By Jacob SeligmannSteffen Grarup Presented By Leon Gendler ([email protected]) Incremental...

41
By Jacob Seligmann Steffen Grarup Presented By Leon Gendler ([email protected]) Incremental Mature Incremental Mature Garbage Collection Garbage Collection Using the Train Using the Train Algorithm Algorithm

Transcript of By Jacob SeligmannSteffen Grarup Presented By Leon Gendler ([email protected]) Incremental...

Page 1: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

By

Jacob Seligmann Steffen GrarupPresented By

Leon Gendler ([email protected])

Incremental Mature Garbage Incremental Mature Garbage Collection Using the Train Collection Using the Train

AlgorithmAlgorithm

Page 2: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Incremental Mature Garbage Incremental Mature Garbage CollectionCollection

By

Jacob Seligmann Steffen Grarup

Incremental Collection of Mature Incremental Collection of Mature ObjectsObjects

By

Richard L. Hudson J. Eliot B. Moss

Page 3: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

ContentsContents

Definitions and Traditional techniques Train Algorithm - Basic Ideas Example Technical and Implementation issues Measurement results

Page 4: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

DefinitionsDefinitionsGarbage - Space occupied by data objects which

can not be accessed.

Root Set - Global variables, local variables in the stack and registers used by active procedures.

Live Objects - The set of objects on any directed path of pointers from any member of the “root set”.

Page 5: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Garbage Collection TechniquesGarbage Collection Techniques

Traditional Techniques: Reference Counting

Mark (Sweep, Compact)

Copy

Page 6: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

ProblemsProblems

Undetected Cycles (Ref. Count)

Fragmentation (M-S) Cost proportional to heap size (M-S) Locality of references (M-C) Extensive Paging

Page 7: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Garbage Collection TechniquesGarbage Collection Techniques

IncrementalSmall units of G.C. between small units of

program execution.– Real-Time (Non-Disruptive).– Synchronizing the mutator with the collector

Generational“Most objects live a very short time, while a

small percentage of them live much longer”

Page 8: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Train Algorithm - DefinitionTrain Algorithm - Definition

IncrementalIncremental garbage collection scheme for

achieving non-disruptivenon-disruptive reclamation of the

oldest generationaloldest generational area.

Page 9: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Train Algorithm - MethodTrain Algorithm - Method

Dividing mature object spacemature object space into a number of fixed-sized blocks and collecting collecting one block at each invocationone block at each invocation.

Page 10: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Key ContributionKey Contribution

Recognizing and reclaiming All GarbageAll Garbage while only processing a single block at a time.

Non-Disruptive collection– Upper bound on the pause time. (fixed number

of objects in one block)

Page 11: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Train MetaphorTrain Metaphor

Memory Block Car

Set of Blocks Train

Page 12: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Train Metaphor Train Metaphor (cont.)(cont.)

The trains are ordered by giving them sequence numbers as they are created.

Order : One block precedes another if it belongs to a lower (older) train or has a lower car numbering in the same train.

Page 13: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.
Page 14: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

IntuitionIntuition

Constantly clustering sets of related objects

Eventually, any linked garbage structure collapses into the same train, no matter how complex.

Page 15: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

The VisionThe Vision

Page 16: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Car Collection StrategyCar Collection Strategy

Process the lowest numbered car:

1.Check references into the car’s trainif non exist, reclaim the whole train

2.Move objects referenced from outside M.O.AM.O.A. (Mature Object Area) to another (older) train.

Page 17: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Car Collection Strategy Car Collection Strategy (cont.)(cont.)

3.Move objects to the train that references them.

4.Scan the evacuated objects for pointers to the car being collected and repeat step 3.

Page 18: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Car Collection Strategy Car Collection Strategy (cont.)(cont.)

5.Objects referenced form further cars in the same train are moved to the last car.

6.Evacuate the car.

None of the objects remaining in the car are referenced form outside.

Page 19: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

ExampleExample

Page 20: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.
Page 21: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.
Page 22: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.
Page 23: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.
Page 24: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.
Page 25: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.
Page 26: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

CorrectnessCorrectness

“Garbage Trains” - The set of trains holding a garbage structure.

As each “Garbage Train” is processed, either garbage reclaimed or moved to a higher train.

When the highest “Garbage Train” is reached, the garbage structure will be entirely in the train.

Page 27: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Fixing an ErrorFixing an Error

RequirementRequirement: All trains are eventually processed.

Page 28: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Technical IssuesTechnical Issues

Remembered set of references from outside the car pointing into the car.

Optimize by recording references only from higher numbered cars to lower numbered car.

Page 29: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Technical Issues Technical Issues (cont.)(cont.)

““Write Barrier” - Write Barrier” - Pointer assignment run-time check.

Record pointer assignment from older to younger generations and between mature objects to other mature objects.

Train Table

Page 30: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Popular objectsPopular objects

Large remembered set. Moving is expensive.

Sol. 1: Indirect addressing– Run-time overhead.

Sol. 2: Do not collect car, move to the end of the newest train.– Leaves garbage in or referenced from those cars.– Scatters garbage across several trains.

Page 31: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

The ImplementationThe Implementation

The Original Mjolner BETA system:The Original Mjolner BETA system: Two generations:

– I.O.A.I.O.A. reclaimed using Copy collector– M.O.A.M.O.A. reclaimed using Mark-Sweep– Arrays of pointer-less objects kept separately.

Single Remembered Set (Hash Table) for references from old to young objects

Write Barrier

Page 32: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

New BETA SystemNew BETA System

Reclamation using the Train Algorithm. 64Kb car size. A Train Table. Remembered set for each car:

– References to young generation– References from higher cars

Page 33: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Evacuation StrategyEvacuation Strategy

Objects referenced from several trains– Last car of the first encountered train– Last car of newest train (if referenced from

outside M.O.A.M.O.A.) Promoted objects

– Last car of the newest train– Create new train if “fill limit” exceeded

Page 34: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Invocation FrequencyInvocation Frequency

Incremental– must be called often enough so that storage

resources are never exhausted. Garbage Ratio

– Memory size before and after a collection– Adjust collection frequency accordingly

Page 35: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Invocation Frequency Invocation Frequency (cont.)(cont.)

Difficult to calculate– parts of memory are not yet processed– only small area collected at each invocation

An object counter for each train.

Page 36: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

The ResultsThe Results

Collection pauses

Time Overhead– Young Generation Collector– Old Generation Collector

Storage Overhead

Page 37: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Collection PausesCollection Pauses

Page 38: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Time OverheadTime Overhead

Old Generation Collection– An increase of 20% in collection time.

Young Generation Collection– Roots are found in several remembered sets.– Object Promotion caused many set insertions

Overall Results:Overall Results:Total execution time increased by 1%.Total execution time increased by 1%.

Page 39: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Storage OverheadStorage Overhead

Car remembered set Overhead– 6-8K per car– editor: 350 entries, compiler: 700 entries

Car size - 64K Train Table - 256K (64K entries)

– One word for the train number – One for the car number

Page 40: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

Storage Overhead Storage Overhead (cont.)(cont.)

Overall Result:Overall Result:

Old generation storage overhead of 10% - 20%

Application storage overhead of 4% - 8%

Page 41: By Jacob SeligmannSteffen Grarup Presented By Leon Gendler (lgendler@math.tau.ac.il) Incremental Mature Garbage Collection Using the Train Algorithm.

ConclusionsConclusions

Non Disruptive M.O.A. collection Minimizes collection delays to few msc. A very low maximum delay A negligible increase in run time A small increase in storage requirements