close
close

AMD EPYC 9965 “Turin Dense” Delivers Better Performance/Power Efficiency Compared to AmpereOne 192-Core ARM CPU Review

Coremark benchmark with settings of CoreMark Size 666, iterations per second. AmpereOne A192-32X was the fastest.
Coremark benchmark with settings of CoreMark Size 666, iterations per second. AmpereOne A192-32X was the fastest.
Coremark benchmark with settings of CoreMark Size 666, iterations per second. AmpereOne A192-32X was the fastest.

The synthetic Coremark benchmark was one of the few cases where the AmpereOne A192-32X outperformed the EPYC 9965.

Algebraic Multi-Grid Benchmark benchmark with settings of . EPYC 9965 was the fastest.
Algebraic Multi-Grid Benchmark benchmark with settings of . EPYC 9965 was the fastest.
Algebraic Multi-Grid Benchmark benchmark with settings of . EPYC 9965 was the fastest.

The AmpereOne A192-32X is tied to eight channels of DDR5-5200 memory, while AMD EPYC Turin allows 12-channel memory at DDR5-6000 speeds. Ampere Computing is expected to release AmpereOne M with 12-channel DDR5 memory support this quarter, but memory speeds have yet to be confirmed. The increased memory bandwidth for EPYC Turin will benefit workloads such as AMG.

WRF benchmark with Input settings: cone 2.5 km. EPYC 9965 was the fastest.
WRF benchmark with Input settings: cone 2.5 km. EPYC 9965 was the fastest.

The increased memory bandwidth and Zen 5 benefits made the EPYC 9965 far outperform AmpereOne. This EPYC 9965 run is also in a suboptimal configuration due to DDR5 memory throttling in this benchmark due to an overheating issue with one of the DIMMs. But even with this underpowered EPYC 9965 WRF run, it still comes in well ahead of the AmpereOne CPU in this 192 core battle.

LULESH benchmark with settings of . AmpereOne A192-32X was the fastest.
LULESH benchmark with settings of . AmpereOne A192-32X was the fastest.
LULESH benchmark with settings of . AmpereOne A192-32X was the fastest.

The LULESH hydrodynamics benchmark was a rare setback for the EPYC Turin Dense 192 core processor.

LAMMPS Molecular Dynamics Simulator benchmark with settings of Model: 20k Atoms. EPYC 9965 was the fastest.
LAMMPS Molecular Dynamics Simulator benchmark with settings of Model: 20k Atoms. EPYC 9965 was the fastest.
LAMMPS Molecular Dynamics Simulator benchmark with settings of Model: 20k Atoms. EPYC 9965 was the fastest.
miniFE benchmark with settings of Problem size: Small. EPYC 9965 was the fastest.
miniFE benchmark with settings of Problem size: Small. EPYC 9965 was the fastest.
miniFE benchmark with settings of Problem size: Small. EPYC 9965 was the fastest.
GROMACS benchmark with Implementation settings: MPI CPU, Input: water_GMX50_bare. EPYC 9965 was the fastest.
GROMACS benchmark with Implementation settings: MPI CPU, Input: water_GMX50_bare. EPYC 9965 was the fastest.
GROMACS benchmark with Implementation settings: MPI CPU, Input: water_GMX50_bare. EPYC 9965 was the fastest.
QuantLib benchmark with configuration settings: Multi-Threaded. EPYC 9965 was the fastest.
QuantLib benchmark with configuration settings: Multi-Threaded. EPYC 9965 was the fastest.
QuantLib benchmark with configuration settings: Multi-Threaded. EPYC 9965 was the fastest.
GPAW benchmark with input settings: carbon nanotube. EPYC 9965 was the fastest.
GPAW benchmark with input settings: carbon nanotube. EPYC 9965 was the fastest.
High Performance Conjugate Gradient benchmark with settings of XYZ: 144 144 144, RT: 60. EPYC 9965 was the fastest.
High Performance Conjugate Gradient benchmark with settings of XYZ: 144 144 144, RT: 60. EPYC 9965 was the fastest.
High Performance Conjugate Gradient benchmark with settings of XYZ: 144 144 144, RT: 60. EPYC 9965 was the fastest.

The AMD EPYC 9965 Turin Dense processor delivered dominant performance in most HPC benchmarks tested compared to the AmpereOne A192-32X flagship ARM server processor.