mirror of
https://github.com/logos-messaging/nim-sds.git
synced 2026-01-02 14:13:07 +00:00
add benchmarking results, custom murmur3_32
This commit is contained in:
shash256
parent
8006af303c
commit
cd7095aa9c
2
nim-bloom/.gitignore
vendored
2
nim-bloom/.gitignore
vendored
@ -1,6 +1,8 @@
|
||||
nimcache
|
||||
nimcache/*
|
||||
tests/test
|
||||
benches/bench
|
||||
benches/bench_arch_end
|
||||
bloom
|
||||
*.html
|
||||
*.css
|
||||
|
||||
@ -1,91 +1,117 @@
|
||||
# nim-bloom
|
||||
|
||||
A high-performance Bloom filter implementation in Nim. Supports both Nim's built-in MurmurHash2 (default) and an optional 128-bit MurmurHash3 implementation for large-scale use cases.
|
||||
A high-performance Bloom filter implementation in Nim offering standard and custom hash function options with different performance characteristics and false positive rates.
|
||||
|
||||
## Features
|
||||
|
||||
- Fast string element insertion and lookup
|
||||
- Configurable error rates
|
||||
- Choice between standard Nim hash (MurmurHash2) and extended 128-bit MurmurHash3
|
||||
- Optimized for both small and large-scale use cases
|
||||
- Comprehensive test suite
|
||||
- Choice between standard Nim hash and custom MurmurHash3 (128-bit or 32-bit)
|
||||
- Optimized for supporting different use cases of speed and accuracy
|
||||
- Comprehensive test suite and benchmarks
|
||||
|
||||
## Performance
|
||||
## Usage
|
||||
|
||||
Historical benchmark using MurmurHash3 implementation on a 10-year-old Macbook Pro Retina:
|
||||
- ~2.5M insertions/sec (~4.0 seconds for 10M insertions)
|
||||
- ~2.9M lookups/sec (~3.5 seconds for 10M lookups)
|
||||
- Test configuration: 0.001 error rate, Bloom filter size ~20-25MB
|
||||
- Compiled with `-d:release` flag
|
||||
|
||||
These numbers reflect performance outside of CPU cache, as the filter size was intentionally larger than L3 cache. Performance can be several million operations/sec higher with smaller filters that fit in cache.
|
||||
|
||||
Current performance will vary based on:
|
||||
- Choice of hash function (standard Nim hash vs extended MurmurHash3)
|
||||
- Hardware specifications
|
||||
- Data size and memory access patterns
|
||||
- Compiler optimizations
|
||||
|
||||
The default configuration (using Nim's built-in hash) is optimized for typical use cases, while the extended hash option (MurmurHash3) provides better collision resistance for large-scale applications at a slight performance cost.
|
||||
|
||||
## Quickstart
|
||||
|
||||
Basic usage:
|
||||
Basic usage (defaults to MurmurHash3_128):
|
||||
```nim
|
||||
import bloom
|
||||
import bloom2
|
||||
|
||||
# Initialize with default hash (suitable for most uses)
|
||||
var bf = initializeBloomFilter(capacity = 10000, errorRate = 0.001)
|
||||
echo bf # Print Bloom filter characteristics
|
||||
echo bf.lookup("test") # false
|
||||
# Initialize with default hash (MurmurHash3_128)
|
||||
var bf = initializeBloomFilter(capacity = 10000, errorRate = 0.01)
|
||||
|
||||
# Or explicitly specify hash type
|
||||
var bf32 = initializeBloomFilter(
|
||||
capacity = 10000,
|
||||
errorRate = 0.01,
|
||||
hashType = htMurmur32 # Use 32-bit implementation
|
||||
)
|
||||
|
||||
# Basic operations
|
||||
bf.insert("test")
|
||||
assert bf.lookup("test") # true
|
||||
|
||||
# For large-scale usage (>1M elements), consider using extended hash
|
||||
var largeBf = initializeBloomFilter(
|
||||
capacity = 2_000_000,
|
||||
errorRate = 0.001,
|
||||
useExtendedHash = true
|
||||
)
|
||||
assert bf.lookup("test")
|
||||
```
|
||||
|
||||
## Advanced Configuration
|
||||
|
||||
The Bloom filter can be configured in several ways:
|
||||
|
||||
1. Default initialization (automatically calculates optimal parameters):
|
||||
```nim
|
||||
var bf = initializeBloomFilter(capacity = 10000, errorRate = 0.001)
|
||||
```
|
||||
|
||||
2. Specify custom number of hash functions:
|
||||
```nim
|
||||
var bf = initializeBloomFilter(
|
||||
capacity = 10000,
|
||||
errorRate = 0.001,
|
||||
k = 5 # Use 5 hash functions instead of calculated optimal
|
||||
)
|
||||
```
|
||||
|
||||
3. Fully manual configuration:
|
||||
```nim
|
||||
var bf = initializeBloomFilter(
|
||||
capacity = 10000,
|
||||
errorRate = 0.001,
|
||||
k = 5,
|
||||
forceNBitsPerElem = 12,
|
||||
useExtendedHash = false # Use standard hash (default)
|
||||
)
|
||||
```
|
||||
|
||||
Note: When specifying `k`, it must be ≤ 12 unless `forceNBitsPerElem` is also specified. The implementation will raise a `BloomFilterError` if parameters would result in suboptimal performance.
|
||||
|
||||
## Hash Function Selection
|
||||
|
||||
- Default: Uses Nim's built-in hash (MurmurHash2), suitable for most use cases
|
||||
- Extended: Uses 128-bit MurmurHash3, better for large sets (>1M elements) where collision resistance is critical
|
||||
1. Use MurmurHash3_128 (default) when:
|
||||
- You need the best balance of performance and accuracy
|
||||
- Memory isn't severely constrained
|
||||
- Working with large datasets
|
||||
- False positive rates are important
|
||||
|
||||
Choose extended hash by setting `useExtendedHash = true` during initialization.
|
||||
2. Use MurmurHash3_32 when:
|
||||
- Running on 32-bit systems
|
||||
- Memory is constrained
|
||||
- Working with smaller datasets
|
||||
- String concatenation overhead for second hash, causing higher insertion and lookup times, is acceptable.
|
||||
|
||||
3. Use NimHash when:
|
||||
- Consistency with Nim's default hashing is important
|
||||
- Working with smaller datasets where performance is less critical
|
||||
- Future availability of better hash functions or performant implementations
|
||||
|
||||
Nim's Hash Implementation:
|
||||
- Default (no flags): Uses FarmHash implementation
|
||||
- With `-d:nimStringHash2`: Uses Nim's MurmurHash3_32 implementation
|
||||
- Our implementation allows explicit choice regardless of compilation flags and our MurmurHash3_32 performs better because of directly using a native C Implementation
|
||||
|
||||
## Performance Characteristics
|
||||
### For 1M items - Random Strings
|
||||
```
|
||||
Insertion Speed:
|
||||
MurmurHash3_128: ~6.8M ops/sec
|
||||
MurmurHash3_32: ~5.9M ops/sec
|
||||
FarmHash: ~2.1M ops/sec
|
||||
|
||||
False Positive Rates:
|
||||
MurmurHash3_128: ~0.84%
|
||||
MurmurHash3_32: ~0.83%
|
||||
FarmHash: ~0.82%
|
||||
```
|
||||
|
||||
These measurements show MurmurHash3_128's balanced performance profile, offering best speed and competitive false positive rates.
|
||||
|
||||
Performance will vary based on:
|
||||
- Choice of hash function
|
||||
- Hardware specifications
|
||||
- Data size and memory access patterns (inside vs outside cache)
|
||||
- Compiler optimizations
|
||||
|
||||
For detailed benchmarks across different data patterns and sizes, see [benches](benches/).
|
||||
|
||||
## Implementation Details
|
||||
|
||||
### Double Hashing Technique
|
||||
This implmentation uses the Kirsch-Mitzenmacher method to generate k hash values from two initial hashes. The implementation varies by hash type:
|
||||
|
||||
1. MurmurHash3_128:
|
||||
```nim
|
||||
h(i) = abs((hash1 + i * hash2) mod m)
|
||||
```
|
||||
- Uses both 64-bit hashes from 128-bit output
|
||||
- Natural double-hash implementation
|
||||
|
||||
2. MurmurHash3_32:
|
||||
```nim
|
||||
let baseHash = murmurHash32(item, 0'u32)
|
||||
let secondHash = murmurHash32(item & " b", 0'u32)
|
||||
```
|
||||
- Uses string concatention by default for the second hash
|
||||
- Bit Rotation for second hash provides sufficient randomness in some use cases while being much faster than string concatenation (but results in higher FP rate)
|
||||
- Choose between bit rotation or string concatenation as per your use-case.
|
||||
|
||||
3. Nim's Default Hash:
|
||||
```nim
|
||||
let
|
||||
hashA = abs(hash(item)) mod maxValue
|
||||
hashB = abs(hash(item & " b")) mod maxValue
|
||||
h(i) = abs((hashA + n * hashB)) mod maxValue
|
||||
```
|
||||
- Farm Hash or Nim's Murmur Hash based (if compliation flag is passed)
|
||||
- Uses string concatention by default.
|
||||
- Lower FP rate than bit rotation but comes at the cost of higher insertion and lookup times.
|
||||
|
||||
*Tip:* Bit rotation values can be configurable as well. Use prime numbers for better mixing: 7, 11, 13, 17 for 32-bit; 21, 23, 27, 33 for 64-bit. Smaller rotations provides lesser mixing but as faster than higher rotations.
