benchmark_latest

Jump to bottom

github-actions[bot] edited this page Aug 26, 2026 · 28 revisions

Latest known benchmark result, per method

Generated. Do not edit. Produced by .github/workflows/bench-nightly.yml from the wiki's own history, alongside nightly_run.

Not a comparison across methods. Each section below is the last night that method was actually re-run -- whichever night touched the source near it, not necessarily last night, and not the same night as its neighbours here. Every run measures on a GitHub hosted runner whose hardware differs night to night, so a number here says "this is the last known reading", never "faster than the section above it".

Per method

Lodestar.Text.Benchmarks.BatchEmbeddingBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	CorpusSize	Mean	Error	StdDev	Ratio	Gen0	Gen1	Allocated	Alloc Ratio
UnitLoop	1	5.715 μs	0.2390 μs	0.0131 μs	1.00	0.1678	-	2.76 KB	1.00
EmbedBatch	1	5.834 μs	0.9573 μs	0.0525 μs	1.02	0.1907	-	3.16 KB	1.14
EmbedBatchBucketed	1	5.920 μs	0.1700 μs	0.0093 μs	1.04	0.1907	-	3.16 KB	1.14

UnitLoop	8	102.109 μs	8.8185 μs	0.4834 μs	1.00	8.1787	0.2441	134.29 KB	1.00
EmbedBatch	8	70.076 μs	6.4254 μs	0.3522 μs	0.69	7.6904	0.3662	127.31 KB	0.95
EmbedBatchBucketed	8	67.057 μs	4.9788 μs	0.2729 μs	0.66	7.6904	0.3662	127.31 KB	0.95

UnitLoop	32	370.686 μs	23.9348 μs	1.3119 μs	1.00	28.3203	0.9766	468.71 KB	1.00
EmbedBatch	32	241.612 μs	16.1958 μs	0.8877 μs	0.65	26.8555	1.7090	441.32 KB	0.94
EmbedBatchBucketed	32	226.319 μs	13.2666 μs	0.7272 μs	0.61	26.1230	1.7090	427.8 KB	0.91

UnitLoop	128	1,479.332 μs	196.8673 μs	10.7910 μs	1.00	113.2813	5.8594	1874.78 KB	1.00
EmbedBatch	128	928.204 μs	88.8204 μs	4.8685 μs	0.63	107.4219	15.6250	1764.42 KB	0.94
EmbedBatchBucketed	128	870.741 μs	63.0107 μs	3.4538 μs	0.59	103.5156	15.6250	1696.9 KB	0.91

Lodestar.Text.Benchmarks.BlockedTableBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	length	Mean	Error	StdDev	Allocated
Latin	1000	57.89 μs	1.368 μs	0.075 μs	-
Cjk	1000	63.86 μs	1.236 μs	0.068 μs	-
Latin	10000	5,560.06 μs	203.463 μs	11.152 μs	-
Cjk	10000	7,132.48 μs	282.587 μs	15.490 μs	-
Latin	65536	228,821.18 μs	27,374.117 μs	1,500.468 μs	-

Lodestar.Text.Benchmarks.BpeBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Ratio	Gen0	Allocated	Alloc Ratio
Unigram	599.1 ms	45.35 ms	2.49 ms	1.00	32000.0000	519.51 MB	1.00
Bpe	567.6 ms	35.23 ms	1.93 ms	0.95	7000.0000	112.18 MB	0.22

Lodestar.Text.Benchmarks.BpeScalingBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Length	Mean	Error	StdDev	Gen0	Allocated
BpeOnOnePathologicalToken	512	103.5 μs	3.85 μs	0.21 μs	1.2207	20.38 KB
BpeOnOnePathologicalToken	1024	211.0 μs	7.91 μs	0.43 μs	2.4414	39.93 KB
BpeOnOnePathologicalToken	2048	490.5 μs	57.84 μs	3.17 μs	3.9063	78.98 KB
BpeOnOnePathologicalToken	4096	1,026.8 μs	59.38 μs	3.25 μs	7.8125	157.03 KB

Lodestar.Text.Benchmarks.BucketRouteDiagnostics-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Alphabet	Mean	Error	StdDev	Allocated
DpGroup	cjk	19.01 μs	1.883 μs	0.103 μs	-
MyersGroup	cjk	169.25 μs	1.589 μs	0.087 μs	-
DpGroup	latin	10.65 μs	0.050 μs	0.003 μs	-
MyersGroup	latin	111.64 μs	0.207 μs	0.011 μs	-

Lodestar.Text.Benchmarks.FuzzBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Allocated	Alloc Ratio
Ratio	110.2 ns	2.44 ns	0.13 ns	1.00	0.00	-	-	NA
PartialRatio	13,442.7 ns	248.97 ns	13.65 ns	122.02	0.17	-	-	NA
TokenSortRatio	942.3 ns	144.02 ns	7.89 ns	8.55	0.06	0.0782	1312 B	NA
TokenSetRatio	3,247.3 ns	508.35 ns	27.86 ns	29.48	0.22	0.3433	5760 B	NA
WRatio	4,581.7 ns	1,138.23 ns	62.39 ns	41.59	0.49	0.4272	7200 B	NA

