Huawei FusionServer Pro CH121 V5 User Manual page 21

FusionServer Pro CH121 V5 Compute Node
User Guide
l A server must use the same model of DDR4 memory modules, and all the memory
l The DDR4 memory modules of different types (RDIMM and LRDIMM) cannot be used
l Contact your local Huawei sales representative or use the
l
l
l
l
l
l
Issue 06 (2019-08-10)
modules operate at the same speed, which is the smallest value of:
l Memory speed supported by a processor
l Maximum operating speed of a memory module
together.
Compatibility Checker
The memory can be used with Intel
processors. The maximum memory capacity supported varies depending on the
processor model.
– Skylake processors
M processors: 1.5 TB/socket
n
Other processors: 768 GB/socket
n
– Cascade Lake processors
L processors: 4.5 TB/socket
n
M processors: 2 TB/socket
n
Other processors: 1 TB/socket
n
The total memory capacity is the sum of the capacity of all DDR4 memory modules.
The total memory capacity cannot exceed the maximum memory capacity supported by
the CPUs.
Use the Intelligent Computing Compatibility Checker
of a single memory module.
The maximum number of memory modules supported by a server varies depending on
the CPU type, memory type, rank quantity, and operating voltage.
NOTE
Each memory channel supports a maximum of 8 ranks. The number of memory modules
supported by each channel varies depending on the number of ranks supported by each channel:
Number of memory modules supported by each channel ≤ Number of ranks supported by each
memory channel/Number of ranks supported by each memory module
A memory channel supports more than eight ranks for LRDIMMs.
NOTE
A quad-rank LRDIMM generates the same electrical load as a single-rank RDIMM on a memory
bus.
DDR4 memory modules of different specifications (capacity, bit width, rank, and height)
can be used together for capacity expansion purposes. However, the memory Reliability,
Availability, and Serviceability (RAS) features may be affected.
Copyright © Huawei Technologies Co., Ltd.
to determine the components to be used.
® ®
Xeon Scalable Skylake and Cascade Lake
2 Hardware Description
Intelligent Computing
to determine the capacity type
13