|
||||
|
||||
## Testing
|
||||
|
||||
|
||||
123
nim-bloom/benches/bench.nim
Normal file
123
nim-bloom/benches/bench.nim
Normal file
@ -0,0 +1,123 @@
|
||||
import times, random, strutils
|
||||
include bloom
|
||||
|
||||
type
|
||||
DataPattern = enum
|
||||
dpRandom, # Random strings
|
||||
dpSequential, # Sequential numbers
|
||||
dpFixed, # Fixed length strings
|
||||
dpLong, # Long strings
|
||||
dpSpecial # Strings with special characters
|
||||
|
||||
type
|
||||
BenchmarkResult = tuple[
|
||||
insertTime: float,
|
||||
lookupTime: float,
|
||||
falsePositives: int
|
||||
]
|
||||
|
||||
proc generateBenchData(pattern: DataPattern, size: int, isLookupData: bool = false): seq[string] =
|
||||
result = newSeq[string](size)
|
||||
let offset = if isLookupData: size * 2 else: 0 # Ensure lookup data is well separated
|
||||
|
||||
case pattern:
|
||||
of dpRandom:
|
||||
for i in 0..<size:
|
||||
var s = ""
|
||||
for j in 0..rand(5..15):
|
||||
s.add(chr(rand(ord('a')..ord('z'))))
|
||||
result[i] = s
|
||||
of dpSequential:
|
||||
for i in 0..<size:
|
||||
result[i] = $(i + offset) # Add offset for lookup data
|
||||
of dpFixed:
|
||||
for i in 0..<size:
|
||||
result[i] = "fixed" & align($(i + offset), 10, '0')
|
||||
of dpLong:
|
||||
for i in 0..<size:
|
||||
result[i] = repeat("x", 100) & $(i + offset)
|
||||
of dpSpecial:
|
||||
for i in 0..<size:
|
||||
result[i] = "test@" & $(i + offset) & "#$%^&*" & $rand(1000)
|
||||
|
||||
proc benchmarkHashType(hashType: HashType, size: int, errorRate: float,
|
||||
data: seq[string], lookupData: seq[string]): BenchmarkResult =
|
||||
# Initialize Bloom filter and run benchmark for given hash type
|
||||
var bf = initializeBloomFilter(size, errorRate, hashType = hashType)
|
||||
|
||||
# Measure insert time
|
||||
let startInsert = cpuTime()
|
||||
for item in data:
|
||||
bf.insert(item)
|
||||
let insertTime = cpuTime() - startInsert
|
||||
|
||||
# Measure lookup time and count false positives
|
||||
var falsePositives = 0
|
||||
let startLookup = cpuTime()
|
||||
for item in lookupData:
|
||||
if bf.lookup(item): falsePositives.inc
|
||||
let lookupTime = cpuTime() - startLookup
|
||||
|
||||
result = (insertTime, lookupTime, falsePositives)
|
||||
|
||||
proc printResults(hashName: string, result: BenchmarkResult,
|
||||
dataSize: int, lookupDataSize: int) =
|
||||
echo "\n", hashName, " Results:"
|
||||
echo " Insert time: ", result.insertTime, "s (", dataSize.float/result.insertTime, " ops/sec)"
|
||||
echo " Lookup time: ", result.lookupTime, "s (", lookupDataSize.float/result.lookupTime, " ops/sec)"
|
||||
echo " False positives: ", result.falsePositives, " (",
|
||||
result.falsePositives.float / lookupDataSize.float * 100, "%)"
|
||||
|
||||
proc runBenchmark(size: int, errorRate: float, pattern: DataPattern, name: string) =
|
||||
echo "\n=== Benchmark: ", name, " ==="
|
||||
echo "Size: ", size, " items"
|
||||
echo "Pattern: ", pattern
|
||||
|
||||
# Generate test data
|
||||
let data = generateBenchData(pattern, size, false)
|
||||
let lookupData = generateBenchData(pattern, size div 2, true)
|
||||
|
||||
# Run benchmarks for each hash type
|
||||
let nimHashResult = benchmarkHashType(htNimHash, size, errorRate, data, lookupData)
|
||||
let murmur128Result = benchmarkHashType(htMurmur128, size, errorRate, data, lookupData)
|
||||
let murmur32Result = benchmarkHashType(htMurmur32, size, errorRate, data, lookupData)
|
||||
|
||||
# Print individual results
|
||||
printResults("Nim's Default (Farm Hash)", nimHashResult, size, lookupData.len)
|
||||
printResults("MurmurHash3_128", murmur128Result, size, lookupData.len)
|
||||
printResults("MurmurHash3_32", murmur32Result, size, lookupData.len)
|
||||
|
||||
# Print comparisons
|
||||
echo "\nComparison (higher means better/faster):"
|
||||
echo " Insert Speed:"
|
||||
echo " Murmur128 vs NimHash: ", nimHashResult.insertTime/murmur128Result.insertTime, "x faster"
|
||||
echo " Murmur32 vs NimHash: ", nimHashResult.insertTime/murmur32Result.insertTime, "x faster"
|
||||
echo " Murmur128 vs Murmur32: ", murmur32Result.insertTime/murmur128Result.insertTime, "x faster"
|
||||
|
||||
echo " Lookup Speed:"
|
||||
echo " Murmur128 vs NimHash: ", nimHashResult.lookupTime/murmur128Result.lookupTime, "x faster"
|
||||
echo " Murmur32 vs NimHash: ", nimHashResult.lookupTime/murmur32Result.lookupTime, "x faster"
|
||||
echo " Murmur128 vs Murmur32: ", murmur32Result.lookupTime/murmur128Result.lookupTime, "x faster"
|
||||
|
||||
echo " False Positive Rates:"
|
||||
let fpRateNimHash = nimHashResult.falsePositives.float / lookupData.len.float
|
||||
let fpRateMurmur128 = murmur128Result.falsePositives.float / lookupData.len.float
|
||||
let fpRateMurmur32 = murmur32Result.falsePositives.float / lookupData.len.float
|
||||
|
||||
echo " Murmur128 vs NimHash: ", fpRateNimHash/fpRateMurmur128, "x better"
|
||||
echo " Murmur32 vs NimHash: ", fpRateNimHash/fpRateMurmur32, "x better"
|
||||
echo " Murmur128 vs Murmur32: ", fpRateMurmur32/fpRateMurmur128, "x better"
|
||||
|
||||
when isMainModule:
|
||||
const errorRate = 0.01
|
||||
|
||||
# Test each pattern
|
||||
for pattern in [dpRandom, dpSequential, dpFixed, dpLong, dpSpecial]:
|
||||
# Small dataset
|
||||
runBenchmark(10_000, errorRate, pattern, "Small " & $pattern)
|
||||
|
||||
# Medium dataset
|
||||
runBenchmark(100_000, errorRate, pattern, "Medium " & $pattern)
|
||||
|
||||
# Large dataset
|
||||
runBenchmark(1_000_000, errorRate, pattern, "Large " & $pattern)
|
||||
495
nim-bloom/benches/bench_results.txt
Normal file
495
nim-bloom/benches/bench_results.txt
Normal file
@ -0,0 +1,495 @@
|
||||
|
||||
=== Benchmark: Small dpRandom ===
|
||||
Size: 10000 items
|
||||
Pattern: dpRandom
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.006714000000000001s (1489425.081918379 ops/sec)
|
||||
Lookup time: 0.003387999999999995s (1475796.930342387 ops/sec)
|
||||
False positives: 51 (1.02%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.001955999999999999s (5112474.437627814 ops/sec)
|
||||
Lookup time: 0.0009989999999999999s (5005005.005005006 ops/sec)
|
||||
False positives: 25 (0.5%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.002221999999999995s (4500450.045004511 ops/sec)
|
||||
Lookup time: 0.001138s (4393673.110720562 ops/sec)
|
||||
False positives: 43 (0.86%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.432515337423315x faster
|
||||
Murmur32 vs NimHash: 3.021602160216029x faster
|
||||
Murmur128 vs Murmur32: 1.135991820040897x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.391391391391387x faster
|
||||
Murmur32 vs NimHash: 2.977152899824249x faster
|
||||
Murmur128 vs Murmur32: 1.139139139139139x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 2.04x better
|
||||
Murmur32 vs NimHash: 1.186046511627907x better
|
||||
Murmur128 vs Murmur32: 1.72x better
|
||||
|
||||
=== Benchmark: Medium dpRandom ===
|
||||
Size: 100000 items
|
||||
Pattern: dpRandom
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.04622000000000001s (2163565.556036348 ops/sec)
|
||||
Lookup time: 0.02333199999999999s (2142979.59883422 ops/sec)
|
||||
False positives: 402 (0.804%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.013294s (7522190.461862494 ops/sec)
|
||||
Lookup time: 0.006861999999999979s (7286505.392014012 ops/sec)
|
||||
False positives: 405 (0.8099999999999999%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.01558700000000002s (6415602.745877968 ops/sec)
|
||||
Lookup time: 0.008157999999999999s (6128953.174797745 ops/sec)
|
||||
False positives: 406 (0.8120000000000001%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.476756431472846x faster
|
||||
Murmur32 vs NimHash: 2.965291589144798x faster
|
||||
Murmur128 vs Murmur32: 1.172483827290508x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.400174876129417x faster
|
||||
Murmur32 vs NimHash: 2.860014709487619x faster
|
||||
Murmur128 vs Murmur32: 1.188866219761006x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9925925925925927x better
|
||||
Murmur32 vs NimHash: 0.9901477832512315x better
|
||||
Murmur128 vs Murmur32: 1.002469135802469x better
|
||||
|
||||
=== Benchmark: Large dpRandom ===
|
||||
Size: 1000000 items
|
||||
Pattern: dpRandom
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.4711130000000001s (2122632.998877127 ops/sec)
|
||||
Lookup time: 0.2430289999999999s (2057367.639252929 ops/sec)
|
||||
False positives: 4104 (0.8208%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.1467989999999999s (6812035.504329053 ops/sec)
|
||||
Lookup time: 0.07689400000000002s (6502457.929097197 ops/sec)
|
||||
False positives: 4187 (0.8373999999999999%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.1697340000000001s (5891571.517786653 ops/sec)
|
||||
Lookup time: 0.08597399999999999s (5815711.726801126 ops/sec)
|
||||
False positives: 4130 (0.826%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.209238482550974x faster
|
||||
Murmur32 vs NimHash: 2.775595932459024x faster
|
||||
Murmur128 vs Murmur32: 1.156234034291788x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.160571696101125x faster
|
||||
Murmur32 vs NimHash: 2.826773210505501x faster
|
||||
Murmur128 vs Murmur32: 1.118084635992405x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9801767375208981x better
|
||||
Murmur32 vs NimHash: 0.9937046004842615x better
|