Lodestar.Text.Benchmarks.IndelBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Length	Mean	Error	StdDev	Ratio	RatioSD	Allocated	Alloc Ratio
Distance_Utf16	8	30.66 ns	0.782 ns	0.043 ns	1.00	0.00	-	NA
Distance_CodePoint	8	132.26 ns	0.866 ns	0.047 ns	4.31	0.01	-	NA
NormalizedSimilarity_Utf16	8	30.35 ns	0.553 ns	0.030 ns	0.99	0.00	-	NA
SubsequenceLength_Utf16	8	28.33 ns	1.256 ns	0.069 ns	0.92	0.00	-	NA

Distance_Utf16	12	32.51 ns	0.712 ns	0.039 ns	1.00	0.00	-	NA
Distance_CodePoint	12	137.20 ns	1.732 ns	0.095 ns	4.22	0.01	-	NA
NormalizedSimilarity_Utf16	12	33.47 ns	0.854 ns	0.047 ns	1.03	0.00	-	NA
SubsequenceLength_Utf16	12	33.28 ns	0.825 ns	0.045 ns	1.02	0.00	-	NA

Distance_Utf16	16	36.11 ns	0.960 ns	0.053 ns	1.00	0.00	-	NA
Distance_CodePoint	16	146.88 ns	13.243 ns	0.726 ns	4.07	0.02	-	NA
NormalizedSimilarity_Utf16	16	36.84 ns	0.078 ns	0.004 ns	1.02	0.00	-	NA
SubsequenceLength_Utf16	16	35.24 ns	1.241 ns	0.068 ns	0.98	0.00	-	NA

Distance_Utf16	20	38.90 ns	0.685 ns	0.038 ns	1.00	0.00	-	NA
Distance_CodePoint	20	187.75 ns	5.034 ns	0.276 ns	4.83	0.01	-	NA
NormalizedSimilarity_Utf16	20	38.63 ns	0.372 ns	0.020 ns	0.99	0.00	-	NA
SubsequenceLength_Utf16	20	35.05 ns	2.321 ns	0.127 ns	0.90	0.00	-	NA

Distance_Utf16	24	56.75 ns	0.891 ns	0.049 ns	1.00	0.00	-	NA
Distance_CodePoint	24	843.67 ns	234.785 ns	12.869 ns	14.87	0.20	-	NA
NormalizedSimilarity_Utf16	24	60.87 ns	0.947 ns	0.052 ns	1.07	0.00	-	NA
SubsequenceLength_Utf16	24	60.29 ns	4.616 ns	0.253 ns	1.06	0.00	-	NA

Distance_Utf16	32	69.04 ns	3.522 ns	0.193 ns	1.00	0.00	-	NA
Distance_CodePoint	32	1,141.48 ns	1,300.638 ns	71.292 ns	16.53	0.90	-	NA
NormalizedSimilarity_Utf16	32	75.13 ns	4.597 ns	0.252 ns	1.09	0.00	-	NA
SubsequenceLength_Utf16	32	65.48 ns	1.491 ns	0.082 ns	0.95	0.00	-	NA

Distance_Utf16	128	875.60 ns	10.179 ns	0.558 ns	1.00	0.00	-	NA
Distance_CodePoint	128	21,160.69 ns	347.192 ns	19.031 ns	24.17	0.02	-	NA
NormalizedSimilarity_Utf16	128	876.07 ns	10.071 ns	0.552 ns	1.00	0.00	-	NA
SubsequenceLength_Utf16	128	874.77 ns	36.457 ns	1.998 ns	1.00	0.00	-	NA

Distance_Utf16	512	8,059.47 ns	25.449 ns	1.395 ns	1.00	0.00	-	NA
Distance_CodePoint	512	307,455.03 ns	53,441.184 ns	2,929.292 ns	38.15	0.31	-	NA
NormalizedSimilarity_Utf16	512	8,126.18 ns	5.873 ns	0.322 ns	1.01	0.00	-	NA
SubsequenceLength_Utf16	512	8,268.05 ns	219.602 ns	12.037 ns	1.03	0.00	-	NA

Lodestar.Text.Benchmarks.LcsGateBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Band	Mean	Error	StdDev	Ratio	RatioSD	Allocated	Alloc Ratio
Dp	8	130.03 ns	1.645 ns	0.090 ns	1.00	0.00	-	NA
Kernel	8	62.21 ns	0.706 ns	0.039 ns	0.48	0.00	-	NA
Dp_Cjk	8	131.38 ns	1.526 ns	0.084 ns	1.01	0.00	-	NA
Kernel_Cjk	8	106.86 ns	4.763 ns	0.261 ns	0.82	0.00	-	NA

Dp	12	228.88 ns	263.103 ns	14.422 ns	1.00	0.08	-	NA
Kernel	12	69.58 ns	2.101 ns	0.115 ns	0.30	0.02	-	NA
Dp_Cjk	12	222.85 ns	23.981 ns	1.314 ns	0.98	0.05	-	NA
Kernel_Cjk	12	118.96 ns	10.672 ns	0.585 ns	0.52	0.03	-	NA

Dp	14	301.99 ns	421.559 ns	23.107 ns	1.00	0.10	-	NA
Kernel	14	74.33 ns	2.256 ns	0.124 ns	0.25	0.02	-	NA
Dp_Cjk	14	310.30 ns	257.836 ns	14.133 ns	1.03	0.08	-	NA
Kernel_Cjk	14	124.62 ns	1.754 ns	0.096 ns	0.41	0.03	-	NA

