May 20, 2021 — STAC recently performed STAC-A2 Benchmark tests on a stack consisting of the STAC-A2 Pack for NEC Vector Engine (Rev A) with 8 x NEC SX-Aurora TSUBASA/Vector Engine Type 20B accelerator cards in an NEC SX-Aurora TSUBASA B300-8 server. The server had patches applied to mitigate Spectre & Meltdown security vulnerabilities.
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).
NEC wished to highlight several results from this report:
- Compared to all publicly reported solutions to date, this Vector Engine-based solution had the fastest cold time in the large problem size Greeks benchmark (STAC-A2.β2.GREEKS.10-100k-1260.TIME.COLD)
- Compared to the previous best results for single-server solutions, this solution was:
- 79% faster in the cold time for the large Greeks benchmark
(STAC-A2.β2.GREEKS.10-100k-1260.TIME.COLD vs. SUT ID INTC210315) - 18% faster in the warm time for the large Greeks benchmark
(STAC-A2.β2.GREEKS.10-100k-1260.TIME.WARM vs. SUT ID NVDA200909)
- 79% faster in the cold time for the large Greeks benchmark
- Compared to the previous best results for a coprocessor-based solution (SUT ID NVDA200909), this solution:
- had 2.07x the speed in the cold time and 1.18x the speed in the warm time for the large Greeks benchmark
(STAC-A2.β2.GREEKS.10-100k-1260.TIME.{COLD,WARM}) - was within 10% of the maximum assets (STAC-A2.β2.GREEKS.MAX_ASSETS)
- had 2.07x the speed in the cold time and 1.18x the speed in the warm time for the large Greeks benchmark
- Compared to the previous best results for CPU-based single-server solutions, this solution delivered:
- 2.31x the throughput (STAC-A2.β2.HPORTFOLIO.SPEED vs. SUT ID INTC190903)
- 2.12x the speed in the warm time for the large Greeks benchmark
(STAC-A2.β2.GREEKS.10-100k-1260.TIME.WARM vs. SUT ID INTC190903) - 1.79x the speed in the cold time for the large Greeks benchmark
(STAC-A2.β2.GREEKS.10-100k-1260.TIME.COLD vs. SUT ID INTC210315) - 1.44x the speed in the warm time for the baseline Greeks benchmark
(STAC-A2.β2.GREEKS.TIME.WARM vs. SUT ID INTC210315)
For details, please see the report at the link above.
Source: STAC