||||
Murmur128 vs Murmur32: 0.9863864342010987x better
|
||||
|
||||
=== Benchmark: Small dpSequential ===
|
||||
Size: 10000 items
|
||||
Pattern: dpSequential
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.004508000000000179s (2218278.615794056 ops/sec)
|
||||
Lookup time: 0.002232000000000012s (2240143.369175615 ops/sec)
|
||||
False positives: 42 (0.84%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.001201000000000008s (8326394.671107357 ops/sec)
|
||||
Lookup time: 0.0006349999999999412s (7874015.748032225 ops/sec)
|
||||
False positives: 47 (0.9400000000000001%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.00151200000000018s (6613756.613755827 ops/sec)
|
||||
Lookup time: 0.0007530000000000037s (6640106.241699835 ops/sec)
|
||||
False positives: 51 (1.02%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.753538717735346x faster
|
||||
Murmur32 vs NimHash: 2.981481481481245x faster
|
||||
Murmur128 vs Murmur32: 1.258950874271582x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.514960629921604x faster
|
||||
Murmur32 vs NimHash: 2.964143426294822x faster
|
||||
Murmur128 vs Murmur32: 1.185826771653659x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.8936170212765957x better
|
||||
Murmur32 vs NimHash: 0.8235294117647057x better
|
||||
Murmur128 vs Murmur32: 1.085106382978724x better
|
||||
|
||||
=== Benchmark: Medium dpSequential ===
|
||||
Size: 100000 items
|
||||
Pattern: dpSequential
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.04492400000000019s (2225981.65791113 ops/sec)
|
||||
Lookup time: 0.02218900000000001s (2253368.78633557 ops/sec)
|
||||
False positives: 436 (0.872%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.012737s (7851142.341210647 ops/sec)
|
||||
Lookup time: 0.006547999999999998s (7635919.364691511 ops/sec)
|
||||
False positives: 422 (0.844%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.01476000000000011s (6775067.750677458 ops/sec)
|
||||
Lookup time: 0.008092999999999684s (6178178.672927462 ops/sec)
|
||||
False positives: 446 (0.8920000000000001%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.527047185365486x faster
|
||||
Murmur32 vs NimHash: 3.043631436314354x faster
|
||||
Murmur128 vs Murmur32: 1.1588286095627x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.388668295662801x faster
|
||||
Murmur32 vs NimHash: 2.741752131471751x faster
|
||||
Murmur128 vs Murmur32: 1.23594990836892x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 1.033175355450237x better
|
||||
Murmur32 vs NimHash: 0.9775784753363228x better
|
||||
Murmur128 vs Murmur32: 1.056872037914692x better
|
||||
|
||||
=== Benchmark: Large dpSequential ===
|
||||
Size: 1000000 items
|
||||
Pattern: dpSequential
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.4540329999999999s (2202483.079423743 ops/sec)
|
||||
Lookup time: 0.2223580000000003s (2248626.089459338 ops/sec)
|
||||
False positives: 4037 (0.8074%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.1384119999999998s (7224807.097650506 ops/sec)
|
||||
Lookup time: 0.07243599999999972s (6902645.093599894 ops/sec)
|
||||
False positives: 4205 (0.8410000000000001%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.1643889999999999s (6083132.083046923 ops/sec)
|
||||
Lookup time: 0.08048799999999989s (6212106.152469941 ops/sec)
|
||||
False positives: 4024 (0.8048%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.280300840967551x faster
|
||||
Murmur32 vs NimHash: 2.761942709062043x faster
|
||||
Murmur128 vs Murmur32: 1.187678813975668x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.069716715445374x faster
|
||||
Murmur32 vs NimHash: 2.762622999701826x faster
|
||||
Murmur128 vs Murmur32: 1.111160196587335x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9600475624256836x better
|
||||
Murmur32 vs NimHash: 1.003230616302187x better
|
||||
Murmur128 vs Murmur32: 0.9569560047562424x better
|
||||
|
||||
=== Benchmark: Small dpFixed ===
|
||||
Size: 10000 items
|
||||
Pattern: dpFixed
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.004486999999999686s (2228660.574994585 ops/sec)
|
||||
Lookup time: 0.002098999999999851s (2382086.707956339 ops/sec)
|
||||
False positives: 43 (0.86%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.001263000000000236s (7917656.373711901 ops/sec)
|
||||
Lookup time: 0.0006620000000001625s (7552870.090632588 ops/sec)
|
||||
False positives: 47 (0.9400000000000001%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.001565000000000261s (6389776.357826412 ops/sec)
|
||||
Lookup time: 0.0007539999999996994s (6631299.734750655 ops/sec)
|
||||
False positives: 35 (0.7000000000000001%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.552652414884281x faster
|
||||
Murmur32 vs NimHash: 2.86709265175651x faster
|
||||
Murmur128 vs Murmur32: 1.239113222486119x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.170694864047335x faster
|
||||
Murmur32 vs NimHash: 2.783819628648127x faster
|
||||
Murmur128 vs Murmur32: 1.13897280966694x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9148936170212766x better
|
||||
Murmur32 vs NimHash: 1.228571428571429x better
|
||||
Murmur128 vs Murmur32: 0.7446808510638298x better
|
||||
|
||||
=== Benchmark: Medium dpFixed ===
|
||||
Size: 100000 items
|
||||
Pattern: dpFixed
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.04369800000000001s (2288434.253283903 ops/sec)
|
||||
Lookup time: 0.02203400000000011s (2269220.295906316 ops/sec)
|
||||
False positives: 411 (0.822%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.013401s (7462129.691814045 ops/sec)
|
||||
Lookup time: 0.00719499999999984s (6949270.32661586 ops/sec)
|
||||
False positives: 440 (0.88%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.01487600000000011s (6722237.160526973 ops/sec)
|
||||
Lookup time: 0.007677999999999852s (6512112.529304632 ops/sec)
|
||||
False positives: 433 (0.8659999999999999%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.260801432728903x faster
|
||||
Murmur32 vs NimHash: 2.937483194407078x faster
|
||||
Murmur128 vs Murmur32: 1.110066412954266x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.062404447533092x faster
|
||||
Murmur32 vs NimHash: 2.86975774941398x faster
|
||||
Murmur128 vs Murmur32: 1.067129951355111x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.934090909090909x better
|
||||
Murmur32 vs NimHash: 0.9491916859122402x better
|
||||
Murmur128 vs Murmur32: 0.9840909090909089x better
|
||||
|
||||
=== Benchmark: Large dpFixed ===
|
||||
Size: 1000000 items
|
||||
Pattern: dpFixed
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.4461910000000002s (2241192.673092912 ops/sec)
|
||||
Lookup time: 0.2275340000000003s (2197473.784137752 ops/sec)
|
||||
False positives: 4084 (0.8168%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.1576560000000002s (6342923.834170589 ops/sec)
|
||||
Lookup time: 0.07750000000000057s (6451612.903225759 ops/sec)
|
||||
False positives: 4109 (0.8218%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.1656709999999997s (6036059.418968932 ops/sec)
|
||||
Lookup time: 0.08378900000000034s (5967370.418551337 ops/sec)
|
||||
False positives: 4085 (0.8170000000000001%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 2.830155528492411x faster
|
||||
Murmur32 vs NimHash: 2.693235388209168x faster
|
||||
Murmur128 vs Murmur32: 1.050838534530874x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 2.935922580645144x faster
|
||||
Murmur32 vs NimHash: 2.715559321629324x faster
|
||||
Murmur128 vs Murmur32: 1.081148387096771x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9939157945972257x better
|
||||
Murmur32 vs NimHash: 0.9997552019583843x better
|
||||
Murmur128 vs Murmur32: 0.9941591628133366x better
|
||||
|
||||
=== Benchmark: Small dpLong ===
|
||||
Size: 10000 items
|
||||
Pattern: dpLong
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.009589999999999321s (1042752.86757046 ops/sec)
|
||||
Lookup time: 0.005217000000000027s (958405.2137243577 ops/sec)
|
||||
False positives: 39 (0.7799999999999999%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.001508000000000287s (6631299.734746749 ops/sec)
|
||||
Lookup time: 0.0007299999999998974s (6849315.068494113 ops/sec)
|
||||
False positives: 56 (1.12%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.001934000000000324s (5170630.816958802 ops/sec)
|
||||
Lookup time: 0.0009790000000000632s (5107252.298263205 ops/sec)
|
||||
False positives: 40 (0.8%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 6.359416445621682x faster
|
||||
Murmur32 vs NimHash: 4.95863495346314x faster
|
||||
Murmur128 vs Murmur32: 1.282493368700236x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 7.146575342466795x faster
|
||||
Murmur32 vs NimHash: 5.328907048007856x faster
|
||||
Murmur128 vs Murmur32: 1.341095890411234x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.6964285714285714x better
|
||||
Murmur32 vs NimHash: 0.975x better
|
||||
Murmur128 vs Murmur32: 0.7142857142857143x better
|
||||
|
||||
=== Benchmark: Medium dpLong ===
|
||||
Size: 100000 items
|
||||
Pattern: dpLong
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.09930899999999987s (1006958.080335117 ops/sec)
|
||||
Lookup time: 0.04954899999999984s (1009102.100950577 ops/sec)
|
||||
False positives: 393 (0.786%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.0148350000000006s (6740815.63869201 ops/sec)
