Skip to content

assume_p_const = false is extremely slow #40

@MarcoArtiano

Description

@MarcoArtiano

assume_p_const = false

───────────────────────────────────────────────────────────────────────────────────────────
                Trixi.jl                          Time                    Allocations      
                                         ───────────────────────   ────────────────────────
            Tot / % measured:                  334s / 100.0%            276GiB / 100.0%    

Section                          ncalls     time    %tot     avg     alloc    %tot      avg
───────────────────────────────────────────────────────────────────────────────────────────
alg                               1.88k     334s  100.0%   178ms    276GiB  100.0%   151MiB
  krylov solve stage 2              625     124s   37.3%   199ms   88.9GiB   32.2%   146MiB
    ~krylov solve stage 2~          625     113s   33.9%   181ms   88.9GiB   32.2%   146MiB
    parabolic rhs!                1.37k    11.5s    3.4%  8.36ms   8.70KiB    0.0%    6.51B
      calculate gradient          1.37k    5.21s    1.6%  3.80ms   3.33KiB    0.0%    2.49B
        volume integral           1.37k    2.16s    0.6%  1.58ms     0.00B    0.0%    0.00B
        surface integral          1.37k    1.34s    0.4%   978μs     0.00B    0.0%    0.00B
        reset gradients           1.37k    505ms    0.2%   369μs     0.00B    0.0%    0.00B
        interface flux            1.37k    457ms    0.1%   334μs     0.00B    0.0%    0.00B
        prolong2interfaces        1.37k    405ms    0.1%   296μs     0.00B    0.0%    0.00B
        Jacobian                  1.37k    275ms    0.1%   201μs     0.00B    0.0%    0.00B
        ~calculate gradient~      1.37k   30.1ms    0.0%  22.0μs   3.33KiB    0.0%    2.49B
        boundary flux             1.37k   13.6ms    0.0%  9.91μs     0.00B    0.0%    0.00B
        prolong2boundaries        1.37k   12.7ms    0.0%  9.24μs     0.00B    0.0%    0.00B
        prolong2mortars           1.37k   2.89ms    0.0%  2.11μs     0.00B    0.0%    0.00B
        mortar flux               1.37k   2.54ms    0.0%  1.86μs     0.00B    0.0%    0.00B
      calculate viscous fluxes    1.37k    2.34s    0.7%  1.71ms     0.00B    0.0%    0.00B
      prolong2interfaces          1.37k    1.25s    0.4%   910μs     0.00B    0.0%    0.00B
      volume integral             1.37k    1.03s    0.3%   749μs     0.00B    0.0%    0.00B
      transform variables         1.37k    550ms    0.2%   401μs     0.00B    0.0%    0.00B
      interface flux              1.37k    424ms    0.1%   310μs     0.00B    0.0%    0.00B
      reset ∂u/∂t                 1.37k    247ms    0.1%   181μs     0.00B    0.0%    0.00B
      surface integral            1.37k    224ms    0.1%   164μs     0.00B    0.0%    0.00B
      Jacobian                    1.37k   99.0ms    0.0%  72.3μs     0.00B    0.0%    0.00B
      ~parabolic rhs!~            1.37k   39.4ms    0.0%  28.7μs   5.38KiB    0.0%    4.02B
      prolong2boundaries          1.37k   31.5ms    0.0%  23.0μs     0.00B    0.0%    0.00B
      boundary flux               1.37k   13.7ms    0.0%  10.0μs     0.00B    0.0%    0.00B
      prolong2mortars             1.37k   3.10ms    0.0%  2.26μs     0.00B    0.0%    0.00B
      mortar flux                 1.37k   2.47ms    0.0%  1.81μs     0.00B    0.0%    0.00B
  krylov solve stage 3              625     106s   31.8%   170ms   86.0GiB   31.1%   141MiB
    ~krylov solve stage 3~          625    95.0s   28.5%   152ms   86.0GiB   31.1%   141MiB
    parabolic rhs!                1.32k    11.0s    3.3%  8.30ms   8.70KiB    0.0%    6.74B
      calculate gradient          1.32k    5.00s    1.5%  3.78ms   3.33KiB    0.0%    2.58B
        volume integral           1.32k    2.08s    0.6%  1.57ms     0.00B    0.0%    0.00B
        surface integral          1.32k    1.29s    0.4%   977μs     0.00B    0.0%    0.00B
        reset gradients           1.32k    480ms    0.1%   363μs     0.00B    0.0%    0.00B
        interface flux            1.32k    439ms    0.1%   332μs     0.00B    0.0%    0.00B
        prolong2interfaces        1.32k    389ms    0.1%   294μs     0.00B    0.0%    0.00B
        Jacobian                  1.32k    261ms    0.1%   197μs     0.00B    0.0%    0.00B
        ~calculate gradient~      1.32k   28.7ms    0.0%  21.7μs   3.33KiB    0.0%    2.58B
        boundary flux             1.32k   12.6ms    0.0%  9.53μs     0.00B    0.0%    0.00B
