Skip to content

Navigation Menu

Appearance settings

View all features
- BY COMPANY SIZE
  Enterprises
  Small and medium teams
  Startups
  Nonprofits
- BY USE CASE
  App Modernization
  DevSecOps
  DevOps
  CI/CD
  View all use cases
- BY INDUSTRY
  Healthcare
  Financial services
  Manufacturing
  Government
  View all industries
View all solutions
- EXPLORE BY TOPIC
  AI
  Software Development
  DevOps
  Security
  View all topics
- EXPLORE BY TYPE
  Customer stories
  Events & webinars
  Ebooks & reports
  Business insights
  GitHub Skills
- SUPPORT & SERVICES
  Documentation
  Customer support
  Community forum
  Trust center
  Partners
View all resources
- COMMUNITY
  GitHub SponsorsFund open source developers
- PROGRAMS
  Security Lab
  Maintainer Community
  Accelerator
  GitHub Stars
  Archive Program
- REPOSITORIES
  Topics
  Trending
  Collections
- ENTERPRISE SOLUTIONS
  Enterprise platformAI-powered developer platform
- AVAILABLE ADD-ONS
  GitHub Advanced SecurityEnterprise-grade security features
  Copilot for BusinessEnterprise-grade AI features
  Premium SupportEnterprise-grade 24/7 support
Pricing

Search code, repositories, users, issues, pull requests...

Search

Clear

Search syntax tips

Provide feedback

We read every piece of feedback, and take your input very seriously.

Include my email address so I can be contacted

Saved searches

Use saved searches to filter your results more quickly

Name

Query

To see all available qualifiers, see our documentation.

Appearance settings

You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session.

Dismiss alert

joshuaisaact / elastic-hash Public

Notifications You must be signed in to change notification settings
Fork 0
Star 2

Code
Issues
Pull requests 8
Actions
Projects
Security and quality
Insights

Additional navigation options

Code
Issues
Pull requests
Actions
Projects
Security and quality
Insights

Optimize lookup: 5x faster than std.HashMap at 1M elements#1

Open

joshuaisaact wants to merge 85 commits intomainjoshuaisaact/elastic-hash:mainfrom

autoresearch/lookup-optimizationjoshuaisaact/elastic-hash:autoresearch/lookup-optimizationCopy head branch name to clipboard

Conversation Commits85 (85)Checks Files changed

Open

Optimize lookup: 5x faster than std.HashMap at 1M elements#1
joshuaisaact wants to merge 85 commits intomainfrom
autoresearch/lookup-optimization