|
||||
Lookup time: 0.007865000000000677s (6357279.084551265 ops/sec)
|
||||
False positives: 422 (0.844%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.0197699999999994s (5058168.942842845 ops/sec)
|
||||
Lookup time: 0.01105200000000028s (4524068.041983236 ops/sec)
|
||||
False positives: 404 (0.8080000000000001%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 6.69423660262864x faster
|
||||
Murmur32 vs NimHash: 5.023216995447794x faster
|
||||
Murmur128 vs Murmur32: 1.33265925176937x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 6.299936427208592x faster
|
||||
Murmur32 vs NimHash: 4.483260948244532x faster
|
||||
Murmur128 vs Murmur32: 1.405212968849248x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9312796208530807x better
|
||||
Murmur32 vs NimHash: 0.9727722772277229x better
|
||||
Murmur128 vs Murmur32: 0.9573459715639812x better
|
||||
|
||||
=== Benchmark: Large dpLong ===
|
||||
Size: 1000000 items
|
||||
Pattern: dpLong
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 1.032455000000001s (968565.2159174002 ops/sec)
|
||||
Lookup time: 0.5242619999999993s (953721.6124762058 ops/sec)
|
||||
False positives: 4143 (0.8286%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.1612590000000003s (6201204.273869975 ops/sec)
|
||||
Lookup time: 0.08462199999999953s (5908628.961735751 ops/sec)
|
||||
False positives: 4044 (0.8088%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.2145720000000004s (4660440.318401273 ops/sec)
|
||||
Lookup time: 0.1070960000000003s (4668708.448494795 ops/sec)
|
||||
False positives: 4039 (0.8078%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 6.402464358578429x faster
|
||||
Murmur32 vs NimHash: 4.81169490893499x faster
|
||||
Murmur128 vs Murmur32: 1.330604803452831x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 6.195339273475009x faster
|
||||
Murmur32 vs NimHash: 4.895252857249551x faster
|
||||
Murmur128 vs Murmur32: 1.265581054572108x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 1.024480712166172x better
|
||||
Murmur32 vs NimHash: 1.025748947759346x better
|
||||
Murmur128 vs Murmur32: 0.9987636003956479x better
|
||||
|
||||
=== Benchmark: Small dpSpecial ===
|
||||
Size: 10000 items
|
||||
Pattern: dpSpecial
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.005548999999998472s (1802126.509281448 ops/sec)
|
||||
Lookup time: 0.002686999999999884s (1860811.313732868 ops/sec)
|
||||
False positives: 39 (0.7799999999999999%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.001593999999998985s (6273525.721459452 ops/sec)
|
||||
Lookup time: 0.0008440000000007331s (5924170.616108599 ops/sec)
|
||||
False positives: 33 (0.66%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.001865999999999701s (5359056.806003002 ops/sec)
|
||||
Lookup time: 0.0009710000000016095s (5149330.587015152 ops/sec)
|
||||
False positives: 51 (1.02%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.481179422836891x faster
|
||||
Murmur32 vs NimHash: 2.973740621650247x faster
|
||||
Murmur128 vs Murmur32: 1.170639899624146x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.183649289096623x faster
|
||||
Murmur32 vs NimHash: 2.767250257461823x faster
|
||||
Murmur128 vs Murmur32: 1.150473933650197x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 1.181818181818182x better
|
||||
Murmur32 vs NimHash: 0.7647058823529411x better
|
||||
Murmur128 vs Murmur32: 1.545454545454546x better
|
||||
|
||||
=== Benchmark: Medium dpSpecial ===
|
||||
Size: 100000 items
|
||||
Pattern: dpSpecial
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.04318999999999917s (2315350.775642554 ops/sec)
|
||||
Lookup time: 0.02338899999999988s (2137757.065287111 ops/sec)
|
||||
False positives: 435 (0.8699999999999999%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.01633100000000098s (6123323.740125772 ops/sec)
|
||||
Lookup time: 0.007922000000000651s (6311537.490532176 ops/sec)
|
||||
False positives: 405 (0.8099999999999999%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.01658399999999993s (6029908.345393174 ops/sec)
|
||||
Lookup time: 0.00944400000000023s (5294366.793731341 ops/sec)
|
||||
False positives: 368 (0.736%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 2.64466352336027x faster
|
||||
Murmur32 vs NimHash: 2.604317414375262x faster
|
||||
Murmur128 vs Murmur32: 1.015492009062454x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 2.952411007321126x faster
|
||||
Murmur32 vs NimHash: 2.476598898771634x faster
|
||||
Murmur128 vs Murmur32: 1.192123201211746x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 1.074074074074074x better
|
||||
Murmur32 vs NimHash: 1.182065217391304x better
|
||||
Murmur128 vs Murmur32: 0.9086419753086421x better
|
||||
|
||||
=== Benchmark: Large dpSpecial ===
|
||||
Size: 1000000 items
|
||||
Pattern: dpSpecial
|
||||
|
||||
Nim's Default (Farm Hash) Results:
|
||||
Insert time: 0.4443290000000015s (2250584.589347075 ops/sec)
|
||||
Lookup time: 0.2233900000000002s (2238238.058999953 ops/sec)
|
||||
False positives: 4096 (0.8191999999999999%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.1440049999999999s (6944203.326273397 ops/sec)
|
||||
Lookup time: 0.07467199999999963s (6695950.289265086 ops/sec)
|
||||
False positives: 4137 (0.8274%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.1650030000000005s (6060495.869772048 ops/sec)
|
||||
Lookup time: 0.08332200000000078s (6000816.110991038 ops/sec)
|
||||
False positives: 4222 (0.8444%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.085510919759742x faster
|
||||
Murmur32 vs NimHash: 2.692854069319953x faster
|
||||
Murmur128 vs Murmur32: 1.145814381445092x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 2.991616670237858x faster
|
||||
Murmur32 vs NimHash: 2.681044622068578x faster
|
||||
Murmur128 vs Murmur32: 1.115839940004302x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9900894367899443x better
|
||||
Murmur32 vs NimHash: 0.9701563240170534x better
|
||||
Murmur128 vs Murmur32: 1.020546289581822x better
|
||||
495
nim-bloom/benches/bench_results_nimStringHash2.txt
Normal file
495
nim-bloom/benches/bench_results_nimStringHash2.txt
Normal file
@ -0,0 +1,495 @@
|
||||
|
||||
=== Benchmark: Small dpRandom ===
|
||||
Size: 10000 items
|
||||
Pattern: dpRandom
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.004986000000000001s (2005615.724027276 ops/sec)
|
||||
Lookup time: 0.00238s (2100840.336134454 ops/sec)
|
||||
False positives: 43 (0.86%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.001388s (7204610.951008644 ops/sec)
|
||||
Lookup time: 0.000954s (5241090.146750525 ops/sec)
|
||||
False positives: 25 (0.5%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.001744000000000002s (5733944.954128432 ops/sec)
|
||||
Lookup time: 0.000780999999999997s (6402048.655569807 ops/sec)
|
||||
False positives: 43 (0.86%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.592219020172911x faster
|
||||
Murmur32 vs NimHash: 2.858944954128437x faster
|
||||
Murmur128 vs Murmur32: 1.256484149855909x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 2.49475890985325x faster
|
||||
Murmur32 vs NimHash: 3.047375160051228x faster
|
||||
Murmur128 vs Murmur32: 0.8186582809224288x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 1.72x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 1.72x better
|
||||
|
||||
=== Benchmark: Medium dpRandom ===
|
||||
Size: 100000 items
|
||||
Pattern: dpRandom
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.050316s (1987439.383098816 ops/sec)
|
||||
Lookup time: 0.02490799999999999s (2007387.184840213 ops/sec)
|
||||
False positives: 406 (0.8120000000000001%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.01539100000000002s (6497303.618998109 ops/sec)
|
||||
Lookup time: 0.008371999999999991s (5972288.580984239 ops/sec)
|
||||
False positives: 405 (0.8099999999999999%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.01614399999999999s (6194251.734390489 ops/sec)
|
||||
Lookup time: 0.008244000000000001s (6065016.982047549 ops/sec)
|
||||
False positives: 406 (0.8120000000000001%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.269183288935089x faster
|
||||
Murmur32 vs NimHash: 3.116699702675918x faster
|
||||
Murmur128 vs Murmur32: 1.048924696251054x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 2.975155279503107x faster
|
||||
Murmur32 vs NimHash: 3.021348859776805x faster
|
||||
Murmur128 vs Murmur32: 0.9847109412326817x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 1.002469135802469x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 1.002469135802469x better
|
||||
|
||||
=== Benchmark: Large dpRandom ===
|
||||
Size: 1000000 items
|
||||
Pattern: dpRandom
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.522163s (1915110.798735261 ops/sec)
|
||||
Lookup time: 0.257452s (1942109.597128785 ops/sec)
|
||||
False positives: 4130 (0.826%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.1587529999999999s (6299093.560436657 ops/sec)
|
||||
Lookup time: 0.07841999999999993s (6375924.509053818 ops/sec)
|
||||
False positives: 4187 (0.8373999999999999%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.170582s (5862283.242077123 ops/sec)
|
||||
Lookup time: 0.08690500000000001s (5753408.894770151 ops/sec)
|
||||
False positives: 4130 (0.826%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.289153590798286x faster
