Analyzing the Performance Impact of HPC Workloads with Gramine+SGX on 3rd Generation Xeon Scalable Processors.

Intel SGX, which provides a strongly isolated execution environment for running applications to protect their code and data from privileged adversaries, is hopeful for increased integrity and confidentiality in multi-user systems. However, SGX is rarely used in the field of HPC despite the increasing importance of data protection. This is because SGX had limited memory resources and required supercomputing users to modify their code for execution. Lately, both 3rd generation Xeon scalable processors, which have many advanced data protection features including increased SGX memory, and Gramine 1.0, which is the first production-ready version of a library OS for the execution of unmodified applications on SGX, were released. However, the impact of these state-of-the-art technologies on HPC workloads is not well investigated. In this paper, we provide the first performance analysis of HPC workloads with Gramine and SGX on 3rd generation Xeon scalable processors. Our analysis starts with some microbenchmarks and is then extended to various HPC workloads. Our experimental results show that Gramine+SGX shows a small performance overhead (4-17%) for both compute-intensive and memory-bandwidth-sensitive workloads but a bit large performance overhead (up to 170%) for a memory-latency-sensitive workload. In addition, we show that the combination of Gramine and a 3rd generation Xeon scalable processor leads to a slowdown of 1.5x on average (up to 4.4x) for many HPC workloads. This number is an order of magnitude smaller than that reported in the previous work using the combination of the former-generation SGX toolchain and processor.