Platforms · Measurement record
Recordgnr-2026-11-05Outputs validatedFixed
Intel Xeon 6 (Granite Rapids)
Measured on 5 November 2026 on a whole machine, with the published method. 26 double-precision measurements of the tuned code for this platform: five launches each, every launch recorded.
- Published method
- Run and checked by Xcelerit
- Suite version: November 2026 round
- Double precision
- 10 measurements repeated on a second machine
- Power limit
- 600 W per socket
- Processors
- Intel Xeon 6 P-core (Granite Rapids), 2 × 96 cores
- Memory
- 755 GiB
- Operating system
- Ubuntu 24.04.5 LTS
- Compiler
- Intel oneAPI icpx 2026.2 -O3 -qopenmp -march=graniterapids -mprefer-vector-width=512
- Libraries
- oneMKL
- Placement
- OMP_PLACES=threads OMP_PROC_BIND=close, all threads
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 | 25.54 | 41,057,923 | 9.573 Wh/bn | ±1.00% | 0.77% | Inspect |
| Barrier options, Monte Carlo | 16,000 options | 16.15 | 990,817 | 0.3534 Wh/M | ±1.00% | 0.97% | Inspect |
| American options, binomial lattice | 512K options | 120.65 | 4,345,423 | 73.35 Wh/bn | ±1.00% | 2.81% | Inspect |
| European options, Black-Scholes formula | 128M options | 578.89 | 23,185,527,929 | 14.46 Wh/tn | ±0.46% | 0.44% | Inspect |
All 26 measurements
| Workload | Code | Size | Time (ms) | Within-launch CI | Launch spread | Average warm calls per launch | Published from | Other machine (ms) |
|---|---|---|---|---|---|---|---|---|
| Swaption portfolio, LMM | Tuned code | 16K | 0.482 | ±1.00% | 0.56% | 11,646 | – | – |
| Swaption portfolio, LMM | Tuned code | 32K | 0.849 | ±1.00% | 0.67% | 6,279 | machine 2 | machine 1: 0.863 (+1.7%, launch spread 1.1%) |
| Swaption portfolio, LMM | Tuned code | 64K | 1.645 | ±1.00% | 0.23% | 2,584 | machine 2 | machine 1: 1.877 (+14.1%, launch spread 31.0%) |
| Swaption portfolio, LMM | Tuned code | 128K | 3.238 | ±1.00% | 0.83% | 1,720 | – | – |
| Swaption portfolio, LMM | Tuned code | 256K | 6.471 | ±1.00% | 0.11% | 1,724 | – | – |
| Swaption portfolio, LMM | Tuned code | 512K | 12.839 | ±1.00% | 0.31% | 1,324 | – | – |
| Swaption portfolio, LMM | Tuned code | 1,024K | 25.539 | ±1.00% | 0.77% | 865 | – | – |
| Swaption portfolio, LMM | Tuned code | 2,048K | 51.295 | ±1.00% | 0.50% | 686 | – | – |
| Barrier options, Monte Carlo | Tuned code | 2,000 | 2.204 | ±1.00% | 1.75% | 1,525 | machine 2 | machine 1: 2.273 (+3.1%, launch spread 5.9%) |
| Barrier options, Monte Carlo | Tuned code | 4,000 | 4.268 | ±1.00% | 5.64% · noisy | 965 | machine 2 | machine 1: 4.473 (+4.8%, launch spread 17.8%) |
| Barrier options, Monte Carlo | Tuned code | 8,000 | 9.277 | ±0.94% | 16.09% · noisy | 514 | machine 2 | machine 1: 8.836 (-4.8%, launch spread 16.6%) |
| Barrier options, Monte Carlo | Tuned code | 16,000 | 16.148 | ±1.00% | 0.97% | 856 | machine 2 | machine 1: 16.888 (+4.6%, launch spread 4.1%) |
| Barrier options, Monte Carlo | Tuned code | 32,000 | 32.069 | ±1.00% | 1.02% | 673 | machine 2 | machine 1: 34.044 (+6.2%, launch spread 12.3%) |
| American options, binomial lattice | Tuned code | 16K | 3.971 | ±1.00% | 0.34% | 802 | – | – |
| American options, binomial lattice | Tuned code | 32K | 7.630 | ±1.00% | 1.79% | 953 | – | – |
| American options, binomial lattice | Tuned code | 64K | 15.269 | ±1.00% | 2.20% | 1,217 | – | – |
| American options, binomial lattice | Tuned code | 128K | 30.112 | ±1.00% | 0.14% | 865 | – | – |
| American options, binomial lattice | Tuned code | 256K | 59.792 | ±1.00% | 0.79% | 491 | – | – |
| American options, binomial lattice | Tuned code | 512K | 120.653 | ±1.00% | 2.81% | 228 | – | – |
| American options, binomial lattice | Tuned code | 1,024K | 233.569 | ±1.00% | 0.82% | 72 | – | – |
| American options, binomial lattice | Tuned code | 2,048K | 456.926 | ±0.98% | 1.40% | 23 | – | – |
| European options, Black-Scholes formula | Tuned code | 16M | 47.883 | ±1.00% | 1.14% | 268 | machine 2 | machine 1: 51.250 (+7.0%, launch spread 23.4%) |
| European options, Black-Scholes formula | Tuned code | 32M | 101.704 | ±1.00% | 1.54% | 401 | machine 2 | machine 1: 102.991 (+1.3%, launch spread 14.6%) |
| European options, Black-Scholes formula | Tuned code | 64M | 236.936 | ±1.00% | 2.15% | 170 | machine 2 | machine 1: 261.342 (+10.3%, launch spread 31.5%) |
| European options, Black-Scholes formula | Tuned code | 128M | 578.886 | ±0.46% | 0.44% | 30 | – | – |
| European options, Black-Scholes formula | Tuned code | 256M | 1,206.331 | ±0.57% | 0.34% | 30 | – | – |
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 | Identical to double precision | Pass |
| Barrier options, Monte Carlo | Agrees to 13 significant digits | 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 gnr-2026-11-05.
SHA-256a6c9840ca8084d75c3c13c5c58659b000930ab03185c166e96896aef95ef527b
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