Dp	16	388.24 ns	438.564 ns	24.039 ns	1.00	0.08	-	NA
Kernel	16	77.51 ns	1.032 ns	0.057 ns	0.20	0.01	-	NA
Dp_Cjk	16	385.92 ns	587.254 ns	32.189 ns	1.00	0.09	-	NA
Kernel_Cjk	16	132.04 ns	12.225 ns	0.670 ns	0.34	0.02	-	NA

Dp	18	736.76 ns	707.736 ns	38.793 ns	1.00	0.07	-	NA
Kernel	18	84.23 ns	2.239 ns	0.123 ns	0.11	0.01	-	NA
Dp_Cjk	18	703.36 ns	319.178 ns	17.495 ns	0.96	0.05	-	NA
Kernel_Cjk	18	136.16 ns	17.261 ns	0.946 ns	0.19	0.01	-	NA

Dp	20	822.71 ns	130.715 ns	7.165 ns	1.00	0.01	-	NA
Kernel	20	87.30 ns	2.288 ns	0.125 ns	0.11	0.00	-	NA
Dp_Cjk	20	825.31 ns	43.357 ns	2.377 ns	1.00	0.01	-	NA
Kernel_Cjk	20	141.20 ns	1.313 ns	0.072 ns	0.17	0.00	-	NA

Dp	24	994.10 ns	4.037 ns	0.221 ns	1.00	0.00	-	NA
Kernel	24	96.78 ns	1.192 ns	0.065 ns	0.10	0.00	-	NA
Dp_Cjk	24	996.96 ns	127.000 ns	6.961 ns	1.00	0.01	-	NA
Kernel_Cjk	24	152.87 ns	1.806 ns	0.099 ns	0.15	0.00	-	NA

Dp	32	1,645.81 ns	54.450 ns	2.985 ns	1.00	0.00	-	NA
Kernel	32	115.41 ns	2.331 ns	0.128 ns	0.07	0.00	-	NA
Dp_Cjk	32	1,641.21 ns	752.876 ns	41.268 ns	1.00	0.02	-	NA
Kernel_Cjk	32	179.32 ns	2.340 ns	0.128 ns	0.11	0.00	-	NA

Dp	48	3,278.12 ns	366.328 ns	20.080 ns	1.00	0.01	-	NA
Kernel	48	144.48 ns	2.064 ns	0.113 ns	0.04	0.00	-	NA
Dp_Cjk	48	3,291.89 ns	1,013.427 ns	55.549 ns	1.00	0.02	-	NA
Kernel_Cjk	48	322.82 ns	957.700 ns	52.495 ns	0.10	0.01	-	NA

Dp	64	5,376.79 ns	970.348 ns	53.188 ns	1.00	0.01	-	NA
Kernel	64	173.02 ns	47.951 ns	2.628 ns	0.03	0.00	-	NA
Dp_Cjk	64	5,133.25 ns	168.402 ns	9.231 ns	0.95	0.01	-	NA
Kernel_Cjk	64	314.60 ns	3.706 ns	0.203 ns	0.06	0.00	-	NA

Dp	96	11,520.78 ns	6,303.697 ns	345.527 ns	1.00	0.04	-	NA
Kernel	96	736.08 ns	5.229 ns	0.287 ns	0.06	0.00	-	NA
Dp_Cjk	96	11,302.81 ns	4,067.584 ns	222.958 ns	0.98	0.03	-	NA
Kernel_Cjk	96	1,080.21 ns	7.670 ns	0.420 ns	0.09	0.00	-	NA

Lodestar.Text.Benchmarks.LevenshteinBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Length	Mean	Error	StdDev	Ratio	RatioSD	Allocated	Alloc Ratio
Distance_Utf16	8	27.97 ns	0.572 ns	0.031 ns	1.00	0.00	-	NA
Distance_CodePoint	8	125.70 ns	18.999 ns	1.041 ns	4.49	0.03	-	NA
NormalizedSimilarity_Utf16	8	28.25 ns	5.317 ns	0.291 ns	1.01	0.01	-	NA

Distance_Utf16	64	312.33 ns	3.984 ns	0.218 ns	1.00	0.00	-	NA
Distance_CodePoint	64	711.65 ns	64.585 ns	3.540 ns	2.28	0.01	-	NA
NormalizedSimilarity_Utf16	64	313.36 ns	4.843 ns	0.265 ns	1.00	0.00	-	NA

Distance_Utf16	512	15,532.55 ns	112.370 ns	6.159 ns	1.00	0.00	-	NA
Distance_CodePoint	512	18,692.32 ns	1,858.174 ns	101.853 ns	1.20	0.01	-	NA
NormalizedSimilarity_Utf16	512	15,413.86 ns	169.538 ns	9.293 ns	0.99	0.00	-	NA

Lodestar.Text.Benchmarks.LevenshteinCodePointBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Length	Distinct	Mean	Error	StdDev	Ratio	RatioSD	Allocated	Alloc Ratio
Distance_CodePoint	16	32	354.2 ns	4.39 ns	0.24 ns	1.00	0.00	-	NA
Distance_Utf16	16	32	255.7 ns	6.48 ns	0.36 ns	0.72	0.00	-	NA

Distance_CodePoint	16	512	354.6 ns	4.68 ns	0.26 ns	1.00	0.00	-	NA
Distance_Utf16	16	512	255.9 ns	2.48 ns	0.14 ns	0.72	0.00	-	NA

