All 18 Memory & Garbage Collection questions
More in Memory & Garbage Collection
What causes managed memory leaks despite the GC? How do Span<T> and ArrayPool reduce GC pressure? How does the difference between the Server GC and Workstation GC modes affect a .NET application's throughput and latency characteristics? How would you use a memory profiler (or dotnet-counters / dotnet-gcdump) to diagnose excessive garbage collection pressure in a production .NET application? How do WeakReference<T> objects work, and in what scenario would you use one instead of a normal strong reference? How would you use ArrayPool<T> to reduce garbage collection pressure in a high-throughput, buffer-allocating hot path?
Advertisement
C# › Memory & Garbage Collection

What is the Large Object Heap (LOH), and why do objects allocated there behave differently from regular small objects?

Advanced

Answer

Objects 85,000 bytes or larger are allocated on a SEPARATE Large Object Heap, which is NOT compacted by default during garbage collection (unlike the regular generational heap) — this can lead to HEAP FRAGMENTATION over time as large objects are freed and re-allocated, since the LOH's free space between surviving large objects can't always be reused efficiently.
byte[] largeArray = new byte[100_000]; // allocated on the LOH, not the regular generational heap
// Repeatedly allocating/freeing large arrays like this can fragment the LOH over time

// Mitigation: enable LOH compaction explicitly when needed
GCSettings.LargeObjectHeapCompactionMode = GCLargeObjectHeapCompactionMode.CompactOnce;
Real-world example Diagnosing a long-running service's gradually increasing memory footprint traced to LOH fragmentation from repeated large buffer allocations.

Common follow-ups: How does using ArrayPool<T> help avoid the repeated large allocations that cause this LOH fragmentation problem in the first place?

Arrays Span<T> & Memory<T>
All 18 Memory & Garbage Collection questions