ENH: Add a bulk mask read to MaskCompare and use it for every mask read

Aperta
#1,895 0 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 6 giorni

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà
5/5
Tempo stimato
Più di una settimana
Idoneità per principianti
35/100
Tipo di issue
Funzionalità
Chiarezza
Abbastanza chiara
Stato di attività
Attiva
Stack tecnologico
cpp
Ambito
backend, data

Direzione di ricerca

Start with src/simplnx/Utilities/MaskCompareUtilities.hpp and .cpp to understand the existing mask interfaces, then inspect the listed OrientationAnalysis and SimplnxCore algorithms that bulk-read typed mask stores. Add a bulk read that reports transfer failures, route those algorithms through MaskCompare, and add the rule to docs/Code_Style_Guide.md. Done means in-core Release performance and out-of-core [ooc] tests meet the stated requirements.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

enhancement Out-of-Core

Summary

Request: give MaskCompare a bulk read of a range of mask values that works for bool and uint8 masks, on in-memory and out-of-core stores, without making in-core algorithms slower. Then change every algorithm that reads a mask to read it through MaskCompare, instead of reading the typed mask stores itself.

Why: today a mask can be read in two different ways, and both are copied around the code base.

  • MaskCompare (MaskCompareUtilities.hpp) handles bool and uint8 masks behind one interface, but it reads one element at a time through the virtual isTrue(index), bothTrue() and bothFalse(). On an out-of-core store, each element read can load a whole chunk, so algorithms that are optimized for out-of-core data do not use it for their main loops.
  • Those algorithms read the mask in bulk instead. Each one checks whether the mask is a UInt8Array or a BoolArray, calls copyIntoBuffer() on the matching store, and copies a bool mask through a contiguous bool scratch buffer, because std::vector<bool> is not contiguous. Each algorithm writes this branch itself.
  • A new algorithm then has to choose between a convenient helper that is slow out-of-core and a hand-written copy of the bulk pattern. Code that reads a mask only with getDataAs<BoolArray>() gets nullptr for a uint8 mask and silently ignores it.

Where it applies:

  • MaskCompare, BoolMaskCompare, UInt8MaskCompare and InstantiateMaskCompare() in src/simplnx/Utilities/MaskCompareUtilities.hpp and .cpp.
  • 27 files call InstantiateMaskCompare() (at 6dd86c67d).
  • These algorithms bulk-read the typed mask stores themselves. Several of them also use MaskCompare for element access, and some handle only one of the two mask types. Each one needs to be checked:
    • OrientationAnalysis: AlignSectionsMisorientation, AlignSectionsMutualInformation, BadDataNeighborOrientationCheckScanline, CAxisSegmentFeatures, ComputeFZQuaternions, EBSDSegmentFeatures, WriteStatsGenOdfAngleFile
    • SimplnxCore: AlignSectionsFeatureCentroid, ComputeArrayHistogram, ComputeArrayHistogramByFeature, ComputeArrayStatistics, ComputeFeaturePhasesBinary, CreateColorMapScanline, DBSCANScanline, ErodeDilateMask, ExtractVertexGeometry
  • Out-of-tree plugins, such as SimplnxReview, may contain the same pattern.

Requirements:

  • The bulk read returns a Result for a failed transfer.
  • In-core performance must not regress: compare the changed algorithms before and after in an in-core Release build.
  • Out-of-core behavior must not regress: run the [ooc] tests in the out-of-core build.
  • When MaskCompare has the bulk read, add a rule to docs/Code_Style_Guide.md that algorithms read masks through MaskCompare, in the same change.

The exact API (the signature of the bulk read, and whether it fills a uint8 or a bool buffer) is still open.

Lingua principale
C++
Stelle
18
Fork
15
Merge medio
4g 23h
PR unite (30g)
42

Preparare l'ambiente

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di BlueQuartzSoftware/simplnx

Tutte le issue di BlueQuartzSoftware/simplnx

Issue simili

Altre issue su C++

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.