Distance_CodePoint	24	32	474.8 ns	9.96 ns	0.55 ns	1.00	0.00	-	NA
Distance_Utf16	24	32	349.6 ns	3.40 ns	0.19 ns	0.74	0.00	-	NA

Distance_CodePoint	24	512	446.7 ns	17.84 ns	0.98 ns	1.00	0.00	-	NA
Distance_Utf16	24	512	346.6 ns	0.81 ns	0.04 ns	0.78	0.00	-	NA

Distance_CodePoint	32	32	547.9 ns	96.74 ns	5.30 ns	1.00	0.01	-	NA
Distance_Utf16	32	32	444.6 ns	26.39 ns	1.45 ns	0.81	0.01	-	NA

Distance_CodePoint	32	512	536.6 ns	12.32 ns	0.68 ns	1.00	0.00	-	NA
Distance_Utf16	32	512	452.4 ns	101.23 ns	5.55 ns	0.84	0.01	-	NA

Distance_CodePoint	40	32	637.8 ns	4.68 ns	0.26 ns	1.00	0.00	-	NA
Distance_Utf16	40	32	1,339.8 ns	57.00 ns	3.12 ns	2.10	0.00	-	NA

Distance_CodePoint	40	512	659.6 ns	345.12 ns	18.92 ns	1.00	0.04	-	NA
Distance_Utf16	40	512	1,311.8 ns	15.32 ns	0.84 ns	1.99	0.05	-	NA

Distance_CodePoint	128	32	2,581.3 ns	416.70 ns	22.84 ns	1.00	0.01	-	NA
Distance_Utf16	128	32	5,788.4 ns	260.26 ns	14.27 ns	2.24	0.02	-	NA

Distance_CodePoint	128	512	2,583.8 ns	42.22 ns	2.31 ns	1.00	0.00	-	NA
Distance_Utf16	128	512	5,773.0 ns	190.99 ns	10.47 ns	2.23	0.00	-	NA

Distance_CodePoint	512	32	19,748.3 ns	439.07 ns	24.07 ns	1.00	0.00	-	NA
Distance_Utf16	512	32	65,252.5 ns	1,445.21 ns	79.22 ns	3.30	0.00	-	NA

Distance_CodePoint	512	512	434,101.8 ns	67,313.58 ns	3,689.68 ns	1.00	0.01	-	NA
Distance_Utf16	512	512	69,178.6 ns	864.36 ns	47.38 ns	0.16	0.00	-	NA

Lodestar.Text.Benchmarks.MetricsBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Samples	Classes	Mean	Error	StdDev	Gen0	Allocated
Matrix	1000	2	6.992 μs	1.4504 μs	0.0795 μs	0.0153	312 B
MatrixWeighted	1000	2	6.922 μs	0.2477 μs	0.0136 μs	0.0153	312 B
AccuracyScore	1000	2	1.033 μs	0.0791 μs	0.0043 μs	-	-
F1Macro	1000	2	6.944 μs	0.2435 μs	0.0133 μs	0.0229	472 B
Report	1000	2	9.772 μs	0.0972 μs	0.0053 μs	0.3815	6520 B
Matrix	1000	10	7.222 μs	0.1891 μs	0.0104 μs	0.0687	1248 B
MatrixWeighted	1000	10	7.257 μs	0.2306 μs	0.0126 μs	0.0687	1248 B
AccuracyScore	1000	10	1.030 μs	0.0041 μs	0.0002 μs	-	-
F1Macro	1000	10	7.705 μs	0.2471 μs	0.0135 μs	0.0916	1664 B
Report	1000	10	14.107 μs	0.1306 μs	0.0072 μs	0.9155	15496 B
Matrix	100000	2	854.652 μs	89.7319 μs	4.9185 μs	-	313 B
MatrixWeighted	100000	2	820.976 μs	127.5164 μs	6.9896 μs	-	313 B
AccuracyScore	100000	2	134.878 μs	33.3763 μs	1.8295 μs	-	-
F1Macro	100000	2	821.090 μs	189.4737 μs	10.3857 μs	-	473 B
Report	100000	2	791.617 μs	53.6832 μs	2.9426 μs	-	6545 B
Matrix	100000	10	951.857 μs	132.1174 μs	7.2418 μs	-	1249 B
MatrixWeighted	100000	10	965.790 μs	242.8810 μs	13.3131 μs	-	1249 B
AccuracyScore	100000	10	231.060 μs	293.5387 μs	16.0898 μs	-	-
F1Macro	100000	10	977.732 μs	36.7259 μs	2.0131 μs	-	1665 B
Report	100000	10	971.064 μs	157.6669 μs	8.6423 μs	-	15841 B
Matrix	1000000	2	8,731.954 μs	1,373.8078 μs	75.3030 μs	-	324 B
MatrixWeighted	1000000	2	8,964.577 μs	799.4884 μs	43.8227 μs	-	324 B
AccuracyScore	1000000	2	1,916.196 μs	53.4906 μs	2.9320 μs	-	-
F1Macro	1000000	2	8,476.928 μs	241.8293 μs	13.2555 μs	-	484 B
Report	1000000	2	8,789.936 μs	100.2981 μs	5.4977 μs	-	6572 B
Matrix	1000000	10	10,095.983 μs	972.8520 μs	53.3253 μs	-	1260 B
MatrixWeighted	1000000	10	10,157.126 μs	678.6928 μs	37.2014 μs	-	1255 B
AccuracyScore	1000000	10	3,110.696 μs	40.6531 μs	2.2283 μs	-	-
F1Macro	1000000	10	9,985.869 μs	98.7127 μs	5.4108 μs	-	1676 B
Report	1000000	10	9,929.414 μs	222.3812 μs	12.1895 μs	-	15892 B

