Decimal128 by Miguel

Kotlin Benchmark Results

IEEE 754-2019 compliant decimal128 high-performance software solution created by Miguel.

Living document — as-measured results. Category codes, profiles, modes & method: Benchmark Key.

This is the Kotlin view of decimal128 as-measured, band by band, with explicit ratios. It opens with the realistic financial-mix (P-fin) headline, then the per-operation band characterization (P-gen) and FMA. In Kotlin, d128 is measured against its in-language idiom peer BigDecimal, with the libbid universal reference on the full-width bands. It is data only — the categories, magnitude profiles, units, and methodology are defined in the Benchmark Key (and, authoritatively, BenchmarkMatrix.md). The cross-port d128 band-shape matrices (all ports, no alternatives) live in Port-Comparison Benchmark Results; the full index of per-language pages is on the Benchmarks hub.

Summary — Ratio Range by Operation

The ratio for Kotlin’s idiom peer on x86_64 (Intel i9-9880H): ratio = BigDecimal / Miguel (> 1× ⇒ d128 faster), broken out by operation.

  Add Subtract Multiply Divide
ratio = BigDecimal / Miguel 4× 3× 3× 1.8× – 50×

FinMix — realistic financial mix (P-fin)

The headline: one realistic 64-bit financial operation mix — a MIX add/sub stream, mul CP/WP, div CD/WD/ET/PT — ratio = alt / Miguel (> 1 ⇒ d128 faster). This is the profile closest to real financial code.

arm64 (M3 Pro).

port op cat profile arch mode ns alt alt ns ratio run notes
kotlin add MIX P-fin arm64 thru‡ 4.89 BigDecimal 20.24 4.14× Rkosw2 compact idiom peer
kotlin sub MIX P-fin arm64 thru‡ 6.38 BigDecimal 24.81 3.89× Rkosw2 compact idiom peer
kotlin mul CP P-fin arm64 thru‡ 5.40 BigDecimal 12.56 2.33× Rkosw2 compact idiom peer
kotlin mul WP P-fin arm64 thru‡ 30.03 BigDecimal 67.12 2.24× Rkosw2 compact idiom peer
kotlin div CD P-fin arm64 thru‡ 39.46 BigDecimal 140.61 3.56× Rkosw2 compact idiom peer
kotlin div WD P-fin arm64 thru‡ 52.20 BigDecimal 111.89 2.14× Rkosw2 compact idiom peer
kotlin div ET P-fin arm64 thru‡ 18.24 BigDecimal 507.64 27.83× Rkosw2 compact idiom peer
kotlin div PT P-fin arm64 thru‡ 15.15 BigDecimal 488.01 32.21× Rkosw2 compact idiom peer

x86_64 (Intel i9-9880H).

port op cat profile arch mode ns alt alt ns ratio run notes
kotlin add MIX P-fin x86_64 thru‡ 17.01 BigDecimal 68.32 4.02× xRkosw2 compact idiom peer
kotlin sub MIX P-fin x86_64 thru‡ 22.95 BigDecimal 87.69 3.82× xRkosw2 compact idiom peer
kotlin mul CP P-fin x86_64 thru‡ 15.82 BigDecimal 53.95 3.41× xRkosw2 compact idiom peer
kotlin mul WP P-fin x86_64 thru‡ 65.20 BigDecimal 209.30 3.21× xRkosw2 compact idiom peer
kotlin div CD P-fin x86_64 thru‡ 131.31 BigDecimal 527.45 4.02× xRkosw2 compact idiom peer
kotlin div WD P-fin x86_64 thru‡ 162.38 BigDecimal 300.21 1.85× xRkosw2 compact idiom peer
kotlin div ET P-fin x86_64 thru‡ 66.93 BigDecimal 1822.78 27.23× xRkosw2 compact idiom peer
kotlin div PT P-fin x86_64 thru‡ 34.67 BigDecimal 1757.49 50.69× xRkosw2 compact idiom peer

Add — SQ · NQ · MQ · OQ · FQ

Swept 4096-input average per band (bare thru; ns/op = Time/4096 over the shared decimal128-resources/swept/ corpus, byte-identical operands every port). Relational peer table with explicit ratios; the cross-port d128 band-shape matrices are in Port-Comparison Benchmark Results.

arm64 (M3 Pro).

