[SPARK-37593] Reduce default page size by LONG_ARRAY_OFFSET if G1GC and ON_HEAP are used

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 3.3.0
• Fix Version/s: 3.3.0
• Component/s: Spark Core, SQL
• Target Version/s: 3.3.0
• Assignee: EdisonWang
• Reporter: EdisonWang
• Created: 09/Dec/21 09:18
• Updated: 01/Mar/22 20:11
• Resolved: 01/Mar/22 20:11

Description

Spark's tungsten memory model usually tries to allocate memory by one `page` each time and allocated by long[pageSizeBytes/8] in HeapMemoryAllocator.allocate. 

Remember that java long array needs extra object header (usually 16 bytes in 64bit system), so the really bytes allocated is pageSize+16.

Assume that the G1HeapRegionSize is 4M and pageSizeBytes is 4M as well. Since every time we need to allocate 4M+16byte memory, so two regions are used with one region only occupies 16byte. Then there are about 50% memory waste.
It can happenes under different combinations of G1HeapRegionSize (varies from 1M to 32M) and pageSizeBytes (varies from 1M to 64M).

Issue Links

• links to: [Github] Pull Request #34846 (WangGuangxin)
• links to: [Github] Pull Request #34846 (WangGuangxin)