|
||||
Murmur32 vs NimHash: 3.061067404532718x faster
|
||||
Murmur128 vs Murmur32: 1.074511977726406x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.282989033409847x faster
|
||||
Murmur32 vs NimHash: 2.96245325355273x faster
|
||||
Murmur128 vs Murmur32: 1.108199438918644x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9863864342010987x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 0.9863864342010987x better
|
||||
|
||||
=== Benchmark: Small dpSequential ===
|
||||
Size: 10000 items
|
||||
Pattern: dpSequential
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.004666000000000059s (2143163.309044122 ops/sec)
|
||||
Lookup time: 0.002341000000000149s (2135839.384878122 ops/sec)
|
||||
False positives: 51 (1.02%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.001309999999999922s (7633587.786259995 ops/sec)
|
||||
Lookup time: 0.0007189999999999142s (6954102.920724057 ops/sec)
|
||||
False positives: 47 (0.9400000000000001%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.001513000000000098s (6609385.327164147 ops/sec)
|
||||
Lookup time: 0.0007410000000001027s (6747638.32658476 ops/sec)
|
||||
False positives: 51 (1.02%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.561832061068959x faster
|
||||
Murmur32 vs NimHash: 3.08393919365483x faster
|
||||
Murmur128 vs Murmur32: 1.154961832061212x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.25591098748321x faster
|
||||
Murmur32 vs NimHash: 3.159244264507185x faster
|
||||
Murmur128 vs Murmur32: 1.030598052851448x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 1.085106382978724x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 1.085106382978724x better
|
||||
|
||||
=== Benchmark: Medium dpSequential ===
|
||||
Size: 100000 items
|
||||
Pattern: dpSequential
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.0480590000000003s (2080775.713185863 ops/sec)
|
||||
Lookup time: 0.02284599999999992s (2188566.926376616 ops/sec)
|
||||
False positives: 446 (0.8920000000000001%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.01357099999999978s (7368653.746960551 ops/sec)
|
||||
Lookup time: 0.008311999999999653s (6015399.422521907 ops/sec)
|
||||
False positives: 422 (0.844%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.01522699999999988s (6567281.802062178 ops/sec)
|
||||
Lookup time: 0.007773999999999948s (6431695.394906141 ops/sec)
|
||||
False positives: 446 (0.8920000000000001%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.541301304251793x faster
|
||||
Murmur32 vs NimHash: 3.156169961253082x faster
|
||||
Murmur128 vs Murmur32: 1.122024906049674x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 2.7485563041387x faster
|
||||
Murmur32 vs NimHash: 2.938770259840504x faster
|
||||
Murmur128 vs Murmur32: 0.9352743022136998x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 1.056872037914692x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 1.056872037914692x better
|
||||
|
||||
=== Benchmark: Large dpSequential ===
|
||||
Size: 1000000 items
|
||||
Pattern: dpSequential
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.4705080000000001s (2125362.374284815 ops/sec)
|
||||
Lookup time: 0.233136s (2144670.921693775 ops/sec)
|
||||
False positives: 4024 (0.8048%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.1423380000000001s (7025530.778850338 ops/sec)
|
||||
Lookup time: 0.07601600000000008s (6577562.61839612 ops/sec)
|
||||
False positives: 4205 (0.8410000000000001%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.1583019999999999s (6317039.582570027 ops/sec)
|
||||
Lookup time: 0.08089100000000027s (6181157.359904048 ops/sec)
|
||||
False positives: 4024 (0.8048%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.305568435695316x faster
|
||||
Murmur32 vs NimHash: 2.972217659915859x faster
|
||||
Murmur128 vs Murmur32: 1.112155573353566x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.066933277204796x faster
|
||||
Murmur32 vs NimHash: 2.88210060451718x faster
|
||||
Murmur128 vs Murmur32: 1.064131235529365x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9569560047562424x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 0.9569560047562424x better
|
||||
|
||||
=== Benchmark: Small dpFixed ===
|
||||
Size: 10000 items
|
||||
Pattern: dpFixed
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.004584000000000366s (2181500.872600175 ops/sec)
|
||||
Lookup time: 0.002313999999999705s (2160760.587727155 ops/sec)
|
||||
False positives: 35 (0.7000000000000001%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.001360000000000028s (7352941.176470438 ops/sec)
|
||||
Lookup time: 0.0006840000000001289s (7309941.520466458 ops/sec)
|
||||
False positives: 47 (0.9400000000000001%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.001682999999999879s (5941770.647653428 ops/sec)
|
||||
Lookup time: 0.0007459999999999134s (6702412.868633485 ops/sec)
|
||||
False positives: 35 (0.7000000000000001%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.370588235294318x faster
|
||||
Murmur32 vs NimHash: 2.723707664884548x faster
|
||||
Murmur128 vs Murmur32: 1.237499999999886x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.383040935671446x faster
|
||||
Murmur32 vs NimHash: 3.101876675603182x faster
|
||||
Murmur128 vs Murmur32: 1.090643274853469x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.7446808510638298x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 0.7446808510638298x better
|
||||
|
||||
=== Benchmark: Medium dpFixed ===
|
||||
Size: 100000 items
|
||||
Pattern: dpFixed
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.04620999999999986s (2164033.758926646 ops/sec)
|
||||
Lookup time: 0.02306600000000003s (2167692.707881728 ops/sec)
|
||||
False positives: 433 (0.8659999999999999%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.01405199999999995s (7116424.708226611 ops/sec)
|
||||
Lookup time: 0.007185999999999915s (6957973.838018452 ops/sec)
|
||||
False positives: 440 (0.88%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.01542299999999974s (6483822.86195952 ops/sec)
|
||||
Lookup time: 0.007922999999999902s (6310740.880979504 ops/sec)
|
||||
False positives: 433 (0.8659999999999999%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.288499857671507x faster
|
||||
Murmur32 vs NimHash: 2.996174544511485x faster
|
||||
Murmur128 vs Murmur32: 1.097566182749772x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.209852490954677x faster
|
||||
Murmur32 vs NimHash: 2.911270983213469x faster
|
||||
Murmur128 vs Murmur32: 1.10256053437239x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9840909090909089x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 0.9840909090909089x better
|
||||
|
||||
=== Benchmark: Large dpFixed ===
|
||||
Size: 1000000 items
|
||||
Pattern: dpFixed
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.4785590000000002s (2089606.506198817 ops/sec)
|
||||
Lookup time: 0.2445870000000001s (2044262.368809462 ops/sec)
|
||||
False positives: 4085 (0.8170000000000001%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.1526930000000002s (6549088.694308178 ops/sec)
|
||||
Lookup time: 0.07818300000000011s (6395252.164792849 ops/sec)
|
||||
False positives: 4109 (0.8218%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.1651470000000002s (6055211.417706643 ops/sec)
|
||||
Lookup time: 0.08524499999999957s (5865446.653762713 ops/sec)
|
||||
False positives: 4085 (0.8170000000000001%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.134125336459429x faster
|
||||
Murmur32 vs NimHash: 2.897775920846275x faster
|
||||
Murmur128 vs Murmur32: 1.081562350598914x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.128391082460378x faster
|
||||
Murmur32 vs NimHash: 2.869224001407723x faster
|
||||
Murmur128 vs Murmur32: 1.090326541575527x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9941591628133366x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 0.9941591628133366x better
|
||||
|
||||
=== Benchmark: Small dpLong ===
|
||||
Size: 10000 items
|
||||
Pattern: dpLong
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.008112999999999815s (1232589.670898586 ops/sec)
|
||||
Lookup time: 0.003777000000000363s (1323801.959226772 ops/sec)
|
||||
False positives: 40 (0.8%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.001597000000000293s (6261740.763931226 ops/sec)
|
||||
Lookup time: 0.0007489999999998886s (6675567.423231967 ops/sec)
|
||||
False positives: 56 (1.12%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.001947999999999617s (5133470.2258737 ops/sec)
|
||||
Lookup time: 0.001180999999999877s (4233700.254022458 ops/sec)
|
||||
False positives: 40 (0.8%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 5.080150281777287x faster
|
||||
Murmur32 vs NimHash: 4.164784394251238x faster
|
||||
Murmur128 vs Murmur32: 1.219787100813563x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 5.042723631509913x faster
|
||||
Murmur32 vs NimHash: 3.198137171888872x faster
|
||||
Murmur128 vs Murmur32: 1.576769025367226x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.7142857142857143x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 0.7142857142857143x better
|
||||
|
||||
=== Benchmark: Medium dpLong ===
|
||||
Size: 100000 items
|
||||
Pattern: dpLong
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.07904900000000037s (1265038.140899942 ops/sec)
|
||||
