7
$\begingroup$

My MacbookAir M1 provided inspiring results for BenchmarkReport[] ... a score of 3.18. I was excited to run this on my new MBPro with M1Max... and got a very low score, 0.21

{"MachineName" -> "office-mbp", "System" -> "Mac OS X ARM (64-bit)",
 "BenchmarkName" -> "WolframMark", "FullVersionNumber" -> "12.3.1",
 "Date" -> "December 16, 2021", "BenchmarkResult" -> 1.352, "TotalTime" -> 10.235,
 "Results" -> {{"Data Fitting", 0.371}, {"Digits of Pi", 0.166},
               {"Discrete Fourier Transform", 0.315}, {"Eigenvalues of a Matrix", 2.085},
               {"Elementary Functions", 0.498}, {"Gamma Function", 0.213},
               {"Large Integer Multiplication", 0.181}, {"Matrix Arithmetic", 0.257},
               {"Matrix Multiplication", 0.264}, {"Matrix Transpose", 0.196},
               {"Numerical Integration", 0.873}, {"Polynomial Expansion", 0.052},
               {"Random Number Sort", 0.396}, {"Singular Value Decomposition", 3.359},
               {"Solving a Linear System", 1.009}}}

A puzzle to me. Any thoughts?

$\endgroup$
10
  • 1
    $\begingroup$ Try the full benchmark report and see if one operation in particular is holding everything back (maybe driver or library issue), or if all steps are slow. $\endgroup$
    – MassDefect
    Dec 16, 2021 at 17:50
  • 2
    $\begingroup$ Perhaps you can post the full results of Benchmark[] as well. $\endgroup$
    – Carl Lange
    Dec 16, 2021 at 18:10
  • 1
    $\begingroup$ Related thread on the Wolfram Community site. $\endgroup$ Dec 16, 2021 at 20:22
  • 2
    $\begingroup$ To be clear, the link from @Nate applies to V12.3. V13 is universal, so there is a single download that contains both Intel and Apple Silicon versions and automatically runs the correct one. $\endgroup$ Dec 16, 2021 at 23:04
  • 3
    $\begingroup$ For 12.3, this has been improved with a paclet update (it may however take a couple of days to be delivered depending on when the last check for updates was). The expected score (assuming the machine is otherwise idle) should be around 3.4 and similar to the V13 result. $\endgroup$
    – ilian
    Dec 17, 2021 at 1:17

1 Answer 1

2
$\begingroup$

Site questions in principle should be answerable, even if practically that's not always possible. "Any thoughts?" is pretty broad, but since this question has not been closed, here is what seems the most authoritative response from the comments:

For 12.3, this has been improved with a paclet update (it may however take a couple of days to be delivered depending on when the last check for updates was). The expected score (assuming the machine is otherwise idle) should be around 3.4 and similar to the V13 result. – ilian Dec 17, 2021 at 1:17

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.