Mac Mini Core i7 3.2 GHz Late 2018

Forums Benchmark Discussion Mac Mini Core i7 3.2 GHz Late 2018

This topic contains 7 replies, has 2 voices, and was last updated by Admin 3 months, 2 weeks ago.

Viewing 8 posts - 1 through 8 (of 8 total)
  • Author
    Posts
  • #1346

    Admin
    Keymaster

    There’s a Mac Mini Core i7 3.2 GHz Late 2018 in the benchmark with a score of 221 tracks. It’s pretty much at the top of the benchmark, in green. How realistic is that? Does anyone else have this Mac?

    I don’t want to delete the result yet, but I may if we find out other people who have this Mac have a much lower score.

    #1428

    amuckrunner
    Participant

    Hello,
    I’m the person reported 221 tracks.
    I was surprised with the result, too. So, little bit suspicious about project/global preference.

    My test environment was,
    Mac: late 2018 Mac mini core i7 3.2GHz, 16GB RAM, 1TB SSD
    Audio interface: Audient ID4
    Sampling rate: 44.1kHz
    IO buffer size: 256
    processing threads: 12
    process buffer size: large
    multithreading: playback & live tracks
    summing: high precision(64bit)
    rewire behavior: off

    tested with ‘looping bar 1 & 2’, and ‘looping bar 2 only’. There was no difference.

    Today, after the reading the Admin’s comment, I just tested with 64 IO buffer.
    I could reach even 235 tracks.

    Am I doing something wrong?
    I don’t want to corrupt the benchmark table.

    #1431

    Admin
    Keymaster

    First, thanks for reporting this back. 🙂

    It looks like there’s other people who had similar issues: https://www.gearslutz.com/board/apple-logic-pro/371545-logic-pro-multicore-benchmarktest-89.html#post10710606

    Apparently some IRs were missing in that case, thus the person was able to reach a very high score. I think the problem is the project file of the benchmark test. Where did you download the test? The one that can be downloaded here is the original benchmark test and I believe it is dated by now, because some IRs are apparently not downloaded by default in Logic Pro X, hence, not everyone can run the test the same way.

    Someone has found a new test on this site here: https://www.logicprohelp.com/forum/viewtopic.php?t=138612

    There’s a discussion about a new test here: https://logicbenchmarks.com/forums/topic/does-the-test-need-an-update/#post-1191

    I sent a message to the person on LogicProHelp if I can use their test, but haven’t received an answer yet.

    The thing is, the test must be compatible with all new versions of Logic Pro, but shouldn’t make all the previous test results obsolete. If we end up with ten different benchmark tests, the comparison is impossible. So I believe the new test should not be too heavily modified, it should just correct the bugs from the old test with the most minimalistic approach. I don’t know if the LogicProHelp test does that, but maybe someone else can give their input if that test is suited fo replacement or not. 🙂

    #1437

    Admin
    Keymaster

    @amuckrunner, I think I fixed the test. 🙂

    Can you try this out for me? Download the test here and tell us your score!

    #1438

    amuckrunner
    Participant

    Dear Admin,
    I just tested with the new project file before going to bed.
    The result is 225 tracks with 256 buffer size.

    About the IR thing, I just skimmed the threads you linked and please note that I’m very new to the Logic Pro x and I play guitar. that is I have no knowledge on EXS24, sampler stuff.

    To my knowledge, only ‘space designer’ plugin uses IR. And I found that the Space designer loads “0.8s_boom room”. Bypassing the space designer audibly removes reverb. So, the results seems to be correct.

    I think that the result is quite aligned with geekbench’s multicore score.

    I hope it’s correct as I just bought this machine and it’s my first desktop mac. 🙂

    P.S. Sorry for my poor English.

    #1439

    Admin
    Keymaster

    Don’t worry for your English, it’s at least as good as mine. 🙂

    I’m actually surprised about how powerful that Mac Mini is. I hadn’t seen it has scored so high on Geekbench. I guess your score is real!

    But did you bypass the Space Designer finally or not? And did you get any messages that some IRs (or other files) could not be found?

    #1450

    amuckrunner
    Participant

    Thanks for the updated benchmark.

    More detailed procedure of testing is as follows.

    * The effect of the Space designer is tested only for a single track and reverted to ‘on’ state in testing.
    * No warning or error message in loading the project other than ‘error initializing core MIDI’. I don’t think the core MIDI stuff affects the test result. I want to resolve the core MIDI error, though.
    * Buffer size varied from 64 to 256. There was no significant difference in crackling threshold(220~230 tracks).
    * CPU utilization shown in Mac’s monitoring reaches around 78%.
    * # of Processing threads: 12 (due to 6 cores)
    * process buffer size: large.
    * other settings are same as mentioned above.(Audient ID4, multithreading, summing, rewire behavior, etc.)

    * At the maximum # of tracks, activating the other windows(e.g. Safari, monitor) instantly introduces stop playing and system overrun message in LPX. So I think that the measured # tracks truly reached the maximum possible.

    Thanks for this benchmark and your help.

    Although this debugging/discussion was initiated by suspicion, there are take-aways for both you and me.
    For you, perfecting benchmark’s compatibility.
    For me, justifying the huge price of Mini. It is so expensive.
    🙂

    Thanks!

    • This reply was modified 3 months, 2 weeks ago by amuckrunner.
    #1452

    Admin
    Keymaster

    I honestly can’t say if the core midi error affects the result or not, but other than that it looks good! Your Mac Mini is a beast. We haven’t had any Mac Minis so high in the benchmark. iMacs started to catch up Mac Pros, but now even Mac Minis overtake them. 🙂

    You’re welcome for the test!

Viewing 8 posts - 1 through 8 (of 8 total)

You must be logged in to reply to this topic.