2 Replies Latest reply on Jul 6, 2016 11:25 AM by Intel Corporation

    Theory craft for a gaming rig using a Xeon workstation motherboard.


      I am looking into building a system with the Asus Z10PE-D8 WS. The primary use for this will be gaming however. I am aware that overclocking a core i7 would be a more cost effective solution however, i want to build around the bus bandwidths associated with the board. There are other cons i do not like, such as having to run buffered RDIMMs while not being able to overclock the Xeons themselves as needed. So 2400mhz Dimms in a quad channel configuration. The good news is however i will have more dram capacity to set when creating custom game configurations through external scripting and game configuration data.

      I am thinking of using [1] 3.5 Ghz Xeon E5 2637 V4 Broadwell EP =). I am looking more for lower thread count ultra high performance so i chose one with a high clock rate. But the real reason why i want to use this CPU and Motherboard is QPI. DMI 2.0 has a max bandwidth rating of 5 GT's. QPI for the single CPU configuration chip i am eye balling, is 9.6GT's, almost double the bandwidth rating from CPU to PCI-E vs DMI 2.0.


      The reason why i am making this thread, if any Intel engineers have the answers for me =). How much more effective is QPI for dual video card or expansion slot scaling over DMI 2.0 in terms of potential bottle necking for the CPU?

      I have read that QPI is a point to point bidirectional connection between 2 or more cpus. But what does this mean for Expansion bus data access on a 1 CPU configuration over DMI 2.0?