Lookup time: 0.04046299999999992s (1235696.809430841 ops/sec)
|
||||
False positives: 404 (0.8080000000000001%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.015733s (6356066.86582343 ops/sec)
|
||||
Lookup time: 0.007823000000000135s (6391409.945033764 ops/sec)
|
||||
False positives: 422 (0.844%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.02339799999999936s (4273869.561501099 ops/sec)
|
||||
Lookup time: 0.01039399999999979s (4810467.577448624 ops/sec)
|
||||
False positives: 404 (0.8080000000000001%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 5.024407296764786x faster
|
||||
Murmur32 vs NimHash: 3.37845114967102x faster
|
||||
Murmur128 vs Murmur32: 1.487192525265326x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 5.172312412118013x faster
|
||||
Murmur32 vs NimHash: 3.892918991726066x faster
|
||||
Murmur128 vs Murmur32: 1.328646299373592x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9573459715639812x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 0.9573459715639812x better
|
||||
|
||||
=== Benchmark: Large dpLong ===
|
||||
Size: 1000000 items
|
||||
Pattern: dpLong
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.814165s (1228252.25844884 ops/sec)
|
||||
Lookup time: 0.4165090000000005s (1200454.251888913 ops/sec)
|
||||
False positives: 4039 (0.8078%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.1629300000000002s (6137605.10648744 ops/sec)
|
||||
Lookup time: 0.08406800000000025s (5947566.255888073 ops/sec)
|
||||
False positives: 4044 (0.8088%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.2405530000000002s (4157088.042967658 ops/sec)
|
||||
Lookup time: 0.1411719999999992s (3541778.822996082 ops/sec)
|
||||
False positives: 4039 (0.8078%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 4.997023261523347x faster
|
||||
Murmur32 vs NimHash: 3.384555586502763x faster
|
||||
Murmur128 vs Murmur32: 1.476419321180874x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 4.954429747347376x faster
|
||||
Murmur32 vs NimHash: 2.950365511574554x faster
|
||||
Murmur128 vs Murmur32: 1.679259646952452x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9987636003956479x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 0.9987636003956479x better
|
||||
|
||||
=== Benchmark: Small dpSpecial ===
|
||||
Size: 10000 items
|
||||
Pattern: dpSpecial
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.007111000000000089s (1406271.97299956 ops/sec)
|
||||
Lookup time: 0.003309000000000673s (1511030.522816254 ops/sec)
|
||||
False positives: 51 (1.02%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.001907000000000103s (5243838.489774232 ops/sec)
|
||||
Lookup time: 0.0009350000000001302s (5347593.582886956 ops/sec)
|
||||
False positives: 33 (0.66%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.001981000000000677s (5047955.577989189 ops/sec)
|
||||
Lookup time: 0.0009660000000000224s (5175983.436852882 ops/sec)
|
||||
False positives: 51 (1.02%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.728893550078503x faster
|
||||
Murmur32 vs NimHash: 3.589601211508158x faster
|
||||
Murmur128 vs Murmur32: 1.03880440482463x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.539037433155307x faster
|
||||
Murmur32 vs NimHash: 3.425465838509933x faster
|
||||
Murmur128 vs Murmur32: 1.033155080213784x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 1.545454545454546x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 1.545454545454546x better
|
||||
|
||||
=== Benchmark: Medium dpSpecial ===
|
||||
Size: 100000 items
|
||||
Pattern: dpSpecial
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.05804300000000051s (1722860.637802993 ops/sec)
|
||||
Lookup time: 0.03600399999999837s (1388734.585046169 ops/sec)
|
||||
False positives: 368 (0.736%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.01871899999999904s (5342165.713980721 ops/sec)
|
||||
Lookup time: 0.008850999999999942s (5649079.200090422 ops/sec)
|
||||
False positives: 405 (0.8099999999999999%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.01748200000000111s (5720169.317011422 ops/sec)
|
||||
Lookup time: 0.008943999999999619s (5590339.892665712 ops/sec)
|
||||
False positives: 368 (0.736%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 3.100753245365857x faster
|
||||
Murmur32 vs NimHash: 3.320157876672968x faster
|
||||
Murmur128 vs Murmur32: 0.9339174101181689x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 4.067788950400927x faster
|
||||
Murmur32 vs NimHash: 4.025491949910544x faster
|
||||
Murmur128 vs Murmur32: 1.010507287312132x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 0.9086419753086421x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 0.9086419753086421x better
|
||||
|
||||
=== Benchmark: Large dpSpecial ===
|
||||
Size: 1000000 items
|
||||
Pattern: dpSpecial
|
||||
|
||||
Nim's Default (MurmurHash3_32) Results:
|
||||
Insert time: 0.5143779999999989s (1944095.58729184 ops/sec)
|
||||
Lookup time: 0.2718509999999998s (1839242.820515651 ops/sec)
|
||||
False positives: 4222 (0.8444%)
|
||||
|
||||
MurmurHash3_128 Results:
|
||||
Insert time: 0.1766860000000001s (5659757.988748398 ops/sec)
|
||||
Lookup time: 0.08288000000000117s (6032818.532818447 ops/sec)
|
||||
False positives: 4137 (0.8274%)
|
||||
|
||||
MurmurHash3_32 Results:
|
||||
Insert time: 0.1758810000000004s (5685662.464962092 ops/sec)
|
||||
Lookup time: 0.09260200000000118s (5399451.415736092 ops/sec)
|
||||
False positives: 4222 (0.8444%)
|
||||
|
||||
Comparison (higher means better/faster):
|
||||
Insert Speed:
|
||||
Murmur128 vs NimHash: 2.911254994736417x faster
|
||||
Murmur32 vs NimHash: 2.924579687402265x faster
|
||||
Murmur128 vs Murmur32: 0.9954438948190591x faster
|
||||
Lookup Speed:
|
||||
Murmur128 vs NimHash: 3.280055501930454x faster
|
||||
Murmur32 vs NimHash: 2.935692533638543x faster
|
||||
Murmur128 vs Murmur32: 1.117302123552122x faster
|
||||
False Positive Rates:
|
||||
Murmur128 vs NimHash: 1.020546289581822x better
|
||||
Murmur32 vs NimHash: 1.0x better
|
||||
Murmur128 vs Murmur32: 1.020546289581822x better
|
||||
1
nim-bloom/benches/config.nims
Normal file
1
nim-bloom/benches/config.nims
Normal file
@ -0,0 +1 @@
|
||||
switch("path", "$projectDir/../src")
|
||||
@ -3,28 +3,44 @@ import hashes
|
||||
import strutils
|
||||
import private/probabilities
|
||||
|
||||
# Import MurmurHash3 code for large-scale use cases
|
||||
# Import MurmurHash3 code with both 128-bit and 32-bit implementations
|
||||
{.compile: "murmur3.c".}
|
||||
|
||||
type
|
||||
HashType* = enum
|
||||
htMurmur128, # Default: MurmurHash3_x64_128
|
||||
htMurmur32, # MurmurHash3_x86_32
|
||||
htNimHash # Nim's Default Hash (currently Farm Hash)
|
||||
|
||||
BloomFilterError* = object of CatchableError
|
||||
|
||||
MurmurHashes = array[0..1, int]
|
||||
|
||||
BloomFilter* = object
|
||||
capacity*: int
|
||||
errorRate*: float
|
||||
kHashes*: int
|
||||
mBits*: int
|
||||
intArray: seq[int]
|
||||
useExtendedHash*: bool # Use 128-bit MurmurHash3 for very large filters
|
||||
hashType*: HashType
|
||||
|
||||
{.push overflowChecks: off.} # Turn off overflow checks for hashing operations
|
||||
|
||||
proc rawMurmurHash(key: cstring, len: int, seed: uint32,
|
||||
proc rawMurmurHash128(key: cstring, len: int, seed: uint32,
|
||||
outHashes: var MurmurHashes): void {.
|
||||
importc: "MurmurHash3_x64_128".}
|
||||
|
||||
proc murmurHash(key: string, seed = 0'u32): MurmurHashes =
|
||||
rawMurmurHash(key, key.len, seed, result)
|
||||
proc rawMurmurHash32(key: cstring, len: int, seed: uint32,
|
||||
outHashes: ptr uint32): void {.
|
||||
importc: "MurmurHash3_x86_32".}
|
||||
|
||||
proc murmurHash128(key: string, seed = 0'u32): MurmurHashes =
|
||||
rawMurmurHash128(key, key.len, seed, result)
|
||||
|
||||
proc murmurHash32(key: string, seed = 0'u32): uint32 =
|
||||
var result: uint32
|
||||
rawMurmurHash32(key, key.len, seed, addr result)
|
||||
result
|
||||
|
||||
proc hashN(item: string, n: int, maxValue: int): int =
|
||||
## Get the nth hash using Nim's built-in hash function using
|
||||
@ -32,10 +48,16 @@ proc hashN(item: string, n: int, maxValue: int): int =
|
||||
## http://www.eecs.harvard.edu/~kirsch/pubs/bbbf/rsa.pdf
|
||||
let
|
||||
hashA = abs(hash(item)) mod maxValue # Use abs to handle negative hashes
|
||||
hashB = abs(hash(item & " b")) mod maxValue
|
||||
hashB = abs(hash(item & " b")) mod maxValue # string concatenation
|
||||
abs((hashA + n * hashB)) mod maxValue
|
||||
# # Use bit rotation for second hash instead of string concatenation if speed if preferred over FP-rate
|
||||
# # Rotate left by 21 bits (lower the rotation, higher the speed but higher the FP-rate too)
|
||||
# hashB = abs(
|
||||
# ((h shl 21) or (h shr (sizeof(int) * 8 - 21)))
|
||||
# ) mod maxValue
|
||||
# abs((hashA + n.int64 * hashB)) mod maxValue
|
||||
|
||||
{.pop.} # Restore overflow checks
|
||||
{.pop.}
|
||||
|
||||
proc getMOverNBitsForK(k: int, targetError: float,
|
||||
probabilityTable = kErrors): int =
|
||||
@ -49,17 +71,24 @@ proc getMOverNBitsForK(k: int, targetError: float,
|
||||
return mOverN
|
||||
|
||||
raise newException(BloomFilterError,
|
||||
"Specified value of k and error rate for which is not achievable using less than 4 bytes / element.")
|
||||
"Specified value of k and error rate not achievable using less than 4 bytes / element.")