port op cat profile arch mode ours alt alt ns ratio run notes
kotlin add SQss P-gen arm64 thru‡ 4.35 BigDecimal 18.63 4.28× Rkosw2 compact idiom peer
kotlin add SQos P-gen arm64 thru‡ 6.19 BigDecimal 21.41 3.46× Rkosw2 compact idiom peer
kotlin add NQss P-gen arm64 thru‡ 8.88 BigDecimal 29.82 3.36× Rkosw2 compact idiom peer
kotlin add NQos P-gen arm64 thru‡ 9.92 BigDecimal 30.52 3.08× Rkosw2 compact idiom peer
kotlin add MQss P-gen arm64 thru‡ 13.86 BigDecimal 29.88 2.16× Rkosw2 compact idiom peer
kotlin add MQos P-gen arm64 thru‡ 27.28 BigDecimal 30.51 1.12× Rkosw2 compact idiom peer
kotlin add OQss P-gen arm64 thru‡ 24.97 BigDecimal 72.04 2.89× Rkosw2 compact idiom peer
kotlin add OQos P-gen arm64 thru‡ 44.47 BigDecimal 75.38 1.70× Rkosw2 compact idiom peer
kotlin add FQss P-gen arm64 thru‡ 16.52 BigDecimal 75.55 4.57× Rkosw2 compact idiom peer
kotlin add FQos P-gen arm64 thru‡ 28.76 BigDecimal 84.90 2.95× Rkosw2 compact idiom peer

x86_64 (Intel i9-9880H).

port op cat profile arch mode ours alt alt ns ratio run notes
kotlin add SQss P-gen x86_64 thru‡ 14.73 BigDecimal 63.75 4.33× xRkosw2 compact idiom peer
kotlin add SQos P-gen x86_64 thru‡ 23.50 BigDecimal 75.59 3.22× xRkosw2 compact idiom peer
kotlin add NQss P-gen x86_64 thru‡ 24.56 BigDecimal 102.39 4.17× xRkosw2 compact idiom peer
kotlin add NQos P-gen x86_64 thru‡ 32.30 BigDecimal 112.49 3.48× xRkosw2 compact idiom peer
kotlin add MQss P-gen x86_64 thru‡ 33.27 BigDecimal 108.23 3.25× xRkosw2 compact idiom peer
kotlin add MQos P-gen x86_64 thru‡ 51.28 BigDecimal 120.87 2.36× xRkosw2 compact idiom peer
kotlin add OQss P-gen x86_64 thru‡ 70.56 BigDecimal 216.17 3.06× xRkosw2 compact idiom peer
kotlin add OQos P-gen x86_64 thru‡ 101.24 BigDecimal 215.61 2.13× xRkosw2 compact idiom peer
kotlin add FQss P-gen x86_64 thru‡ 43.35 BigDecimal 242.44 5.59× xRkosw2 compact idiom peer
kotlin add FQos P-gen x86_64 thru‡ 48.66 BigDecimal 265.04 5.45× xRkosw2 compact idiom peer

Subtract — SQ · NQ · MQ · OQ · FQ

Swept 4096-input average per band (bare thru; ns/op = Time/4096 over the shared decimal128-resources/swept/ corpus, byte-identical operands every port). Relational peer table with explicit ratios; the cross-port d128 band-shape matrices are in Port-Comparison Benchmark Results.

arm64 (M3 Pro).

port op cat profile arch mode ours alt alt ns ratio run notes
kotlin sub SQss P-gen arm64 thru‡ 5.20 BigDecimal 22.48 4.32× Rkosw2 compact idiom peer
kotlin sub SQos P-gen arm64 thru‡ 3.66 BigDecimal 22.28 6.09× Rkosw2 compact idiom peer
kotlin sub NQss P-gen arm64 thru‡ 9.35 BigDecimal 34.26 3.66× Rkosw2 compact idiom peer
kotlin sub NQos P-gen arm64 thru‡ 8.70 BigDecimal 34.24 3.94× Rkosw2 compact idiom peer
kotlin sub MQss P-gen arm64 thru‡ 27.38 BigDecimal 33.52 1.22× Rkosw2 compact idiom peer
kotlin sub MQos P-gen arm64 thru‡ 13.23 BigDecimal 33.62 2.54× Rkosw2 compact idiom peer
kotlin sub OQss P-gen arm64 thru‡ 43.18 BigDecimal 83.36 1.93× Rkosw2 compact idiom peer
kotlin sub OQos P-gen arm64 thru‡ 25.00 BigDecimal 80.21 3.21× Rkosw2 compact idiom peer
kotlin sub FQss P-gen arm64 thru‡ 28.39 BigDecimal 89.90 3.17× Rkosw2 compact idiom peer
kotlin sub FQos P-gen arm64 thru‡ 18.11 BigDecimal 81.90 4.52× Rkosw2 compact idiom peer

x86_64 (Intel i9-9880H).