Lodestar.Text.Benchmarks.MyersGateBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Band	Mean	Error	StdDev	Ratio	RatioSD	Allocated	Alloc Ratio
Dp	4	81.89 ns	1.326 ns	0.073 ns	1.00	0.00	-	NA
Kernel	4	80.97 ns	1.432 ns	0.078 ns	0.99	0.00	-	NA
Dp_Cjk	4	81.76 ns	3.229 ns	0.177 ns	1.00	0.00	-	NA
Kernel_Cjk	4	81.60 ns	0.581 ns	0.032 ns	1.00	0.00	-	NA

Dp	6	120.96 ns	7.269 ns	0.398 ns	1.00	0.00	-	NA
Kernel	6	86.52 ns	0.818 ns	0.045 ns	0.72	0.00	-	NA
Dp_Cjk	6	120.53 ns	4.682 ns	0.257 ns	1.00	0.00	-	NA
Kernel_Cjk	6	165.84 ns	8.437 ns	0.462 ns	1.37	0.01	-	NA

Dp	8	157.83 ns	12.843 ns	0.704 ns	1.00	0.01	-	NA
Kernel	8	101.48 ns	15.216 ns	0.834 ns	0.64	0.01	-	NA
Dp_Cjk	8	156.45 ns	8.731 ns	0.479 ns	0.99	0.00	-	NA
Kernel_Cjk	8	195.06 ns	9.856 ns	0.540 ns	1.24	0.01	-	NA

Dp	10	210.55 ns	20.597 ns	1.129 ns	1.00	0.01	-	NA
Kernel	10	107.97 ns	0.910 ns	0.050 ns	0.51	0.00	-	NA
Dp_Cjk	10	208.91 ns	5.480 ns	0.300 ns	0.99	0.00	-	NA
Kernel_Cjk	10	168.48 ns	0.386 ns	0.021 ns	0.80	0.00	-	NA

Dp	12	269.92 ns	2.152 ns	0.118 ns	1.00	0.00	-	NA
Kernel	12	121.10 ns	3.062 ns	0.168 ns	0.45	0.00	-	NA
Dp_Cjk	12	269.65 ns	19.716 ns	1.081 ns	1.00	0.00	-	NA
Kernel_Cjk	12	177.51 ns	1.397 ns	0.077 ns	0.66	0.00	-	NA

Dp	16	427.05 ns	27.782 ns	1.523 ns	1.00	0.00	-	NA
Kernel	16	142.84 ns	1.922 ns	0.105 ns	0.33	0.00	-	NA
Dp_Cjk	16	420.61 ns	11.422 ns	0.626 ns	0.98	0.00	-	NA
Kernel_Cjk	16	200.08 ns	12.666 ns	0.694 ns	0.47	0.00	-	NA

Dp	24	862.54 ns	8.385 ns	0.460 ns	1.00	0.00	-	NA
Kernel	24	181.29 ns	4.116 ns	0.226 ns	0.21	0.00	-	NA
Dp_Cjk	24	864.33 ns	7.882 ns	0.432 ns	1.00	0.00	-	NA
Kernel_Cjk	24	246.08 ns	5.033 ns	0.276 ns	0.29	0.00	-	NA

Dp	32	1,481.14 ns	37.663 ns	2.064 ns	1.00	0.00	-	NA
Kernel	32	219.64 ns	19.510 ns	1.069 ns	0.15	0.00	-	NA
Dp_Cjk	32	1,482.53 ns	35.542 ns	1.948 ns	1.00	0.00	-	NA
Kernel_Cjk	32	296.87 ns	2.654 ns	0.145 ns	0.20	0.00	-	NA

Dp	48	3,445.28 ns	5,843.255 ns	320.289 ns	1.01	0.11	-	NA
Kernel	48	302.06 ns	13.648 ns	0.748 ns	0.09	0.01	-	NA
Dp_Cjk	48	3,243.60 ns	125.317 ns	6.869 ns	0.95	0.07	-	NA
Kernel_Cjk	48	390.99 ns	7.971 ns	0.437 ns	0.11	0.01	-	NA

Dp	64	5,666.65 ns	483.223 ns	26.487 ns	1.00	0.01	-	NA
Kernel	64	380.87 ns	12.894 ns	0.707 ns	0.07	0.00	-	NA
Dp_Cjk	64	5,651.98 ns	211.230 ns	11.578 ns	1.00	0.00	-	NA
Kernel_Cjk	64	496.90 ns	3.711 ns	0.203 ns	0.09	0.00	-	NA

Dp	96	12,548.57 ns	288.869 ns	15.834 ns	1.00	0.00	-	NA
Kernel	96	1,280.89 ns	14.553 ns	0.798 ns	0.10	0.00	-	NA
Dp_Cjk	96	12,511.21 ns	156.476 ns	8.577 ns	1.00	0.00	-	NA
Kernel_Cjk	96	1,558.38 ns	31.192 ns	1.710 ns	0.12	0.00	-	NA

