Brand new Cisco UCS M5 C-Series

This is the second and last part of a small blog series about Cisco’s brand new UCS M5 series. In this blog: C-Series and GPUs.

Cisco UCS C220 M5 Rack Server


Record-breaking performance in a dense 1RU form factor for a wide range of workloads including virtualization, web, collaboration, cloud, and bare metal applications

What’s new

  • Support for Intel® Xeon® Scalable processors
  • Support for 2666 MHz DDR4 DIMMS with new capacity point of 128GB
  • GPU support (max 1 per server) – post-FCS
  • Increased storage density – up to 10 SFF per 1RU
  • 10GbE Embedded LOM
  • Modular Internal FlexStorage Option: Mirrored SD Card or M.2 SATA @ FCS with SW RAID
  • New SSD and NVMe Options
  • Post FCS: 3D XPoint Memory, M.2 NVMe 

Cisco UCS C240 M5 Rack Server


Unrivaled performance for a wide range of enterprise workloads including big data analytics, collaboration, databases, virtualization, and high performance applications

What’s new

  • Support for Intel® Xeon® Scalable processors
  • Support for 2666 MHz DDR4 DIMMS with new capacity point of 128GB
  • Hot-swap rear SFF drives with centralized RAID
  • 10GbE Embedded LOM
  • Modular Internal FlexStorage Option: Mirrored SD Card or M.2 SATA @ FCS with SW RAID
  • New SSD and NVMe Options
  • Post FCS: 3D XPoint Memory, M.2 NVMe 

Cisco UCS C480 M5 4S Rack Server

Designed for the most demanding server workloads such as in-memory database, EDA, Machine Learning, Data Analytics, and CPU / GPU rendering
  • Stand-alone or UCS-managed operations
  • Ultimate scale-up platform

New Drive Sleds for C-Series M5 Rack Servers

  • Smaller 2.5-in sled design enables better density (up to 10 drives) in the 1U server
  • Older M4 generation drive PIDs will not work with new M5 chassis
  • M5 blade servers will continue to use the older generation PIDs
  • 3.5-in sled will be redesigned to match the 2.5-inch brand identity

M4 – M5 comparison

GPUs

  • Manageability: UCS Manager and CIMC integration – Firmware, vBios, Service Profiles
  • Performance: Run at PCIe Gen 3 speed –Cisco + NVIDIA engineering collaboration fixed issues
  • Flexibility/Performance: Supported with all high-bin CPUs – significant mechanical/thermal work done by Cisco to overcome these limitations
  • Performance: Test up to 35c in Heat chambers with all max configs. Ensure no GPU throttling (requires fan algorithm integration and for some servers, custom baffling).