8 vs. 12 Channel DDR5-6000 Memory Performance With AMD 5th Gen EPYC

Written by Michael Larabel in Memory on 20 November 2024 at 11:40 AM EST. Page 2 of 7. 29 Comments.
RAMspeed SMP benchmark with settings of Type: Copy, Benchmark: Integer. 12c DDR5-6000 was the fastest.
Stream benchmark with settings of Type: Add. 12c DDR5-6000 was the fastest.
MBW benchmark with settings of Test: Memory Copy, Array Size: 4096 MiB. 12c DDR5-6000 was the fastest.

Within the synthetic RAM benchmarks obviously is a clear difference out of eight versus twelve memory channels, but for this testing focusing more on the real-world workloads.

High Performance Conjugate Gradient benchmark with settings of X Y Z: 144 144 144, RT: 60. 12c DDR5-6000 was the fastest.

Within the commonly used HPCG benchmark that is very memory bandwidth intensive, going from eight to twelve memory channels increased the score by 50% with the additional four DDR5-6000 RDIMMs and keeping to the same EPYC 9655 processor.

NAS Parallel Benchmarks benchmark with settings of Test / Class: BT.C. 12c DDR5-6000 was the fastest.
NAS Parallel Benchmarks benchmark with settings of Test / Class: CG.C. 12c DDR5-6000 was the fastest.
NAS Parallel Benchmarks benchmark with settings of Test / Class: FT.C. 12c DDR5-6000 was the fastest.
NAS Parallel Benchmarks benchmark with settings of Test / Class: IS.D. 12c DDR5-6000 was the fastest.
NAS Parallel Benchmarks benchmark with settings of Test / Class: SP.C. 12c DDR5-6000 was the fastest.

Across the NPB benchmarks there are also the very clear performance benefits out of making use of the full 12 memory channel support found with EPYC 9005 series processors.

NAMD benchmark with settings of Input: ATPase with 327,506 Atoms. 12c DDR5-6000 was the fastest.
NAMD benchmark with settings of Input: STMV with 1,066,628 Atoms. 12c DDR5-6000 was the fastest.
Algebraic Multi-Grid Benchmark benchmark with settings of . 12c DDR5-6000 was the fastest.

Depending upon the HPC workload though and the size of the data set, in some areas the 12 memory channel performance is less impactful than others.

Xcompact3d Incompact3d benchmark with settings of Input: X3D-benchmarking input.i3d. 12c DDR5-6000 was the fastest.
OpenFOAM benchmark with settings of Input: drivaerFastback, Small Mesh Size, Execution Time. 12c DDR5-6000 was the fastest.
OpenFOAM benchmark with settings of Input: drivaerFastback, Medium Mesh Size, Execution Time. 12c DDR5-6000 was the fastest.
Graph500 benchmark with settings of Scale: 26. 12c DDR5-6000 was the fastest.
Graph500 benchmark with settings of Scale: 26. 12c DDR5-6000 was the fastest.
GROMACS benchmark with settings of Implementation: MPI CPU, Input: water_GMX50_bare. 12c DDR5-6000 was the fastest.
GPAW benchmark with settings of Input: Carbon Nanotube. 12c DDR5-6000 was the fastest.

For most HPC deployments though you'll be best off going for populating all 12 memory channels -- and wanting to pick out a server/motherboard solution allowing the full 12 channel DDR5-6000/DDR5-6400 memory support rather than those limited to eight channels.

Related Articles
Crucial Pro DDR5-6400 32GB Kit Performance With Intel Arrow Lake
Intel Core Ultra 9 285K Linux Memory DDR5 Performance Testing
AMD EPYC 9755 DDR5-4800 vs. DDR5-6000 Memory Performance
MRDIMM 8800MT/s vs. DDR5-6400 Memory Performance With Intel Xeon 6
AMD Ryzen 9 9950X DDR5-6000 / DDR5-6400 / DDR5-8000 Memory Performance
AMD Ryzen 7 9700X Performance With DDR5-8000