Lodestar.Text.Benchmarks.PersistenceBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
VocabTxt	4.285 ms	4.4435 ms	0.2436 ms	117.1875	109.3750	39.0625	3.62 MB
TokenizerJsonWordPiece	10.789 ms	1.2890 ms	0.0707 ms	156.2500	125.0000	31.2500	5.72 MB
TokenizerJsonUnigram	11.128 ms	0.8632 ms	0.0473 ms	93.7500	78.1250	31.2500	4.64 MB
SpieceModel	3.916 ms	1.2111 ms	0.0664 ms	109.3750	101.5625	31.2500	3.36 MB
TfidfSave	2.018 ms	0.5765 ms	0.0316 ms	33.2031	27.3438	27.3438	2.09 MB
TfidfLoad	4.537 ms	1.0388 ms	0.0569 ms	85.9375	78.1250	23.4375	2.86 MB
EmbeddingIndexSave	4.869 ms	0.4277 ms	0.0234 ms	453.1250	453.1250	453.1250	39.64 MB
EmbeddingIndexLoad	5.306 ms	0.4166 ms	0.0228 ms	500.0000	468.7500	437.5000	35.35 MB

Lodestar.Text.Benchmarks.StopWordBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Documents	Mean	Error	StdDev	Ratio	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Count	200	7.492 ms	0.7557 ms	0.0414 ms	1.00	500.0000	234.3750	62.5000	7.92 MB	1.00
CountWithStopWords	200	6.175 ms	0.6579 ms	0.0361 ms	0.82	390.6250	187.5000	-	6.34 MB	0.80
Hashing	200	7.425 ms	1.4131 ms	0.0775 ms	0.99	507.8125	171.8750	70.3125	8.08 MB	1.02
HashingWithStopWords	200	6.403 ms	0.7136 ms	0.0391 ms	0.85	406.2500	148.4375	-	6.52 MB	0.82

Count	1000	30.227 ms	1.3520 ms	0.0741 ms	1.00	2718.7500	1031.2500	531.2500	38.64 MB	1.00
CountWithStopWords	1000	23.323 ms	3.2614 ms	0.1788 ms	0.77	1968.7500	781.2500	250.0000	30.99 MB	0.80
Hashing	1000	29.070 ms	3.0746 ms	0.1685 ms	0.96	2562.5000	750.0000	593.7500	39.55 MB	1.02
HashingWithStopWords	1000	24.339 ms	2.6587 ms	0.1457 ms	0.81	2031.2500	625.0000	250.0000	31.83 MB	0.82

Lodestar.Text.Benchmarks.VectorMathBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Dim	Mean	Error	StdDev	Ratio	Allocated	Alloc Ratio
Dot	384	60.63 ns	2.443 ns	0.134 ns	1.00	-	NA
L2Norm	384	48.63 ns	1.191 ns	0.065 ns	0.80	-	NA

Dot	768	99.81 ns	1.147 ns	0.063 ns	1.00	-	NA
L2Norm	768	90.73 ns	3.587 ns	0.197 ns	0.91	-	NA

Dot	1024	132.71 ns	2.538 ns	0.139 ns	1.00	-	NA
L2Norm	1024	125.10 ns	1.355 ns	0.074 ns	0.94	-	NA

Lodestar.Text.Benchmarks.VectorizerBenchmarks-report-github

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

BenchmarkDotNet v0.14.0, Ubuntu 24.04.4 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.400
  [Host]   : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.11 (10.0.1126.37416), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Documents	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Count	200	3.011 ms	2.0515 ms	0.1124 ms	1.00	0.05	93.7500	39.0625	-	1.6 MB	1.00
Tfidf	200	3.038 ms	1.9607 ms	0.1075 ms	1.01	0.04	101.5625	39.0625	-	1.63 MB	1.02
CountBigrams	200	3.847 ms	0.5933 ms	0.0325 ms	1.28	0.04	171.8750	109.3750	-	2.78 MB	1.74
Hashing	200	3.037 ms	1.9245 ms	0.1055 ms	1.01	0.04	93.7500	23.4375	-	1.6 MB	1.00

Count	1000	7.058 ms	1.4429 ms	0.0791 ms	1.00	0.01	484.3750	343.7500	62.5000	7.83 MB	1.00
Tfidf	1000	7.239 ms	1.1104 ms	0.0609 ms	1.03	0.01	484.3750	312.5000	93.7500	7.97 MB	1.02
CountBigrams	1000	11.635 ms	0.6589 ms	0.0361 ms	1.65	0.02	906.2500	375.0000	265.6250	13.42 MB	1.71
Hashing	1000	7.093 ms	0.1322 ms	0.0072 ms	1.01	0.01	492.1875	156.2500	70.3125	7.85 MB	1.00

compare-indel

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

Python: rapidfuzz 3.14.5 (py 3.12.14)
C#:     Lodestar.Text on .NET 10.0.11 (mode Utf16Unit)

alphabet	length	Python ns/pair	C# ns/pair	speedup (py/C#)
latin	8	114.0	23.7	4.80x C# faster
latin	32	161.8	67.9	2.38x C# faster
latin	128	478.5	843.4	1.76x Py faster
latin	512	4872.0	8258.7	1.70x Py faster
cjk	8	130.7	23.8	5.50x C# faster
cjk	32	238.1	143.6	1.66x C# faster
cjk	128	2049.6	1663.3	1.23x C# faster
cjk	512	16675.1	11766.7	1.42x C# faster

Note: Indel is len(a)+len(b)-2*LCS on both sides, so this compares the subsequence kernels. Lodestar's is Hyyro's bit-parallel LLCS above a pattern of 8 and a rolling-row dynamic program below it (#273).

