AMD EPYC 2 Rome What We Know Will Change the Game

Discussion in 'STH Main Site Posts' started by Patrick Kennedy, Nov 6, 2018.

  1. #1
    eva2000, realtomatoes and i386 like this.
  2. realtomatoes

    realtomatoes Active Member

    Joined:
    Oct 3, 2016
    Messages:
    239
    Likes Received:
    31
    yep, i needed to come here to *drool*
     
    #2
  3. Edu

    Edu Member

    Joined:
    Aug 8, 2017
    Messages:
    37
    Likes Received:
    5
    I seems the new architecture increases off-die bandwidth and reduced off-die latency. This will benefit non NUMA aware applications, like databases. However, since all DRAM access has to happen off-die, the latency for NUMA aware applications will certainly increase. Maybe, they are using something like TSMC's InFO to connect the dies, which would allow them more bandwidth.
     
    #3
  4. MiniKnight

    MiniKnight Well-Known Member

    Joined:
    Mar 30, 2012
    Messages:
    2,760
    Likes Received:
    780
    If you're building a new per socket licensing cluster next year, you'd be crazy not to buy Rome. That's insane.
     
    #4
    capn_pineapple and realtomatoes like this.
  5. zir_blazer

    zir_blazer Active Member

    Joined:
    Dec 5, 2016
    Messages:
    152
    Likes Received:
    45
    Was it confirmed that the Memory Controller is on the I/O chip? Somehow it seems horrible from a latency standpoint, but the TR-W series showed that pure compute cores could scale "well enough".
    Also, this would make a single Rome fully UMA. Latency to memory should be uniform across all different dies, as all the CPU dies have to do one hop to the I/O die to get to the RAM. If anything, you still have the Thread to CPU allocation due to SMT, cache sharing, inter CCX latency, etc, but the RAM should be uniform.
     
    #5
  6. Edu

    Edu Member

    Joined:
    Aug 8, 2017
    Messages:
    37
    Likes Received:
    5
    Yes it was confirmed. Indeed RAM access will be more uniform, but that hardly matters. Really, you just want access to be fast as possible.
     
    #6
  7. zir_blazer

    zir_blazer Active Member

    Joined:
    Dec 5, 2016
    Messages:
    152
    Likes Received:
    45
    Technically that dramatically simplifies EPYC platform. Memory Channels and intersocket I/O are now single extremely wide ones instead of being a multitude of narrow paths where each die had multiple I/O controllers.

    CPU Die -> I/O Die -> Memory Bank/CPU Die or CPU Die -> I/O Die -> I/O Die -> Memory Bank/CPU Die in Dual EPYC setups. Doesn't seems that bad. Reminds me of early Pentium 2/3 era Xeons where you had 4 Processors wired on parallel on the same FSB along with one or two Northbridges with their own Memory Controller and PCI Host Bridge.
    Still, latency should be higher overall. There should be a few scenarios where you have a performance regression due to that...
     
    #7
Similar Threads: EPYC Rome
Forum Title Date
STH Main Site Posts AMD EPYC Rome Details Trickle Out 64 Cores 128 Threads Per Socket Jun 6, 2018
STH Main Site Posts Xilinx Alveo U280 Launched Possibly with AMD EPYC CCIX Support Yesterday at 8:22 AM
STH Main Site Posts New AMD EPYC 7371 Frequency Optimized Processor Launched Nov 13, 2018
STH Main Site Posts Noctua NH-U12S TR4-SP3 AMD EPYC and Threadripper Cooler Review Nov 2, 2018
STH Main Site Posts New Gigabyte AMD EPYC Server 2U4N and Motherboard Options Nov 1, 2018

Share This Page