|
||||
|
||||
proc initializeBloomFilter*(capacity: int, errorRate: float, k = 0,
|
||||
forceNBitsPerElem = 0,
|
||||
useExtendedHash = false): BloomFilter =
|
||||
## Initializes a Bloom filter, using a specified ``capacity``,
|
||||
## ``errorRate``, and – optionally – specific number of k hash functions.
|
||||
## If ``kHashes`` is < 1 (default argument is 0), ``kHashes`` will be
|
||||
## optimally calculated. If capacity > 1M elements, consider setting
|
||||
## useExtendedHash = true to use 128-bit MurmurHash3 for better
|
||||
## collision resistance.
|
||||
hashType = htMurmur128): BloomFilter =
|
||||
## Initializes a Bloom filter with specified parameters.
|
||||
##
|
||||
## Parameters:
|
||||
## - capacity: Expected number of elements to be inserted
|
||||
## - errorRate: Desired false positive rate (e.g., 0.01 for 1%)
|
||||
## - k: Optional number of hash functions. If 0, calculated optimally
|
||||
## See http://pages.cs.wisc.edu/~cao/papers/summary-cache/node8.html for
|
||||
## useful tables on k and m/n (n bits per element) combinations.
|
||||
## - forceNBitsPerElem: Optional override for bits per element
|
||||
## - hashType: Choose hash function:
|
||||
## * htMurmur128: MurmurHash3_x64_128 (default) - recommended
|
||||
## * htMurmur32: MurmurHash3_x86_32
|
||||
## * htNimHash: Nim's Default Hash
|
||||
var
|
||||
kHashes: int
|
||||
nBitsPerElem: int
|
||||
@ -85,29 +114,44 @@ proc initializeBloomFilter*(capacity: int, errorRate: float, k = 0,
|
||||
kHashes: kHashes,
|
||||
mBits: mBits,
|
||||
intArray: newSeq[int](mInts),
|
||||
useExtendedHash: useExtendedHash
|
||||
hashType: hashType
|
||||
)
|
||||
|
||||
proc `$`*(bf: BloomFilter): string =
|
||||
## Prints the capacity, set error rate, number of k hash functions,
|
||||
## and total bits of memory allocated by the Bloom filter.
|
||||
"Bloom filter with $1 capacity, $2 error rate, $3 hash functions, and requiring $4 bits of memory." %
|
||||
## Prints the configuration of the Bloom filter.
|
||||
let hashType = case bf.hashType
|
||||
of htMurmur128: "MurmurHash3_x64_128"
|
||||
of htMurmur32: "MurmurHash3_x86_32"
|
||||
of htNimHash: "NimHashHash"
|
||||
|
||||
"Bloom filter with $1 capacity, $2 error rate, $3 hash functions, and requiring $4 bits of memory. Using $5." %
|
||||
[$bf.capacity,
|
||||
formatFloat(bf.errorRate, format = ffScientific, precision = 1),
|
||||
$bf.kHashes,
|
||||
$(bf.mBits div bf.capacity)]
|
||||
$(bf.mBits div bf.capacity),
|
||||
hashType]
|
||||
|
||||
{.push overflowChecks: off.} # Turn off overflow checks for hash computations
|
||||
|
||||
proc computeHashes(bf: BloomFilter, item: string): seq[int] =
|
||||
var hashes = newSeq[int](bf.kHashes)
|
||||
if bf.useExtendedHash:
|
||||
let murmurHashes = murmurHash(item, 0'u32)
|
||||
|
||||
case bf.hashType
|
||||
of htMurmur128:
|
||||
let murmurHashes = murmurHash128(item, 0'u32)
|
||||
for i in 0..<bf.kHashes:
|
||||
hashes[i] = abs((murmurHashes[0] + i.int64 * murmurHashes[1].int64).int) mod bf.mBits
|
||||
else:
|
||||
hashes[i] = abs((murmurHashes[0].int64 + i.int64 * murmurHashes[1].int64).int) mod bf.mBits
|
||||
of htMurmur32:
|
||||
let baseHash = murmurHash32(item, 0'u32)
|
||||
# let rotated = ((baseHash shl 13) or (baseHash shr (32 - 13)))
|
||||
let secondHash = murmurHash32(item & " b", 0'u32)
|
||||
for i in 0..<bf.kHashes:
|
||||
hashes[i] = abs((baseHash.int64 + i.int64 * secondHash.int64).int) mod bf.mBits
|
||||
|
||||
of htNimHash:
|
||||
for i in 0..<bf.kHashes:
|
||||
hashes[i] = hashN(item, i, bf.mBits)
|
||||
|
||||
hashes
|
||||
|
||||
{.pop.} # Restore overflow checks
|
||||
@ -122,7 +166,7 @@ proc insert*(bf: var BloomFilter, item: string) =
|
||||
bf.intArray[intAddress] = bf.intArray[intAddress] or (1 shl bitOffset)
|
||||
|
||||
proc lookup*(bf: BloomFilter, item: string): bool =
|
||||
## Lookup an item (string) into the Bloom filter.
|
||||
## Lookup an item (string) in the Bloom filter.
|
||||
## If the item is present, ``lookup`` is guaranteed to return ``true``.
|
||||
## If the item is not present, ``lookup`` will return ``false``
|
||||
## with a probability 1 - ``bf.errorRate``.
|
||||
|
||||
@ -1,195 +1,139 @@
|
||||
import unittest
|
||||
import sets
|
||||
import strutils
|
||||
include bloom
|
||||
from random import rand, randomize
|
||||
import times
|
||||
|
||||
suite "murmur":
|
||||
# Test murmurhash 3 when enabled
|
||||
# Test murmurhash3 implementations
|
||||
setup:
|
||||
var hashOutputs: MurmurHashes
|
||||
hashOutputs = [0, 0]
|
||||
rawMurmurHash("hello", 5, 0, hashOutputs)
|
||||
rawMurmurHash128("hello", 5, 0'u32, hashOutputs)
|
||||
|
||||
test "raw":
|
||||
check int(hashOutputs[0]) == -3758069500696749310 # Correct murmur outputs (cast to int64)
|
||||
test "murmur128 raw":
|
||||
check int(hashOutputs[0]) == -3758069500696749310
|
||||
check int(hashOutputs[1]) == 6565844092913065241
|
||||
|
||||
test "wrapped":
|
||||
let hashOutputs2 = murmurHash("hello", 0)
|
||||
test "murmur128 wrapped":
|
||||
let hashOutputs2 = murmurHash128("hello", 0'u32)
|
||||
check hashOutputs2[0] == hashOutputs[0]
|
||||
check hashOutputs2[1] == hashOutputs[1]
|
||||
|
||||
test "seed":
|
||||
let hashOutputs3 = murmurHash("hello", 10)
|
||||
check hashOutputs3[0] != hashOutputs[0]
|
||||
check hashOutputs3[1] != hashOutputs[1]
|
||||
test "murmur32":
|
||||
let hash1 = murmurHash32("hello", 0'u32)
|
||||
let hash2 = murmurHash32("hello", 0'u32)
|
||||
check hash1 == hash2 # Same input should give same output
|
||||
|
||||
let hash3 = murmurHash32("hello", 10'u32)
|
||||
check hash1 != hash3 # Different seeds should give different outputs
|
||||
|
||||
suite "hashing comparison":
|
||||
test "hash distribution":
|
||||
const testSize = 10000
|
||||
var standardCollisions = 0
|
||||
var extendedCollisions = 0
|
||||
suite "hash quality":
|
||||
test "hash type selection":
|
||||
let bfMurmur128 = initializeBloomFilter(100, 0.01, hashType = htMurmur128)
|
||||
let bfMurmur32 = initializeBloomFilter(100, 0.01, hashType = htMurmur32)
|
||||
let bfNimHash = initializeBloomFilter(100, 0.01, hashType = htNimHash)
|
||||
|
||||
var bfStandard = initializeBloomFilter(testSize, 0.01, useExtendedHash = false)
|
||||
var bfExtended = initializeBloomFilter(testSize, 0.01, useExtendedHash = true)
|
||||
|
||||
# Generate test data
|
||||
var testData = newSeq[string](testSize)
|
||||
for i in 0..<testSize:
|
||||
testData[i] = $i & "salt" & $rand(1000000)
|
||||
|
||||
# Test standard hash
|
||||
var startTime = cpuTime()
|
||||
for item in testData:
|
||||
bfStandard.insert(item)
|
||||
let standardTime = cpuTime() - startTime
|
||||
|
||||
# Test extended hash
|
||||
startTime = cpuTime()
|
||||
for item in testData:
|
||||
bfExtended.insert(item)
|
||||
let extendedTime = cpuTime() - startTime
|
||||
|
||||
echo "Standard hash time: ", standardTime
|
||||
echo "Extended hash time: ", extendedTime
|
||||
check bfMurmur128.hashType == htMurmur128
|
||||
check bfMurmur32.hashType == htMurmur32
|
||||
check bfNimHash.hashType == htNimHash
|
||||
|
||||
test "hash implementation switch":
|
||||
# Create two filters with different hash implementations
|
||||
let standardBf = initializeBloomFilter(1000, 0.01, useExtendedHash = false)
|
||||
let murmurBf = initializeBloomFilter(1000, 0.01, useExtendedHash = true)
|
||||
test "quality across hash types":
|
||||
const testSize = 10_000
|
||||
let patterns = @[
|
||||
"shortstr",
|
||||
repeat("a", 1000), # Very long string
|
||||
"special@#$%^&*()", # Special characters
|
||||
"unicode→★∑≈", # Unicode characters
|
||||
repeat("pattern", 10) # Repeating pattern
|
||||
]
|
||||
|
||||
# Insert same elements
|
||||
let testData = ["test1", "test2", "test3", "test4", "test5"]
|
||||
for item in testData:
|
||||
var stdBf = standardBf # Create mutable copies
|
||||
var murBf = murmurBf
|
||||
stdBf.insert(item)
|
||||
murBf.insert(item)
|
||||
for hashType in [htMurmur128, htMurmur32, htNimHash]:
|
||||
var bf = initializeBloomFilter(testSize, 0.01, hashType = hashType)