compare-levenshtein

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

Python: rapidfuzz 3.14.5 (py 3.12.14)
C#:     Lodestar.Text on .NET 10.0.11 (mode Utf16Unit)

alphabet	length	Python ns/pair	C# ns/pair	speedup (py/C#)
latin	8	143.1	18.3	7.83x C# faster
latin	32	256.1	127.6	2.01x C# faster
latin	128	1815.6	1616.4	1.12x C# faster
latin	512	15667.3	18115.4	1.16x Py faster
cjk	8	142.9	18.1	7.90x C# faster
cjk	32	295.9	190.9	1.55x C# faster
cjk	128	3010.5	2534.8	1.19x C# faster
cjk	512	26207.8	21663.9	1.21x C# faster

Note: Python times the realistic per-call loop; rapidfuzz's C core uses the bit-parallel Myers algorithm, so it scales better on long strings.

compare-metrics

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

Python: {'scikit-learn': '1.9.0', 'numpy': '2.5.1'} (py 3.12.14)
C#:     Lodestar on .NET 10.0.11

operation	C# ms	Py ms	wall	C# cpu	Py cpu	cpu
confusion_matrix_n1000_k2	0.009	0.768	84.44x	0.009	0.767	84.44x
accuracy_n1000_k2	0.001	0.402	382.67x	0.001	0.402	382.64x
precision_recall_f1_macro_n1000_k2	0.007	1.392	186.96x	0.007	1.392	186.96x
classification_report_n1000_k2	0.010	5.243	524.78x	0.010	5.242	524.81x
roc_auc_binary_n1000_k2	0.016	1.545	96.49x	0.016	1.544	96.49x
balanced_accuracy_n1000_k2	0.007	0.825	112.14x	0.007	0.825	112.14x
matthews_n1000_k2	0.007	1.550	210.68x	0.007	1.550	210.67x
cohen_kappa_n1000_k2	0.007	0.868	117.69x	0.007	0.868	117.69x
mse_n1000_k2	0.003	0.214	78.08x	0.003	0.214	78.09x
mae_n1000_k2	0.003	0.214	77.97x	0.003	0.214	77.97x
median_ae_n1000_k2	0.006	0.231	38.81x	0.006	0.231	38.81x
r2_n1000_k2	0.003	0.271	99.12x	0.003	0.271	99.10x
confusion_matrix_n1000_k10	0.009	0.779	83.24x	0.009	0.779	83.24x
accuracy_n1000_k10	0.001	0.406	384.43x	0.001	0.406	384.44x
precision_recall_f1_macro_n1000_k10	0.008	1.414	174.70x	0.008	1.414	174.66x
classification_report_n1000_k10	0.014	5.489	381.16x	0.014	5.488	381.16x
roc_auc_ovr_macro_n1000_k10	0.453	8.097	17.86x	0.453	8.096	17.86x
balanced_accuracy_n1000_k10	0.008	0.833	102.00x	0.008	0.833	102.00x
matthews_n1000_k10	0.008	1.592	199.25x	0.008	1.592	199.25x
cohen_kappa_n1000_k10	0.008	0.880	103.96x	0.008	0.880	103.96x
mse_n1000_k10	0.003	0.216	78.99x	0.003	0.216	78.98x
mae_n1000_k10	0.003	0.215	78.97x	0.003	0.215	78.98x
median_ae_n1000_k10	0.006	0.231	38.83x	0.006	0.231	38.83x
r2_n1000_k10	0.003	0.270	98.86x	0.003	0.270	98.87x
confusion_matrix_n100000_k2	1.022	10.948	10.71x	1.022	10.947	10.71x
accuracy_n100000_k2	0.106	3.775	35.70x	0.106	3.775	35.70x
precision_recall_f1_macro_n100000_k2	0.873	12.524	14.34x	0.873	12.522	14.34x
classification_report_n100000_k2	0.826	26.807	32.45x	0.826	26.801	32.45x
roc_auc_binary_n100000_k2	2.877	28.359	9.86x	2.877	28.356	9.85x
balanced_accuracy_n100000_k2	0.826	11.022	13.35x	0.825	11.020	13.35x
matthews_n100000_k2	0.818	22.057	26.95x	0.818	22.055	26.95x
cohen_kappa_n100000_k2	0.812	11.072	13.63x	0.812	11.070	13.63x
mse_n100000_k2	0.266	0.370	1.39x	0.266	0.370	1.39x
mae_n100000_k2	0.266	0.368	1.38x	0.266	0.368	1.38x
median_ae_n100000_k2	0.685	1.881	2.75x	0.706	1.881	2.66x
r2_n100000_k2	0.260	0.588	2.26x	0.260	0.588	2.26x
confusion_matrix_n100000_k10	1.066	11.031	10.35x	1.066	11.030	10.35x
accuracy_n100000_k10	0.108	3.779	35.07x	0.108	3.778	35.08x
precision_recall_f1_macro_n100000_k10	1.032	13.251	12.85x	1.031	13.249	12.85x
classification_report_n100000_k10	1.006	29.774	29.61x	1.006	29.772	29.61x
roc_auc_ovr_macro_n100000_k10	30.465	230.801	7.58x	30.462	230.780	7.58x
balanced_accuracy_n100000_k10	1.005	11.008	10.95x	1.005	11.007	10.95x
matthews_n100000_k10	0.984	22.817	23.19x	0.984	22.814	23.19x
cohen_kappa_n100000_k10	1.008	11.062	10.98x	1.008	11.061	10.98x
mse_n100000_k10	0.267	0.379	1.42x	0.267	0.378	1.42x
mae_n100000_k10	0.267	0.377	1.41x	0.267	0.377	1.41x
median_ae_n100000_k10	0.713	1.887	2.64x	0.756	1.887	2.49x
r2_n100000_k10	0.261	0.597	2.29x	0.261	0.597	2.29x
confusion_matrix_n1000000_k2	9.228	102.493	11.11x	9.227	102.487	11.11x
accuracy_n1000000_k2	2.307	34.206	14.83x	2.307	34.204	14.83x
precision_recall_f1_macro_n1000000_k2	8.972	112.558	12.54x	8.971	112.547	12.55x
classification_report_n1000000_k2	8.938	219.291	24.54x	8.938	219.250	24.53x
roc_auc_binary_n1000000_k2	41.207	305.260	7.41x	41.202	305.230	7.41x
balanced_accuracy_n1000000_k2	8.977	102.598	11.43x	8.975	102.588	11.43x
matthews_n1000000_k2	8.826	206.686	23.42x	8.826	206.660	23.42x
cohen_kappa_n1000000_k2	8.828	102.787	11.64x	8.827	102.780	11.64x
mse_n1000000_k2	2.656	1.820	0.69x	2.656	1.820	0.69x
mae_n1000000_k2	2.656	1.775	0.67x	2.656	1.775	0.67x
median_ae_n1000000_k2	6.199	15.441	2.49x	6.230	15.440	2.48x
r2_n1000000_k2	2.601	3.432	1.32x	2.600	3.432	1.32x
confusion_matrix_n1000000_k10	11.028	102.276	9.27x	11.027	102.268	9.27x
accuracy_n1000000_k10	3.310	34.147	10.32x	3.310	34.142	10.32x
precision_recall_f1_macro_n1000000_k10	10.541	119.275	11.32x	10.541	119.268	11.31x
classification_report_n1000000_k10	10.428	246.366	23.63x	10.427	246.335	23.62x
balanced_accuracy_n1000000_k10	10.432	102.385	9.81x	10.432	102.371	9.81x
matthews_n1000000_k10	10.157	214.215	21.09x	10.156	214.202	21.09x
cohen_kappa_n1000000_k10	10.166	102.661	10.10x	10.165	102.652	10.10x
mse_n1000000_k10	2.653	1.822	0.69x	2.652	1.822	0.69x
mae_n1000000_k10	2.658	1.797	0.68x	2.658	1.797	0.68x
median_ae_n1000000_k10	6.273	15.496	2.47x	6.343	15.494	2.44x
r2_n1000000_k10	2.776	3.472	1.25x	2.776	3.471	1.25x

