Announcement

Collapse
No announcement yet.

Memtest86 V11.0 Pro - Shows different L1/L2 xxx.x GB/s Cache on identical systems

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Memtest86 V11.0 Pro - Shows different L1/L2 xxx.x GB/s Cache on identical systems

    Hey all!

    Something i've wondered about before, but never been able to really question it. Is there an obvious reason why different (identical) systems would boot Memtest differently? They appear to show different speeds (bandwidth?) for the L1/L2 cache numbers.

    Node 1

    Click image for larger version

Name:	CaptureScreen (22).jpg
Views:	29
Size:	74.5 KB
ID:	57937

    Node 2

    Click image for larger version

Name:	CaptureScreen (21).jpg
Views:	30
Size:	75.8 KB
ID:	57941

    Both BIOSes have been flashed at the same time, reset to defaults and then booted in to Memtest. All hardware was acquired from the same batches/suppliers etc.

    I don't suppose it's a major issue, but it's something I've been curious of for a while.

    Any insight would be greatly appreciated.
    Attached Files

  • #2
    L2 numbers are very very close. All benchmarks have a small margin for error. This can be due to thermal or power throttling / ramping and background threads running plus other factors. If you ran the test many times, you'd probably find the numbers jump around ±5%

    The 25% difference in the L1 results is more significant however. We don't know why there is this level of difference between your machines. How consistent is this on the two machines?

    Comment


    • #3
      Thanks for your reply!

      Another slightly strange thing is that the 'slower' one is taking an eternity, compared to the other one.

      Click image for larger version  Name:	CaptureScreen (24).jpg Views:	0 Size:	79.2 KB ID:	57952

      Click image for larger version  Name:	CaptureScreen (25).jpg Views:	0 Size:	87.3 KB ID:	57953

      Comment


      • #4
        Another slightly strange thing is that the 'slower' one is taking an eternity, compared to the other one.
        So maybe it isn't a measurement error and there really is some difference between the machines.
        Might be worth running some other benchmarks. Then start looking for the difference. e.g. max boost clock speed, RAM in different slots, BIOS set memory timings differently.

        Comment

        Working...
        X