방법: 동적 파티션 구현
다음 예제에서는 동적 분할을 구현하고 ForEach의 특정 오버로드 및 PLINQ에서 사용할 수 있는 사용자 지정 System.Collections.Concurrent.OrderablePartitioner<TSource>을 구현하는 방법을 보여 줍니다.
예제
파티션에서 MoveNext를 열거자에 대해 호출할 때마다 열거자는 파티션에 목록 요소를 하나 제공합니다. PLINQ 및 ForEach의 경우 파티션은 Task 인스턴스입니다. 여러 스레드에서 요청이 동시에 발생하기 때문에 현재 인덱스에 대한 액세스는 동기화됩니다.
Imports System.Threading
Imports System.Threading.Tasks
Imports System.Collections.Concurrent
Module Module1
Public Class OrderableListPartitioner(Of TSource)
Inherits OrderablePartitioner(Of TSource)
Private ReadOnly m_input As IList(Of TSource)
Public Sub New(ByVal input As IList(Of TSource))
MyBase.New(True, False, True)
m_input = input
End Sub
' Must override to return true.
Public Overrides ReadOnly Property SupportsDynamicPartitions As Boolean
Get
Return True
End Get
End Property
Public Overrides Function GetOrderablePartitions(ByVal partitionCount As Integer) As IList(Of IEnumerator(Of KeyValuePair(Of Long, TSource)))
Dim dynamicPartitions = GetOrderableDynamicPartitions()
Dim partitions(partitionCount - 1) As IEnumerator(Of KeyValuePair(Of Long, TSource))
For i = 0 To partitionCount - 1
partitions(i) = dynamicPartitions.GetEnumerator()
Next
Return partitions
End Function
Public Overrides Function GetOrderableDynamicPartitions() As IEnumerable(Of KeyValuePair(Of Long, TSource))
Return New ListDynamicPartitions(m_input)
End Function
Private Class ListDynamicPartitions
Implements IEnumerable(Of KeyValuePair(Of Long, TSource))
Private m_input As IList(Of TSource)
Friend Sub New(ByVal input As IList(Of TSource))
m_input = input
End Sub
Public Function GetEnumerator() As IEnumerator(Of KeyValuePair(Of Long, TSource)) Implements IEnumerable(Of KeyValuePair(Of Long, TSource)).GetEnumerator
Return New ListDynamicPartitionsEnumerator(m_input)
End Function
Public Function GetEnumerator1() As IEnumerator Implements IEnumerable.GetEnumerator
Return CType(Me, IEnumerable).GetEnumerator()
End Function
End Class
Private Class ListDynamicPartitionsEnumerator
Implements IEnumerator(Of KeyValuePair(Of Long, TSource))
Private m_input As IList(Of TSource)
Shared m_pos As Integer = 0
Private m_current As KeyValuePair(Of Long, TSource)
Public Sub New(ByVal input As IList(Of TSource))
m_input = input
m_pos = 0
Me.disposedValue = False
End Sub
Public ReadOnly Property Current As KeyValuePair(Of Long, TSource) Implements IEnumerator(Of KeyValuePair(Of Long, TSource)).Current
Get
Return m_current
End Get
End Property
Public ReadOnly Property Current1 As Object Implements IEnumerator.Current
Get
Return Me.Current
End Get
End Property
Public Function MoveNext() As Boolean Implements IEnumerator.MoveNext
Dim elemIndex = Interlocked.Increment(m_pos) - 1
If elemIndex >= m_input.Count Then
Return False
End If
m_current = New KeyValuePair(Of Long, TSource)(elemIndex, m_input(elemIndex))
Return True
End Function
Public Sub Reset() Implements IEnumerator.Reset
m_pos = 0
End Sub
Private disposedValue As Boolean ' To detect redundant calls
Protected Overridable Sub Dispose(ByVal disposing As Boolean)
If Not Me.disposedValue Then
m_input = Nothing
m_current = Nothing
End If
Me.disposedValue = True
End Sub
Public Sub Dispose() Implements IDisposable.Dispose
Dispose(True)
GC.SuppressFinalize(Me)
End Sub
End Class
End Class
Class ConsumerClass
Shared Sub Main()
Console.BufferHeight = 20000
Dim nums = Enumerable.Range(0, 2000).ToArray()
Dim partitioner = New OrderableListPartitioner(Of Integer)(nums)
' Use with Parallel.ForEach
Parallel.ForEach(partitioner, Sub(i) Console.Write("{0}:{1} ", i, Thread.CurrentThread.ManagedThreadId))
Console.WriteLine("PLINQ -----------------------------------")
' create a new partitioner, since Enumerators are not reusable.
