Independently Verified on Kaggle

25.5 DAYS to 2.7 MIN

13,848x faster. Both EXACT.

13,848x

Speedup

104M

Elements Exact

0

Precision Loss

Free

pip install

Execution Time Comparison

CPU Decimal takes DAYS. VLA takes SECONDS. Both give EXACT results.

MatrixElementsCPU DecimalVLA GPUSpeedup
512x512262K4.6 min0.05s6,071x
1024x10241M36.7 min0.2s12,922x
2048x20484M4.9 hrs1.3s13,814x
4096x409617M1.6 DAYS10.1s13,934x
6144x614438M5.5 DAYS34.3s13,885x
8192x819267M13.1 DAYS1.4 min13,843x
10240x10240105M25.5 DAYS2.7 min13,848x

Benchmarked on Tesla T4 (Kaggle), February 2026

Cross-GPU Reproducibility

Same checksum on completely different architectures. This is unprecedented.

10240x10240 Matrix Multiply Checksum

RTX 4070 (Ada Lovelace, sm_89)

6ece6956f187064f

Tesla T4 (Turing, sm_75)

6ece6956f187064f

BIT-IDENTICAL

Different GPU architectures, different memory layouts, same exact result

100%

Reproducible

2

GPU architectures verified

0

Bit differences

VLA Beats Everything

Test: 1e20 + 10,000 ones - 1e20. Expected result: 10,000. Only VLA gets it right.

FP32

8,750

Lost 1,250

FP64

7,500

Lost 2,500

80-bit Extended

9,984

Lost 16

VLA

10,000

EXACT

VLA beats Intel 80-bit extended precision hardware - on a consumer GPU.

FP64 Loses Tens of Thousands of Values

VLA recovers ALL of them.

TestExpectedFP64 LostVLA Lost
1e20 + 10K - 1e209,9981,262 (12.6%)0
1e20 + 100K - 1e2099,99825,022 (25.0%)0
1e20 + 500K - 1e20499,99824,862 (5.0%)0
1e20 + 1M - 1e20999,99833,342 (3.3%)0

Real-World Impact

Financial Transactions

$881,143,573.77

1 million transactions summed

FP64 error: $0.0000001VLA error: $0.00

Patriot Missile Tracking

100 Hours

0.1s increments accumulated

FP64 miss: 0.0000002mVLA miss: 0.0m

Lorenz Chaos System

50,000 Steps

Chaotic trajectory integration

FP64: Standard trajectoryVLA: 0.0 divergence

Orbit Propagation

10 Orbits

ISS-altitude satellite tracking

FP64 drift: 4.29 kmVLA drift: 4.67 km

Try It Yourself

All benchmarks are reproducible. Run them on Kaggle or install locally.

$ pip install simgen-vla