AMD EPYC 9755 DDR5-4800 vs. DDR5-6000 Memory Performance

Written by Michael Larabel in Memory on 14 October 2024 at 10:45 AM EDT. Page 3 of 9. 3 Comments.
ClickHouse benchmark with settings of 100M Rows Hits Dataset, First Run / Cold Cache. DDR5-6000 was the fastest.
ClickHouse benchmark with settings of 100M Rows Hits Dataset, Second Run. DDR5-6000 was the fastest.
ClickHouse benchmark with settings of 100M Rows Hits Dataset, Third Run. DDR5-6000 was the fastest.

There was just about 2% faster performance with the ClickHouse database from running the Samsung DDR5 ECC RDIMMs at DDR5-6000 rather than DDR5-4800.

PostgreSQL benchmark with settings of Scaling Factor: 100, Clients: 1000, Mode: Read Write. DDR5-4800 was the fastest.
PostgreSQL benchmark with settings of Scaling Factor: 100, Clients: 1000, Mode: Read Write, Average Latency. DDR5-4800 was the fastest.
PostgreSQL benchmark with settings of Scaling Factor: 100, Clients: 1000, Mode: Read Only. DDR5-6000 was the fastest.
PostgreSQL benchmark with settings of Scaling Factor: 100, Clients: 1000, Mode: Read Only, Average Latency. DDR5-6000 was the fastest.

The PostgreSQL difference was very small from the DDR5-4800 vs. DDR5-6000 comparison.

Memcached benchmark with settings of Set To Get Ratio: 1:100. DDR5-4800 was the fastest.
RocksDB benchmark with settings of Test: Read While Writing. DDR5-6000 was the fastest.
Speedb benchmark with settings of Test: Random Read. DDR5-6000 was the fastest.
Speedb benchmark with settings of Test: Read While Writing. DDR5-4800 was the fastest.

For many workloads there was no measurable performance loss from using DDR5-4800 speeds with AMD EPYC Turin.

Related Articles