STAC Report: AWS c6i.metal under STAC-A2 (derivatives risk)
Intel Ice Lake in AWS bare metal sets price-performance records.
15 November 2022
As a follow up to recently announced audit results, STAC performed STAC-A2 Benchmark tests on a bare metal AWS instance with a newer generation of Intel processors (Ice Lake). 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 consisting of the STAC-A2 Pack for Intel® oneAPI (Rev N) on a single c6i.metal Amazon Web Services instance with 2 x Intel® Xeon® Platinum 8375C (Ice Lake) processors and 256 GiB of DRAM.
Intel wished to highlight several results from this report:
- Compared to all other publicly reported cloud-based solutions, this solution had:
- The highest options per dollar over a 1 hour burst and a 3 day period
- The highest options per dollar over a 1 year period (reflecting reserve instance pricing discounts)
(STAC-A2.β2.HPORTFOLIO.PRICE_PERF.CONTINUOUS) - Compared to an on-premises solution involving 2 x Intel Xeon Platinum 8280 (Cascade Lake) CPUs and a previous generation of the Intel-authored STAC Pack (SUT ID INTC190402), this solution:
- Had 1.5x the throughput (STAC-A2.β2.HPORTFOLIO.SPEED)
- Was 1.3x the speed in cold runs and 1.4x the speed in warm runs of the baseline problem size
(STAC-A2.β2.GREEKS.TIME.COLD and .WARM) - Was 1.8x the speed in cold runs and 1.4x the speed in warm runs of the large problem size
(STAC-A2.β2.GREEKS.10-100k-1260TIME.COLD and .WARM) - Handled 2x the paths in the max paths test (STAC-A2.β2.GREEKS.MAX_PATHS)
- Compared to a solution involving an AWS c5.metal instance with 2 x Intel Xeon Platinum 8275CL (Cascade Lake) CPUs (SUT ID 221006b), this solution:
- Had 1.5x the throughput (STAC-A2.β2.HPORTFOLIO.SPEED)
- Was 1.3x the speed in cold runs and 1.5x the speed in warm runs of the baseline problem size
(STAC-A2.β2.GREEKS.TIME.COLD and .WARM) - Was 1.4x the speed in cold and warm runs of the large problem size
(STAC-A2.β2.GREEKS.10-100k-1260TIME.COLD and .WARM)
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.
About STAC News
Read the latest about research, events, and other important news from STAC.