Platforms · Measurement record
Recordb200-2026-11-05Outputs validatedFixed
NVIDIA B200
Measured on 5 November 2026 on one GPU, with the published method. 26 double-precision measurements of the tuned code for this platform: five launches each, every launch recorded.
One whole GPU, on the 96 host threads of its memory node. Default power limit, ECC on, clocks not locked.
- Published method
- Run and checked by Xcelerit
- Suite version: November 2026 round
- Double precision
- Accelerator
- NVIDIA B200
- Power limit
- 1000 W
- Driver
- 595.91.07
- Host CPU
- Intel Xeon 5th gen (Emerald Rapids), 2 × 48 cores, 192 threads
- Host memory
- 1,996 GiB
- Operating system
- Ubuntu 24.04.5 LTS
- Compiler
- CUDA 12.9.1, sm_100
- Libraries
- cuRAND
Results at a glance
The sizes the results rank at. Time is the mean of the middle three of five launches; throughput is the problem size over that time (Black-Scholes: options × 100 passes, the work of one call). Sizes in K and M are binary (1K = 1,024); throughput is per second, in decimal units. Tuned code: tuned for this chip under the same rules. Select a row to see the launches behind it.
| Workload | Size | Time (ms) | Throughput (per s) | Energy | Within-launch CI | Launch spread | |
|---|---|---|---|---|---|---|---|
| Swaption portfolio, LMM | 1,024K paths | 13.44 | 78,044,106 | 2.02 Wh/bn | ±0.32% | 0.22% | Inspect |
| Barrier options, Monte Carlo | 16,000 options | 9.56 | 1,674,074 | 0.08497 Wh/M | ±1.00% | 1.27% | Inspect |
| American options, binomial lattice | 512K options | 283.38 | 1,850,124 | 63.06 Wh/bn | ±0.21% | 0.18% | Inspect |
| European options, Black-Scholes formula | 128M options | 254.83 | 52,669,238,337 | 2.814 Wh/tn | ±0.41% | 0.39% | Inspect |
All 26 measurements
| Workload | Code | Size | Time (ms) | Within-launch CI | Launch spread | Average warm calls per launch |
|---|---|---|---|---|---|---|
| Swaption portfolio, LMM | Tuned code | 16K | 0.782 | ±0.50% | 0.15% | 2,559 |
| Swaption portfolio, LMM | Tuned code | 32K | 0.799 | ±0.89% | 0.13% | 3,349 |
| Swaption portfolio, LMM | Tuned code | 64K | 1.528 | ±0.53% | 0.20% | 1,309 |
| Swaption portfolio, LMM | Tuned code | 128K | 2.307 | ±0.44% | 0.17% | 867 |
| Swaption portfolio, LMM | Tuned code | 256K | 3.817 | ±0.62% | 0.43% | 524 |
| Swaption portfolio, LMM | Tuned code | 512K | 7.161 | ±0.41% | 0.25% | 280 |
| Swaption portfolio, LMM | Tuned code | 1,024K | 13.436 | ±0.32% | 0.22% | 149 |
| Swaption portfolio, LMM | Tuned code | 2,048K | 25.828 | ±0.52% | 0.36% | 135 |
| Barrier options, Monte Carlo | Tuned code | 2,000 | 1.858 | ±1.00% | 1.48% | 18,841 |
| Barrier options, Monte Carlo | Tuned code | 4,000 | 2.888 | ±1.00% | 1.15% | 15,151 |
| Barrier options, Monte Carlo | Tuned code | 8,000 | 5.146 | ±1.00% | 1.24% | 6,597 |
| Barrier options, Monte Carlo | Tuned code | 16,000 | 9.558 | ±1.00% | 1.27% | 3,815 |
| Barrier options, Monte Carlo | Tuned code | 32,000 | 18.145 | ±1.00% | 0.33% | 938 |
| American options, binomial lattice | Tuned code | 16K | 9.148 | ±1.00% | 0.76% | 1,781 |
| American options, binomial lattice | Tuned code | 32K | 17.719 | ±1.00% | 0.40% | 576 |
| American options, binomial lattice | Tuned code | 64K | 34.726 | ±1.00% | 0.60% | 203 |
| American options, binomial lattice | Tuned code | 128K | 69.042 | ±0.63% | 0.48% | 38 |
| American options, binomial lattice | Tuned code | 256K | 140.581 | ±0.08% | 0.11% | 15 |
| American options, binomial lattice | Tuned code | 512K | 283.380 | ±0.21% | 0.18% | 10 |
| American options, binomial lattice | Tuned code | 1,024K | 568.200 | ±0.08% | 0.05% | 10 |
| American options, binomial lattice | Tuned code | 2,048K | 1,138.502 | ±0.04% | 0.02% | 10 |
| European options, Black-Scholes formula | Tuned code, pinned host memory | 16M | 34.049 | ±1.00% | 1.24% | 917 |
| European options, Black-Scholes formula | Tuned code, pinned host memory | 32M | 65.190 | ±0.82% | 0.72% | 86 |
| European options, Black-Scholes formula | Tuned code, pinned host memory | 64M | 128.769 | ±0.88% | 0.32% | 22 |
| European options, Black-Scholes formula | Tuned code, pinned host memory | 128M | 254.831 | ±0.41% | 0.39% | 15 |
| European options, Black-Scholes formula | Tuned code, pinned host memory | 256M | 506.828 | ±0.04% | 0.06% | 10 |
Outputs validated before speed
A time only counts if the output is right. At every problem size, the option price the tuned code computes is validated against the reference implementation's on the same machine.
| Workload | Against the reference code | Result |
|---|---|---|
| Swaption portfolio, LMM | Within 0.78% at 128K paths, on this platform's own random-number generator | Pass |
| Barrier options, Monte Carlo | Within 0.18% at 32,000 options, on this platform's own random-number generator | Pass |
| American options, binomial lattice | Identical to double precision | Pass |
| European options, Black-Scholes formula | Identical to double precision | Pass |
Binomial and Black-Scholes are deterministic: the tuned code gives the reference code's prices. Monte Carlo: on GPUs the barrier and LMM codes generate their random numbers differently, so their prices differ only by sampling: within 4 standard errors of the reference implementation's, and fed the same random numbers they agree with the reference implementation's to within 1e-9 (relative).
Every launch behind one measurement
Pick a workload and a size. Each dot is one launch: the mean of its warm calls, run until the 95% confidence interval was within ±1% or until the call limit. The fastest and slowest launch are dropped and the middle three averaged into the result.
Zoom: drag a box on the chart or use the slider under it; reset brings it back.
| Launch | Warm calls | Warm mean (ms) | Range (ms) | Used |
|---|
How time grows with the problem
Time against size on log scales, for every size measured. The dashed line is perfect linear scaling from the smallest size: points under it mean the machine was not yet full at small sizes.
What the record holds
The measurement record behind this page: every launch, the machine and the output validation, and the results as a table.
Full record (JSON) Results (CSV)
Free to cite with attribution: Xcelerit Finance Benchmarks, record b200-2026-11-05.
SHA-256ba0f98dea711bb1a4e78845a279ac4b1e7ecd5d288189b0efa2210d3992eacf3
Records are fixed: once published, a record is never edited. A correction is published as a new record, linked from the old one. Checksums of every record