IBM

1
IBMs X10

• Presentation by Isaac Dooley
• CS498LVK
• Spring 2006

2
Sources

• Report on the Experimental Language X10, Vijay
  Saraswat
• http//domino.research.ibm.com/comm/research_proje
  cts.nsf/pages/x10.index.html
• X10 Tutorial v10.pdf

3
Goal of X10

• The fundamental goal of X10 is to enable
  scalable, high-performance, high-productivity
  transformational programming for high-end
  computers -- for traditional numerical
  computation workloads as well as commercial
  server workloads.

4
Description

• X10 is a type-safe, modern, parallel,
  distributed object-oriented language intended to
  be very easily accessible to Java(TM)
  programmers. It is targeted to future low-end and
  high-end systems with nodes that are built out of
  multi-core SMP chips with non-uniform memory
  hierarchies, and interconnected in scalable
  cluster configurations.

5
Description

• A member of the Partitioned Global Address
  Space (PGAS) family of languages, X10 highlights
  the explicit reification of locality in the form
  of places lightweight activities embodied in
  async, future, foreach, and ateach constructs
  constructs for termination detection (finish) and
  phased computation (clocks) the use of lock-free
  synchronization (atomic blocks) and the
  manipulation of global arrays and data
  structures.

6
Memory Model

• Not Distributed Memory, Not Shared Memory
• Fragmented Memory Model
• Partitioned Global Address Space(GAS)
• Like UPC, Co-Array Fortran, Titanium
• Globally Asynchronous, Locally Synchronous (GALS)

7
Language

• Extended Subset of Java
• Without Java Concurrency(threads, monitors)
• With Distributed Arrays
• With new built-in types
• With Places, Activites, Clocks

8
Language

• Places
• Host some data
• Runs some activities(lightweight threads)
• Must spawn activities to access data at other
  places
• Correspond to an Address Space (SMP node)
• Immutable data
• Freely copied between places on access

9
Language

• Clocks
• Used to order activities
• Distributed Arrays
• Support Collectives

10
Language

• Atomic Sections
• atomic S
• Excecute locally, only accessing local data
• Asynchronous Activities
• async (P) S
• future (P) E

11
Regions

• Distributions are maps from a region to a subset
  of places
• A region 0200,1100 specifies a collection of
  2-D points
• Points are used as array indices
• Operations on regions are provided
• Union, disjunction, set difference

12
Deadlock, Data Races

• X10 guarantee -Any program written with async,
  finish, atomic, foreach, ateach, and clock
  parallel constructs will never deadlock
• Unrestricted use of future and force may lead to
  deadlock, but Restricted use of future and force
  in X10 can preserve guaranteed freedom from
  deadlocks
• To eliminate Data Races, atomic methods and
  blocks should be used

13
Example of future

• public class TutFuture1
• static int fib(final int n)
• if ( n lt 0 ) return 0
• else if ( n 1 ) return 1
• else
• futureltintgt fn_1 future fib(n-1)
  futureltintgt fn_2 future fib(n-2)
  
• return fn_1.force() fn_2.force()
• // fib()
• public static void main(String args)
  
• System.out.println("fib(10) " fib(10))
  
• Example of recursive divide-and- conquer
  parallelism --- calls to fib(n-1) and fib(n-2)
  execute in parallel

14
Example of async

• final int n100
• finish
• async for (int i1 iltn i2 ) oddSum.val
  i
• for (int j2 jltn j2 ) evenSum.val j
  
• // Both oddSum and evenSum have been
  computed now
• System.out.println("oddSum " oddSum.val "
  evenSum " evenSum.val)
• Parent activity creates new child to execute for
  (int i1 iltn i2 ) oddSum.val i
• An async statement returns immediately
• Parent execution proceeds immediately to next
  statement
• Any access to parents local data must be through
  final variables

15
Example of atomic

• finish
• async for (int i1 iltn i2 )
  
• double r 1.0d / i atomic rSum r
• for (int j2 jltn j2 )
• double r 1.0d / j atomic rSum r
  
• An atomic statement/method is conceptually
  executed in a single step, while other activities
  are suspended - Note programmer does not manage
  any locks explicitly
• An atomic section may not include - Blocking
  operations - Creation of activities

16
Example of atomic

• public class TutAtomic2
• const int a new boxedInt(100)
• const int b new boxedInt(100)
• public static atomic void incr_a() a.val
  b.val-- public static atomic void
  decr_a() a.val-- b.val
• public static void main(String args)
• int sum
• finish
• async for (int i1 ilt10 i ) incr_a()
  
• for (int i1 ilt10 i ) decr_a()
  
• atomic sum a.val b.val
• System.out.println("ab " sum)
• Console output ab 200

17
Code for Jacobi 2-D

• public class Jacobi
• const region R 0N1, 0N1
• const region RInner 1N, 1N
• const distribution D distribution.factory.block(
  R)
• const distribution DInner D RInner
• const distribution DBoundary D - RInner
• doubleD B new doubleD (point pi,j)
• return DBoundary.contains(p) ? (N-1)/2
  N(i-1)(j-1) // Exploded Variable
  Declaration, implicitly defined for i,j
• public boolean run()
• int iters 0
• double err
• while(true)
• double. Temp new doubleDInner (point
  i,j)
• return (read(i1,j)read(i-1,j)read(i,j1)read
  (i,j-1))/4.0
• if((err((B DInner) - Temp).abs().sum()) lt
  epsilon) break
• B.update(Temp)
• iters