Announcement

Collapse
No announcement yet.

Why does the Linux X86 version not test memory performance, but the ARM version does?

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

  • Why does the Linux X86 version not test memory performance, but the ARM version does?

    I used passmark Linux X86 platform to test the performance and Linux ARM version to test the performance, but I found that the X86 version could not test the performance of the memory, while the ARM version could test the performance of the memory. May I ask why this is?
    Refer to the image below. The first one is for the X86 platform, which has no Memory mark related test data.
    Click image for larger version

Name:	Hygon C3350_CentOS.png
Views:	35
Size:	83.5 KB
ID:	53854Click image for larger version

Name:	performance.png
Views:	17
Size:	149.4 KB
ID:	53855

  • #2
    Can you check and make sure you are using the latest version of PT Linux:
    https://www.passmark.com/products/pt_linux/download.php

    The legacy version does not include the memory test suite.

    Comment


    • #3
      Originally posted by Simon (PassMark) View Post
      Can you check and make sure you are using the latest version of PT Linux:
      https://www.passmark.com/products/pt_linux/download.php

      The legacy version does not include the memory test suite.
      Thank you for your reply.
      Is there any difference between the latest version and Legacy version in CPU test items and test logic?

      Comment


      • #4
        There were some changes to the CPU integer test, you can find more details here:
        https://www.passmark.com/products/pe...st/history.php

        Legacy version is V10.1.1000

        Comment


        • #5
          Does the latest version of Passmark performance support China ZHAOXIN CPU? As shown in the following figure, I ran the test program, but the test was not actually carried out. May I ask why? The same is true for both UOS and Ubuntu.
          Click image for larger version

Name:	_20221124171211.jpg
Views:	22
Size:	160.8 KB
ID:	53863

          Comment


          • #6
            Tests will not run until started by the user. The control menu is located at the bottom of the program. Use these keys to control the program:
            "A: Run All Tests, C: Run CPU Tests, M: Run Memory Tests, U: Upload Test Results"​


            If it still does not run, try running ptlinux in debug mode with the '-debug' command line argument and send us the generated log file.
            https://www.passmark.com/support/pt_linux_faq.php

            Comment


            • #7
              Thank you for your guidance, I have run the program.

              However, I still have a question about the test results. The following test results show that the two cpus have different number of cores and different frequencies.

              1. According to the test results, the result of core 32 is not as good as that of core 16, is it because the main frequency is relatively low? If this conclusion is followed, intel server Xeon cpu should turn out to be inferior to desktop Core cpu.

              2. Which has more impact on the final result, the memory capacity or the frequency?

              Click image for larger version

Name:	_20221125083704.png
Views:	20
Size:	52.7 KB
ID:	53866

              Comment


              • #8
                According to the test results, the result of core 32 is not as good as that of core 16
                The Hygon 5380 16 Core CPU has a CPUMark benchmark of 22,189.
                The Hygon 7385 32 Core CPU has a CPUMark benchmarkof 38,097

                So I am not sure why you are saying a 16 Core CPU is always better?
                They are very very close for the single threaded result however.

                If you are comparing Hygon 7385 16 to the i9-12900K, then the i9 is better in several areas. This is despite it only having 8 performance Cores (the other 8 are low efficiency cores).
                The 5.2Ghz clock speed only applies for single threaded, where the i9 is more than twice as fast as the 7385. The i9 probably runs at around 3.2Ghz most of the time.

                In short the Hygon 7385 doesn't have great per core performance (compared to current CPUs). But it doesn't follow that all Xeon CPUs are equally bad as the Hygon. Hygon 7385 would be competitive with a Intel Xeon Platinum 8180, released at the start of 2018.

                Which has more impact on the final result, the memory capacity or the frequency
                Memory capacity doesn't impact performance at all. Once you have 'enough' RAM adding more RAM doesn't make the CPU faster.
                RAM frequency can be important (but so are number of channels and timings and which software you are running)

                Comment


                • #9
                  Originally posted by David (PassMark) View Post

                  So I am not sure why you are saying a 16 Core CPU is always better?
                  They are very very close for the single threaded result however.

                  If you are comparing Hygon 7385 16 to the i9-12900K, then the i9 is better in several areas. This is despite it only having 8 performance Cores (the other 8 are low efficiency cores).
                  The 5.2Ghz clock speed only applies for single threaded, where the i9 is more than twice as fast as the 7385. The i9 probably runs at around 3.2Ghz most of the time.
                  Yes, my question is that the comparison is made between the results of Intel I9 and Hygon 7385. As for the above test results, 7385 is better than i9 in some items(exp: The red items in the image below are all good), but the total score is lower.
                  I am trying to find out the key factors that affect the performance result through this comparison, so as to achieve the conditions for the best test results.
                  Click image for larger version

Name:	CPU.png
Views:	18
Size:	27.8 KB
ID:	53869
                  Also, in the intel i9 and hygon 7385 RAM tests shown above, the i9 only has 2 16GB DDR4 RAM (2 RAM channels) and the 7385 has 16 16GB DDR4 RAM (8 RAM channels), but the results are pretty similar.
                  Another point is that the Hygon 7385's memory runs at a reduced frequency to 2666MHz, while the i9's runs at 3200MHz, but does this frequency difference make up for the efficiency of the six memory channels?

                  Comment


                  • #10
                    The Physics and Prime numbers test are the most sensitive to the RAM speed. And the 7385 had a good win in those tests. So it is certainly possible that the win was the result of the 8 channel RAM.

                    You might find the i9 would do better than the results above if you had used a Z690 motherboard with fast DDR5 RAM (instead of slow DDR4). Might have got 150 Million / Primes per sec and 2300 Frames / Sec with better RAM.

                    Comment

                    Working...
                    X