Machine for Gaussian

Our customer inquired about a calculation server that handles the molecular dynamics simulation software "GROMACS" and the quantum chemistry calculation software "Gaussian."
The client is looking for a machine that can run GROMACS and Gaussian at high speed, and the ideal specifications would be close to the following:

CPU: Xeon Gold 6354 (18core 3.0GHz 205W) x2
Memory: 256GB
SSD: One for 960GB SATA x2 system and one for backup
GPU: NVIDIA RTX A6000x2
OS: Alma Linux 8.3

Taking this information into consideration, we proposed the following configuration:

CPU Xeon Gold 6326 (2.90GHz 16 cores) x2
memory 256GB REG ECC
Storage 1 1TB SSD S-ATA
Storage 2 1TB SSD S-ATA
Video Nvidia RTX A6000
network On Board (1000Base-T X2)
Housing + power supply Tower housing + 1200W
OS Alma Linux 8.3
Other CUDA Toolkit 11 installation

As a result of discussions with the customer, we are considering a configuration that prioritizes budget.
Since the main use was for GROMACS, we placed importance on a configuration suitable for GROMACS within the budget.
Additionally, with an emphasis on processing speed, we have selected the following CPUs, memory capacities, and GPUs with high parallel performance.

・Xeon Gold 6326 (2.90GHz 16 cores) x2
・256GB REG ECC
・NVIDIA RTX A6000

Therefore, although it is compatible with GROMACS, if you intend to use it in conjunction with Gaussian16, you will need to be careful about the GPU.

Notes on this configuration

"Gaussian16" officially supports only GPGPU dedicated card products that support double precision arithmetic such as NVIDIA V100 and A100. Also, if you change the GPU to NVIDIA A100, which is supported by Gaussian, you will need to change the configuration of the PC itself to take into account the physical specifications of the GPU.
Note that if you are using Gaussian 16 only for CPU execution, the selection of the GPU does not affect operation.