Commits

Commits on Mar 23, 2026

tier-first search order in get() for better cache locality
joshuaisaact
committed
Revert "tier-first search order in get() for better cache locality"
joshuaisaact
committed
cross-tier prefetching in get() to hide inter-tier latency
joshuaisaact
committed
Revert "cross-tier prefetching in get() to hide inter-tier latency"
joshuaisaact
committed
remove all prefetching from get() to test if it helps or hurts
joshuaisaact
committed
add tier-0 probe-0 fast path in get()
joshuaisaact
committed
Revert "add tier-0 probe-0 fast path in get()"
joshuaisaact
committed
reduce MAX_PROBES from 32 to 24
joshuaisaact
committed
reduce MAX_PROBES from 24 to 20
joshuaisaact
committed
cap get() tier search to 8 tiers
joshuaisaact
committed
reduce MAX_LOOKUP_TIERS from 8 to 6
joshuaisaact
committed
Revert "reduce MAX_LOOKUP_TIERS from 8 to 6"
joshuaisaact
committed
use fixed-size arrays for tier metadata instead of heap slices
joshuaisaact
committed
Revert "use fixed-size arrays for tier metadata instead of heap slices"
joshuaisaact
committed
switch hash to Stafford variant 13 (splitmix64 finalizer)
joshuaisaact
committed
Revert "switch hash to Stafford variant 13 (splitmix64 finalizer)"
joshuaisaact
committed
switch to linear probing for better cache locality
joshuaisaact
committed
add early exit on empty bucket slots in get()
joshuaisaact
committed
Revert "add early exit on empty bucket slots in get()"
joshuaisaact
committed
combined fp+empty SIMD check with per-tier early exit
joshuaisaact
committed
Revert "combined fp+empty SIMD check with per-tier early exit"
joshuaisaact
committed
double tier 0 size to concentrate elements for faster lookup
joshuaisaact
committed
quadruple tier 0 size (2x capacity / BUCKET_SIZE)
joshuaisaact
committed
Revert "quadruple tier 0 size (2x capacity / BUCKET_SIZE)"
joshuaisaact
committed
reduce MAX_LOOKUP_TIERS from 8 to 4 with larger tier 0
joshuaisaact
committed
reduce MAX_LOOKUP_TIERS from 4 to 3
joshuaisaact
committed
reduce MAX_LOOKUP_TIERS from 3 to 2
joshuaisaact
committed
reduce MAX_LOOKUP_TIERS to 1 (tier 0 only)
joshuaisaact
committed
simplify get() to single tier 0 loop, no tier iteration
joshuaisaact
committed
Revert "simplify get() to single tier 0 loop, no tier iteration"
joshuaisaact
committed
use inline for to unroll probe loop in get()
joshuaisaact
committed
Revert "use inline for to unroll probe loop in get()"
joshuaisaact
committed
reduce MAX_PROBES from 20 to 10
joshuaisaact
committed
reduce MAX_PROBES from 10 to 8
joshuaisaact
committed
reduce MAX_PROBES from 8 to 6
joshuaisaact
committed
Revert "reduce MAX_PROBES from 8 to 6"
joshuaisaact
committed
increase BUCKET_SIZE to 32 for AVX2, generic mask types
joshuaisaact
committed
Revert "increase BUCKET_SIZE to 32 for AVX2, generic mask types"
joshuaisaact
committed
try quadratic probing instead of linear
joshuaisaact
committed
Revert "try quadratic probing instead of linear"
joshuaisaact
committed
inline for unroll 8-probe loop and eliminate tier loop
joshuaisaact
committed
Revert "inline for unroll 8-probe loop and eliminate tier loop"
joshuaisaact
committed
simplify get() to tier-0-only with comptime MAX_PROBES loop bound
joshuaisaact
committed
remove insert prefetch
joshuaisaact
committed
delay batch transition to 90% fill (was 75%) to keep more in tier 0
joshuaisaact
committed
limit insertIntoTier probe depth to MAX_PROBES for lookup alignment
joshuaisaact
committed
try 4x tier 0 capacity with probe-limited insert
joshuaisaact
committed
Revert "try 4x tier 0 capacity with probe-limited insert"
joshuaisaact
committed
tighten batch threshold to 0.08 (92% fill)
joshuaisaact
committed
optimize remove() to tier-0-only with comptime loop bound
joshuaisaact
committed
try faster 64-bit multiply hash
joshuaisaact
committed
try single-multiply hash
joshuaisaact
committed
noinline findKeyInBucket to keep hot path tight
joshuaisaact
committed
Revert "noinline findKeyInBucket to keep hot path tight"
joshuaisaact
committed
fast path for single FP match in findKeyInBucket
joshuaisaact
committed
cache tier0 metadata in struct fields for faster get/remove
joshuaisaact
committed
hardcode tier0_start=0, eliminate addition in get/remove
joshuaisaact
committed
remove unused tier0_start field
joshuaisaact
committed
try MAX_PROBES=6 again with all other optimizations
joshuaisaact
committed
Revert "try MAX_PROBES=6 again with all other optimizations"
joshuaisaact
committed
cache bucket mask directly, eliminate subtraction per probe
joshuaisaact
committed
prefetch keys for probe 0 to hide key read latency
joshuaisaact
committed
Revert "prefetch keys for probe 0 to hide key read latency"
joshuaisaact
committed
branchless fingerprint using 7 bits + 1 (range 1-128)
joshuaisaact
committed
Revert "branchless fingerprint using 7 bits + 1 (range 1-128)"
joshuaisaact
committed
reorder struct fields: hot path first for cache line alignment
joshuaisaact
committed
try inline for with precomputed probe array
joshuaisaact
committed
Revert "try inline for with precomputed probe array"
joshuaisaact
committed
use bits 32-39 for fingerprint, less correlation with bucket index
joshuaisaact
committed
try bits 24-31 for fingerprint
joshuaisaact
committed
ultra-fast path: check slot 0 with validity check
joshuaisaact
committed
separate probe 0 check to avoid redundant work in main loop
joshuaisaact
committed
unroll probes 0 and 1 before the loop
joshuaisaact
committed
Revert "unroll probes 0 and 1 before the loop"
joshuaisaact
committed
prefetch values for probe 0 while SIMD runs
joshuaisaact
committed
Revert "prefetch values for probe 0 while SIMD runs"
joshuaisaact
committed
branchless fingerprint using max/min clamp
joshuaisaact
committed
Revert "branchless fingerprint using max/min clamp"
joshuaisaact
committed
batch threshold 0.06 (94% fill)
joshuaisaact
committed
Revert "batch threshold 0.06 (94% fill)"
joshuaisaact
committed
batch threshold 0.12 (88% fill)
joshuaisaact
committed
batch threshold 0.15 (85% fill)
joshuaisaact
committed
Revert "batch threshold 0.15 (85% fill)"
joshuaisaact
committed
add experiment results, benchmark harness, and autoresearch artifacts
joshuaisaact
committed
add rigorous abseil comparison, correct stale README claims
joshuaisaact
committed

Footer

© 2026 GitHub, Inc.

Footer navigation

Terms
Privacy
Security
Status
Community
Docs
Contact

You can’t perform that action at this time.