Dim partitioner2 = New OrderableListPartitioner(Of Integer)(nums)
' Use with PLINQ
Dim query = From num In partitioner2.AsParallel()
Where num Mod 8 = 0
Select num
For Each v In query
Console.Write("{0} ", v)
Next
Console.WriteLine("press any key")
Console.ReadKey()
End Sub
End Class
End Module
//
// An orderable dynamic partitioner for lists
//
class OrderableListPartitioner<TSource> : OrderablePartitioner<TSource>
{
private readonly IList<TSource> m_input;
public OrderableListPartitioner(IList<TSource> input)
: base(true, false, true)
{
m_input = input;
}
// Must override to return true.
public override bool SupportsDynamicPartitions
{
get
{
return true;
}
}
public override IList<IEnumerator<KeyValuePair<long, TSource>>>
GetOrderablePartitions(int partitionCount)
{
var dynamicPartitions = GetOrderableDynamicPartitions();
var partitions =
new IEnumerator<KeyValuePair<long, TSource>>[partitionCount];
for (int i = 0; i < partitionCount; i++)
{
partitions[i] = dynamicPartitions.GetEnumerator();
}
return partitions;
}
public override IEnumerable<KeyValuePair<long, TSource>>
GetOrderableDynamicPartitions()
{
return new ListDynamicPartitions(m_input);
}
private class ListDynamicPartitions
: IEnumerable<KeyValuePair<long, TSource>>
{
private IList<TSource> m_input;
private int m_pos = 0;
internal ListDynamicPartitions(IList<TSource> input)
{
m_input = input;
}
public IEnumerator<KeyValuePair<long, TSource>> GetEnumerator()
{
while (true)
{
// Each task gets the next item in the list. The index is
// incremented in a thread-safe manner to avoid races.
int elemIndex = Interlocked.Increment(ref m_pos) - 1;
if (elemIndex >= m_input.Count)
{
yield break;
}
yield return new KeyValuePair<long, TSource>(
elemIndex, m_input[elemIndex]);
}
}
IEnumerator IEnumerable.GetEnumerator()
{
return
((IEnumerable<KeyValuePair<long, TSource>>)this)
.GetEnumerator();
}
}
}
class ConsumerClass
{
static void Main()
{
var nums = Enumerable.Range(0, 10000).ToArray();
OrderableListPartitioner<int> partitioner = new OrderableListPartitioner<int>(nums);
// Use with Parallel.ForEach
Parallel.ForEach(partitioner, (i) => Console.WriteLine(i));
// Use with PLINQ
var query = from num in partitioner.AsParallel()
where num % 2 == 0
select num;
foreach (var v in query)
Console.WriteLine(v);
}
}
이는 각 청크가 하나의 요소로 구성되어 있는 청크 분할의 예입니다. 한 번에 보다 많은 요소를 제공하면 잠금에 대한 경합을 줄일 수 있고 이론적으로는 수행 속도를 향상시킬 수 있습니다. 그러나 모든 작업이 수행될 때까지 모든 스레드가 유휴 상태에 있지 않도록 하려면 특정 시점에는 큰 청크 때문에 추가적인 부하 분산 논리가 필요할 수 있습니다.