|
||||
var inserted = newSeq[string](testSize)
|
||||
|
||||
# Verify both can find their items
|
||||
check stdBf.lookup(item)
|
||||
check murBf.lookup(item)
|
||||
|
||||
# Verify false positives work as expected for both
|
||||
let nonExistentItem = "definitely-not-in-filter"
|
||||
var falsePositiveStd = standardBf.lookup(nonExistentItem)
|
||||
var falsePositiveMur = murmurBf.lookup(nonExistentItem)
|
||||
|
||||
# Both should maintain their error rates
|
||||
# Run multiple times to get a sample
|
||||
var fpCountStd = 0
|
||||
var fpCountMur = 0
|
||||
for i in 0..1000:
|
||||
let testItem = "test-" & $i
|
||||
if standardBf.lookup(testItem): fpCountStd += 1
|
||||
if murmurBf.lookup(testItem): fpCountMur += 1
|
||||
|
||||
# Both should have similar false positive rates within reasonable bounds
|
||||
let fpRateStd = fpCountStd.float / 1000.0
|
||||
let fpRateMur = fpCountMur.float / 1000.0
|
||||
|
||||
check abs(fpRateStd - fpRateMur) < 0.01 # Should be reasonably close
|
||||
check fpRateStd < standardBf.errorRate * 1.5 # Should not exceed target error rate by too much
|
||||
check fpRateMur < murmurBf.errorRate * 1.5
|
||||
# Test pattern handling
|
||||
for pattern in patterns:
|
||||
bf.insert(pattern)
|
||||
check bf.lookup(pattern)
|
||||
|
||||
# Test general insertion and lookup
|
||||
for i in 0..<testSize:
|
||||
inserted[i] = $i & "test" & $rand(1000)
|
||||
bf.insert(inserted[i])
|
||||
|
||||
# Verify all insertions
|
||||
var lookupErrors = 0
|
||||
for item in inserted:
|
||||
if not bf.lookup(item):
|
||||
lookupErrors.inc
|
||||
check lookupErrors == 0
|
||||
|
||||
# Check false positive rate
|
||||
var falsePositives = 0
|
||||
let fpTestSize = testSize div 2
|
||||
for i in 0..<fpTestSize:
|
||||
let testItem = "notpresent" & $i & $rand(1000)
|
||||
if bf.lookup(testItem):
|
||||
falsePositives.inc
|
||||
|
||||
let fpRate = falsePositives.float / fpTestSize.float
|
||||
check fpRate < bf.errorRate * 1.5 # Allow some margin but should be close to target
|
||||
|
||||
echo "Standard hash false positive rate: ", fpRateStd
|
||||
echo "Murmur hash false positive rate: ", fpRateMur
|
||||
|
||||
suite "bloom":
|
||||
suite "bloom filter":
|
||||
setup:
|
||||
let nElementsToTest = 100000
|
||||
let nElementsToTest = 10000
|
||||
var bf = initializeBloomFilter(capacity = nElementsToTest, errorRate = 0.001)
|
||||
randomize(2882) # Seed the RNG
|
||||
var
|
||||
sampleChars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789"
|
||||
kTestElements = newSeq[string](nElementsToTest)
|
||||
testElements = newSeq[string](nElementsToTest)
|
||||
|
||||
for i in 0..<nElementsToTest:
|
||||
var newString = ""
|
||||
for j in 0..7:
|
||||
newString.add(sampleChars[rand(51)])
|
||||
kTestElements[i] = newString
|
||||
testElements[i] = newString
|
||||
|
||||
for i in 0..<nElementsToTest:
|
||||
bf.insert(kTestElements[i])
|
||||
for item in testElements:
|
||||
bf.insert(item)
|
||||
|
||||
test "init parameters":
|
||||
check(bf.capacity == nElementsToTest)
|
||||
check(bf.errorRate == 0.001)
|
||||
check(bf.kHashes == 10)
|
||||
check(bf.mBits == 15 * nElementsToTest)
|
||||
|
||||
test "hash mode selection":
|
||||
let bf1 = initializeBloomFilter(100, 0.01)
|
||||
check(bf1.useExtendedHash == false)
|
||||
|
||||
let bf2 = initializeBloomFilter(100, 0.01, useExtendedHash = true)
|
||||
check(bf2.useExtendedHash == true)
|
||||
test "initialization parameters":
|
||||
check bf.capacity == nElementsToTest
|
||||
check bf.errorRate == 0.001
|
||||
check bf.kHashes == 10
|
||||
check bf.mBits div bf.capacity == 15 # bits per element
|
||||
|
||||
test "basic operations":
|
||||
# Test empty lookup
|
||||
check(bf.lookup("nothing") == false)
|
||||
check bf.lookup("nonexistent") == false # Test empty lookup
|
||||
|
||||
# Test insert and lookup
|
||||
bf.insert("teststring")
|
||||
check(bf.lookup("teststring") == true)
|
||||
|
||||
# Test multiple inserts
|
||||
bf.insert("test1")
|
||||
bf.insert("test2")
|
||||
check(bf.lookup("test1") == true)
|
||||
check(bf.lookup("test2") == true)
|
||||
check(bf.lookup("test3") == false)
|
||||
|
||||
test "large scale performance":
|
||||
let largeSize = 1_000_000
|
||||
var standardBf = initializeBloomFilter(largeSize, 0.001, useExtendedHash = false)
|
||||
var extendedBf = initializeBloomFilter(largeSize, 0.001, useExtendedHash = true)
|
||||
|
||||
var largeData = newSeq[string](1000)
|
||||
for i in 0..<1000:
|
||||
largeData[i] = $i & "test" & $rand(1000000)
|
||||
|
||||
# Insert and measure false positives for both
|
||||
var startTime = cpuTime()
|
||||
for item in largeData:
|
||||
standardBf.insert(item)
|
||||
let standardTime = cpuTime() - startTime
|
||||
|
||||
startTime = cpuTime()
|
||||
for item in largeData:
|
||||
extendedBf.insert(item)
|
||||
let extendedTime = cpuTime() - startTime
|
||||
|
||||
echo "Standard hash large insert time: ", standardTime
|
||||
echo "Extended hash large insert time: ", extendedTime
|
||||
var bf2 = initializeBloomFilter(100, 0.01)
|
||||
bf2.insert("test string")
|
||||
check bf2.lookup("test string") == true
|
||||
check bf2.lookup("different string") == false
|
||||
|
||||
test "error rate":
|
||||
var falsePositives = 0
|
||||
for i in 0..<nElementsToTest:
|
||||
var falsePositiveString = ""
|
||||
for j in 0..8:
|
||||
falsePositiveString.add(sampleChars[rand(51)])
|
||||
if bf.lookup(falsePositiveString):
|
||||
falsePositives += 1
|
||||
let testSize = nElementsToTest div 2
|
||||
for i in 0..<testSize:
|
||||
var testString = ""
|
||||
for j in 0..8: # Different length than setup
|
||||
testString.add(sampleChars[rand(51)])
|
||||
if bf.lookup(testString):
|
||||
falsePositives.inc
|
||||
|
||||
let actualErrorRate = falsePositives.float / nElementsToTest.float
|
||||
check actualErrorRate < bf.errorRate
|
||||
echo "Actual error rate: ", actualErrorRate
|
||||
echo "Target error rate: ", bf.errorRate
|
||||
|
||||
test "lookup reliability":
|
||||
let actualErrorRate = falsePositives.float / testSize.float
|
||||
check actualErrorRate < bf.errorRate * 2.0 # Allow some margin
|
||||
|
||||
test "perfect recall":
|
||||
var lookupErrors = 0
|
||||
let startTime = cpuTime()
|
||||
for i in 0..<nElementsToTest:
|
||||
if not bf.lookup(kTestElements[i]):
|
||||
lookupErrors += 1
|
||||
let endTime = cpuTime()
|
||||
|
||||
for item in testElements:
|
||||
if not bf.lookup(item):
|
||||
lookupErrors.inc
|
||||
check lookupErrors == 0
|
||||
echo "Lookup time for ", nElementsToTest, " items: ", formatFloat(endTime - startTime, format = ffDecimal, precision = 4), " seconds"
|
||||
|
||||
test "k/(m/n) specification":
|
||||
test "k/m bits specification":
|
||||
expect(BloomFilterError):
|
||||
discard getMOverNBitsForK(k = 2, targetError = 0.00001)
|
||||
|
||||
@ -197,23 +141,13 @@ suite "bloom":
|
||||
check getMOverNBitsForK(k = 7, targetError = 0.01) == 10
|
||||
check getMOverNBitsForK(k = 7, targetError = 0.001) == 16
|
||||
|
||||
test "force params":
|
||||
var bf2 = initializeBloomFilter(10000, 0.001, k = 4, forceNBitsPerElem = 20)
|
||||
check(bf2.capacity == 10000)
|
||||
check(bf2.errorRate == 0.001)
|
||||
check(bf2.kHashes == 4)
|
||||
check(bf2.mBits == 200000)
|
||||
|
||||
test "init error cases":
|
||||
expect(BloomFilterError):
|
||||
discard initializeBloomFilter(1000, 0.00001, k = 2)
|
||||
|
||||
expect(BloomFilterError):
|
||||
discard initializeBloomFilter(1000, 0.00001, k = 13)
|
||||
check bf2.kHashes == 4
|
||||
check bf2.mBits == 200000
|
||||
|
||||
test "string representation":
|
||||
let bf3 = initializeBloomFilter(1000, 0.01, k = 4)
|
||||
let str = $bf3
|
||||
check str.contains("1000")
|
||||
check str.contains("4 hash functions")
|
||||
check str.contains("1.0e-02") # 0.01 in scientific notation
|
||||
check str.contains("1000") # Capacity
|
||||
check str.contains("4 hash") # Hash functions
|
||||
check str.contains("1.0e-02") # Error rate in scientific notation
|
||||
Loading…
x
Reference in New Issue
Block a user