STAC Report: STAC-A2 (derivatives risk) on NVIDIA H100 80GiB in HPE ProLiant XL675d Gen10 Plus

First STAC-A2 results on NVIDIA H100 GPUs.

28 September 2023

STAC recently performed the first STAC-A2 Benchmark tests on NVIDIA H100 GPUs. The system included 8 GPUs in an HPE ProLiant Gen10 Plus Server and CUDA® 12.0. The STAC Report is now available here.

STAC-A2 is the technology benchmark standard based on financial market risk analysis. Designed by quants and technologists from some of the world's largest banks, STAC-A2 reports the performance, scaling, quality, and resource efficiency of any technology stack that is able to handle the workload (Monte Carlo estimation of Heston-based Greeks for a path-dependent, multi-asset option with early exercise).

The stack consisted of the STAC-A2 Pack for CUDA (Rev H) using CUDA 12.0 on x NVIDIA H100 PCIe 80GiB GPUs in an HPE ProLiant XL675d Gen10 Plus running SUSE Linux Enterprise Server 15 SP4.

HPE wished to highlight several results from this report:

• Compared to all publicly reported solutions to date, this NVIDIA H100-based solution set numerous performance and efficiency records, including (but not limited to):
• The first sub-10ms warm time (8.9 ms) in the baseline Greeks benchmark¹
• A cold time in the baseline Greeks benchmark² (38 ms) over 3x faster than any previously reported
• The fastest warm³ (0.51 s) and cold⁴ (1.85 s) times in the large Greeks benchmarks
• The most correlated assets⁵ (400) and most paths⁶ (310,000,000) simulated in 10 minutes
• The best energy efficiency⁷ (311,045 options / kWh)
• Compared to a solution using 8 x NVIDIA A100 SXM4 80GiB GPUs, as well as previous generations of the NVIDIA STAC Pack and CUDA (SUT ID NVDA210914), this solution:
• Was 10x the speed in the cold run of the baseline Greeks benchmark²
• Was 1.38x the speed in the warm runs of the baseline Greeks benchmark¹
• Was 1.38x the speed in the cold run of the large Greeks benchmark⁴
• Was 1.4x the speed in the warm runs of the large Greeks benchmark³
• Was 10% more energy efficient⁷

For details, please see the report at the link above. Premium subscribers have access to the code used in this project as well as the micro-detailed configuration information for the solution. To learn about subscription options, please contact us.

---------------
¹STAC-A2.β2.GREEKS.TIME.WARM
²STAC-A2.β2.GREEKS.TIME.COLD
³STAC-A2.β2.GREEKS.10-100k-1260TIME.WARM
⁴STAC-A2.β2.GREEKS.10-100k-1260TIME.COLD
⁵STAC-A2.β2.GREEKS.MAX_ASSETS
⁶STAC-A2.β2.GREEKS.MAX_PATHS
⁷STAC-A2.β2.HPORTFOLIO.ENERG_EFF