MySQL tests were performed on bare-metal single socket servers with equivalent memory, networking and storage configurations for each of the platforms shown. The processors tested here are: AMD EPYC 7763 "Milan"; Intel Xeon 8380 "Ice Lake"; Ampere Altra Q80-30; Ampere Altra Max M128-30
This test was accomplished using the Sysbench multi-threaded benchmark tool. The tool can create database schemas, populate tables with data, and generate multi-thread load (SQL queries) towards the database server.
Six OLTP tests were run which use the following SQL queries:
sb11-OLTP_RO_10M_8tab-uniform-dst_ranges1-notrx : 5
sb11-OLTP_RO_10M_8tab-uniform-notrx : 1+2+3+4+5
sb11-OLTP_RO_10M_8tab-uniform-p_sel1-notrx : 1
sb11-OLTP_RO_10M_8tab-uniform-s_ranges1-notrx : 2
sb11-OLTP_RW_10M_8tab-uniform-notrx : 1+2+3+4+5+6+7
sb11-OLTP_RW_10M_8tab-uniform-upd_idx1-notrx : 6
Sysbench was deployed on different CPU sockets with Mysql Server and database connection is through TCP/IP socket interface over loopback IP address.
The Mysql server storage was configured as a RAMdisk filesystem.
Since it is realistic to measure throughput under a specified Service Level Agreement (SLA), a 95th percentile latency (p.95) of 1 millisecond was used. This ensures that 95 percent of the requests have a response time of 1 ms in the worst case.
The above benchmark was run on all platforms and six test cases’ TPS results at P.95 of 1 millisecond were collected and the geomean of those was used as the platform’s final score.