AMD Makes A Compelling Case For Budget-Friendly Ryzen Dedicated Servers

Thanks to ServeTheWorld for providing the remote access for this round of benchmarking. (Picture courtesy of ServeTheWorld from their Oslo data center.)

For the Linux benchmarking I carried out remotely (unfortunately having no Xeon E-2300 series hardware locally) in evaluating Xeon E vs. Ryzen performance, the tested processors included:

2nd Gen AMD “Matisse” Zen 2
- Ryzen 5 Pro 3600 (6C, 65W)
- Ryzen 7 Pro 3700 (8C, 65W)
- Ryzen 9 Pro 3900 (12C, 65W)
- Ryzen 5 3600X (6C, 105W)
- Ryzen 7 3800X (8C, 105W)
- Ryzen 9 3900X (12C, 105W)
- Ryzen 9 3950X (16C, 105W)

3rd Gen AMD “Vermeer” Zen 3
- Ryzen 5 5600X (6C, 65W)
- Ryzen 7 5800X (8C, 105W)
- Ryzen 9 5900X (12C, 105W)
- Ryzen 9 5950X (16C, 105W)

2nd Gen Intel “Coffee Lake”
- Xeon E-2278G (8C, 80W)
- Xeon E-2288G (8C, 95W)

3rd Gen Intel “Rocket Lake E”
- Xeon E-2386G (6C, 95W)
- Xeon E-2388G (8C, 95W)

The Ryzen server processors were all tested in an ASRockRack X570D4U-2L2T platform while the Xeon E processors were using the ASRock Rack E3C246DI-NL. All of the servers had 128GB (4 x 32GB DDR4-3200) ECC memory, Samsung NVMe SSD storage, and configured/tested in their ServeTheWorld data center. The selection of processors tested were based on what they had currently available and with supply constraints a limited selection of Xeon E-2300 series processors but at least having the flagship Xeon E-2388G and the flagship prior-generation Xeon E-2288G.

Picture courtesy of ServeTheWorld from their Oslo data center.

Alma Linux 8.5 was used for all of this benchmarking with the Linux 4.18 kernel, given the increasing popularity of Alma Linux as an alternative/successor to CentOS 8, and already becoming popular in the data center.