port op cat profile arch mode ours alt alt ns ratio run notes
kotlin sub SQss P-gen x86_64 thru‡ 18.91 BigDecimal 89.09 4.71× xRkosw2 compact idiom peer
kotlin sub SQos P-gen x86_64 thru‡ 13.99 BigDecimal 76.44 5.46× xRkosw2 compact idiom peer
kotlin sub NQss P-gen x86_64 thru‡ 31.74 BigDecimal 121.25 3.82× xRkosw2 compact idiom peer
kotlin sub NQos P-gen x86_64 thru‡ 27.44 BigDecimal 121.76 4.44× xRkosw2 compact idiom peer
kotlin sub MQss P-gen x86_64 thru‡ 47.73 BigDecimal 127.70 2.68× xRkosw2 compact idiom peer
kotlin sub MQos P-gen x86_64 thru‡ 33.69 BigDecimal 120.10 3.56× xRkosw2 compact idiom peer
kotlin sub OQss P-gen x86_64 thru‡ 106.02 BigDecimal 236.51 2.23× xRkosw2 compact idiom peer
kotlin sub OQos P-gen x86_64 thru‡ 73.52 BigDecimal 225.62 3.07× xRkosw2 compact idiom peer
kotlin sub FQss P-gen x86_64 thru‡ 54.19 BigDecimal 273.80 5.05× xRkosw2 compact idiom peer
kotlin sub FQos P-gen x86_64 thru‡ 41.09 BigDecimal 248.74 6.05× xRkosw2 compact idiom peer

Multiply — CP · WP · XP

Swept 4096-input average per band (bare thru; ns/op = Time/4096 over the shared decimal128-resources/swept/ corpus, byte-identical operands every port). Relational peer table with explicit ratios; the cross-port d128 band-shape matrices are in Port-Comparison Benchmark Results.

arm64 (M3 Pro).

port op cat profile arch mode ours alt alt ns ratio run notes
kotlin mul CP P-gen arm64 thru‡ 5.54 BigDecimal 11.88 2.14× Rkosw2 compact idiom peer
kotlin mul WP P-gen arm64 thru‡ 29.88 BigDecimal 64.15 2.15× Rkosw2 compact idiom peer
kotlin mul XP P-gen arm64 thru‡ 42.63 BigDecimal 160.46 3.76× Rkosw2 compact idiom peer

x86_64 (Intel i9-9880H).

port op cat profile arch mode ours alt alt ns ratio run notes
kotlin mul CP P-gen x86_64 thru‡ 17.36 BigDecimal 51.64 2.97× xRkosw2 compact idiom peer
kotlin mul WP P-gen x86_64 thru‡ 64.37 BigDecimal 210.28 3.27× xRkosw2 compact idiom peer
kotlin mul XP P-gen x86_64 thru‡ 110.91 BigDecimal 372.11 3.36× xRkosw2 compact idiom peer

Divide — CD · WD · XD (+ ET · PT)

Swept 4096-input average per band (bare thru; ns/op = Time/4096 over the shared decimal128-resources/swept/ corpus, byte-identical operands every port). Relational peer table with explicit ratios; the cross-port d128 band-shape matrices are in Port-Comparison Benchmark Results.

arm64 (M3 Pro).

port op cat profile arch mode ours alt alt ns ratio run notes
kotlin div CD P-gen arm64 thru‡ 34.17 BigDecimal 144.51 4.23× Rkosw2 compact idiom peer
kotlin div WD P-gen arm64 thru‡ 49.78 BigDecimal 124.07 2.49× Rkosw2 compact idiom peer
kotlin div XD P-gen arm64 thru‡ 52.47 BigDecimal 220.19 4.20× Rkosw2 compact idiom peer
kotlin div ET P-gen arm64 thru‡ 19.76 BigDecimal 435.10 22.02× Rkosw2 compact idiom peer
kotlin div PT P-gen arm64 thru‡ 11.71 BigDecimal 404.25 34.52× Rkosw2 compact idiom peer

x86_64 (Intel i9-9880H).

port op cat profile arch mode ours alt alt ns ratio run notes
kotlin div CD P-gen x86_64 thru‡ 111.07 BigDecimal 505.28 4.55× xRkosw2 compact idiom peer
kotlin div WD P-gen x86_64 thru‡ 170.68 BigDecimal 338.13 1.98× xRkosw2 compact idiom peer
kotlin div XD P-gen x86_64 thru‡ 194.15 BigDecimal 452.63 2.33× xRkosw2 compact idiom peer
kotlin div ET P-gen x86_64 thru‡ 54.85 BigDecimal 1710.40 31.18× xRkosw2 compact idiom peer
kotlin div PT P-gen x86_64 thru‡ 31.38 BigDecimal 1584.28 50.49× xRkosw2 compact idiom peer

FMA — FN (Barrett) · FF (fits-128)

Swept 4096-input average per band (bare thru; ns/op = Time/4096 over the shared decimal128-resources/swept/ corpus, byte-identical operands every port). Relational peer table with explicit ratios; the cross-port d128 band-shape matrices are in Port-Comparison Benchmark Results.

arm64 (M3 Pro).

port op cat profile arch mode ours alt alt ns ratio run notes
kotlin fma FN FMA arm64 thru‡ 110.65 - - - Rkosw2  
kotlin fma FF FMA arm64 thru‡ 89.52 - - - Rkosw2  

x86_64 (Intel i9-9880H).

port op cat profile arch mode ours alt alt ns ratio run notes
kotlin fma FN FMA x86_64 thru‡ 290.31 - - - xRkosw2  
kotlin fma FF FMA x86_64 thru‡ 229.52 - - - xRkosw2