Passmark rates half the RAM in overall results but does show full 64GB in MemoryMark

[Pat-Abitech]

Hi All,
A bit weird,
Passmark does show the full 64GB of RAM in it's detail but the overall score is very low for the memory.
This is on a new Dell Power Edge T430 Dual XEON E5-2620 2.10GHz with 64GB of DDR RAM running Server 2016 (although Passmark does not show Dual CPU's).
A comparison was done to a Server 2012 with 64GB of RAM(not sure of CPU), the RAM score is almost double.
We thought Passmark might only be scoring one CPU and 1 Channel of RAM even though the individual sections seem correct.
Attempting to attach comparison image below.
Thank you in advance

[David (PassMark)]

Having a lot of RAM doesn't mean it is fast RAM. Quantity and speed and different things.

Using Windows Server 2016 (rather than Server 2012) will not change the speed of the RAM, or the speed of any other hardware for that matter.

Different CPUs can change the RAM throughput however. Slow CPUs can't use all available bandwidth. Other factors with dual Xeons might be the use of ECC RAM or not, registered RAM or not and the use of NUMA.

[Pat-Abitech]

Hi David and thanks for the reply.
Should have mentioned the new 2016 server is ECC Registered (PC4-19200 ECC Registered) as supplied by Dell for this server.
It would be nice to know the CPU on the 2012 server but not ours.
The reason for the question is the server is under performing and the main third party application provider is trying to point the finger at anything but them selves (which they always do no matter what the issue is) they provided the comparison, I have run passmark on the server and got slightly better result but still not where expected.

[David (PassMark)]

ECC Registered RAM is known to be slower than typical desktop RAM, all else being equal.

NUMA can also have a big impact. Around 20% to 60% penalty for dual CPUs running software that isn't NMUMA aware. See,
https://www.passmark.com/forum/perfo...d-threadripper