/putMVar bloqueiam caso a caixa não esteja no estado esperado TVar é disponibilizada por uma implementação de software transactional memory TMVar é uma variante transacional do tipo MVar
- Reforçar a diversidade dos benchmarks TWO MACHINES! No special purpose for 999
- switching from forkIO to forkON can yell big savings, and switching from thread management constructs in Haskell is simple!
- Observar linha verde: dos melhores desempenhos em tempo de execução, dos piores em termos de consumo de energia
- forkOS often bad MVars often good
- The End