ratio > 1 means Lodestar is faster. cpu is the merge gate for this branch (docs/guides/performance.md): every operation, every size, must be >= 1x.

BELOW GATE on processor time: mse_n1000000_k2 0.69x mae_n1000000_k2 0.67x mse_n1000000_k10 0.69x mae_n1000000_k10 0.68x

compare-persistence

As of 2026-08-23, measured at commit 2483f1a00691271083f00baf3835de96bf0a4076.

Python: {'tokenizers': '0.23.1', 'sentencepiece': '0.2.2', 'scikit-learn': '1.9.0', 'numpy': '2.5.1'} (py 3.12.14)
C#:     Lodestar on .NET 10.0.11

operation	C# ms	Py ms	wall	C# cpu	Py cpu	cpu	C# bytes	Py bytes
vocab_txt	5.204	9.890	1.90x	5.462	9.889	1.81x	228,891	228,891
tokenizer_json_wordpiece	12.343	17.594	1.43x	13.144	17.592	1.34x	706,526	706,526
tokenizer_json_unigram	11.317	37.159	3.28x	11.814	37.152	3.14x	1,990,038	1,990,038
spiece_model	4.818	30.384	6.31x	4.981	30.380	6.10x	533,084	533,084
tfidf_save	1.666	2.444	1.47x	1.722	2.444	1.42x	581,787	591,922
tfidf_load	4.704	4.267	0.91x	4.896	4.266	0.87x	581,787	591,922
embedding_index_save	4.670	1.347	0.29x	5.230	1.347	0.26x	20,589,007	15,360,128
embedding_index_load	4.329	1.349	0.31x	4.907	1.348	0.27x	20,589,007	15,360,128
embedding_index_load_file	5.598	0.834	0.15x	6.300	0.833	0.13x	20,589,007	15,360,128
embedding_index_load_memory	3.012	1.399	0.46x	3.407	1.399	0.41x	20,589,007	15,360,128
embedding_index_view_floor	0.000	0.001	76.74x	0.000	0.001	76.80x	20,589,007	15,360,128
embedding_index_save_gzip	454.761	639.466	1.41x	457.526	639.447	1.40x	15,251,458	14,022,374
embedding_index_load_gzip	81.072	72.519	0.89x	82.366	72.512	0.88x	15,251,458	14,022,374

ratio > 1 means Lodestar is faster. cpu is the honest one: elapsed time hides work .NET does on background GC threads; CPython is single-threaded. bytes is what the row wrote or read; a results file from before #378 carries none, and the two columns then disappear rather than read zero.

Lodestar

Project

Clone this wiki locally