        prolong2boundaries        1.32k   12.1ms    0.0%  9.12μs     0.00B    0.0%    0.00B
        prolong2mortars           1.32k   2.80ms    0.0%  2.12μs     0.00B    0.0%    0.00B
        mortar flux               1.32k   2.58ms    0.0%  1.95μs     0.00B    0.0%    0.00B
      calculate viscous fluxes    1.32k    2.25s    0.7%  1.70ms     0.00B    0.0%    0.00B
      prolong2interfaces          1.32k    1.19s    0.4%   896μs     0.00B    0.0%    0.00B
      volume integral             1.32k    989ms    0.3%   747μs     0.00B    0.0%    0.00B
      transform variables         1.32k    523ms    0.2%   395μs     0.00B    0.0%    0.00B
      interface flux              1.32k    404ms    0.1%   306μs     0.00B    0.0%    0.00B
      reset ∂u/∂t                 1.32k    239ms    0.1%   181μs     0.00B    0.0%    0.00B
      surface integral            1.32k    214ms    0.1%   161μs     0.00B    0.0%    0.00B
      Jacobian                    1.32k   93.6ms    0.0%  70.7μs     0.00B    0.0%    0.00B
      ~parabolic rhs!~            1.32k   37.7ms    0.0%  28.5μs   5.38KiB    0.0%    4.16B
      prolong2boundaries          1.32k   29.8ms    0.0%  22.5μs     0.00B    0.0%    0.00B
      boundary flux               1.32k   12.9ms    0.0%  9.77μs     0.00B    0.0%    0.00B
      prolong2mortars             1.32k   3.08ms    0.0%  2.32μs     0.00B    0.0%    0.00B
      mortar flux                 1.32k   2.19ms    0.0%  1.66μs     0.00B    0.0%    0.00B
  krylov solve stage 1              625    59.6s   17.8%  95.3ms   40.9GiB   14.8%  67.0MiB
    ~krylov solve stage 1~          625    54.3s   16.3%  86.9ms   40.9GiB   14.8%  67.0MiB
    parabolic rhs!                  625    5.24s    1.6%  8.38ms   8.70KiB    0.0%    14.3B
      calculate gradient            625    2.38s    0.7%  3.81ms   3.33KiB    0.0%    5.45B
        volume integral             625    995ms    0.3%  1.59ms     0.00B    0.0%    0.00B
        surface integral            625    615ms    0.2%   983μs     0.00B    0.0%    0.00B
        reset gradients             625    231ms    0.1%   369μs     0.00B    0.0%    0.00B
        interface flux              625    207ms    0.1%   332μs     0.00B    0.0%    0.00B
        prolong2interfaces          625    182ms    0.1%   290μs     0.00B    0.0%    0.00B
        Jacobian                    625    125ms    0.0%   200μs     0.00B    0.0%    0.00B
        ~calculate gradient~        625   13.5ms    0.0%  21.6μs   3.33KiB    0.0%    5.45B
        boundary flux               625   5.99ms    0.0%  9.59μs     0.00B    0.0%    0.00B
        prolong2boundaries          625   5.62ms    0.0%  8.99μs     0.00B    0.0%    0.00B
        prolong2mortars             625   1.39ms    0.0%  2.22μs     0.00B    0.0%    0.00B
        mortar flux                 625   1.22ms    0.0%  1.95μs     0.00B    0.0%    0.00B
      calculate viscous fluxes      625    1.07s    0.3%  1.72ms     0.00B    0.0%    0.00B
      prolong2interfaces            625    569ms    0.2%   910μs     0.00B    0.0%    0.00B
      volume integral               625    471ms    0.1%   753μs     0.00B    0.0%    0.00B
      transform variables           625    252ms    0.1%   404μs     0.00B    0.0%    0.00B
      interface flux                625    194ms    0.1%   311μs     0.00B    0.0%    0.00B
      reset ∂u/∂t                   625    110ms    0.0%   176μs     0.00B    0.0%    0.00B
      surface integral              625    101ms    0.0%   162μs     0.00B    0.0%    0.00B
      Jacobian                      625   44.6ms    0.0%  71.4μs     0.00B    0.0%    0.00B
      ~parabolic rhs!~              625   18.1ms    0.0%  28.9μs   5.38KiB    0.0%    8.81B
      prolong2boundaries            625   14.3ms    0.0%  22.9μs     0.00B    0.0%    0.00B
      boundary flux                 625   6.15ms    0.0%  9.84μs     0.00B    0.0%    0.00B
      prolong2mortars               625   1.46ms    0.0%  2.34μs     0.00B    0.0%    0.00B
      mortar flux                   625   1.01ms    0.0%  1.62μs     0.00B    0.0%    0.00B
  parabolic rhs!                  1.88k    15.2s    4.6%  8.10ms   8.70KiB    0.0%    4.75B
    calculate gradient            1.88k    8.85s    2.7%  4.72ms   3.33KiB    0.0%    1.82B
      volume integral             1.88k    3.88s    1.2%  2.07ms     0.00B    0.0%    0.00B
      surface integral            1.88k    2.45s    0.7%  1.31ms     0.00B    0.0%    0.00B
      interface flux              1.88k    775ms    0.2%   414μs     0.00B    0.0%    0.00B
      Jacobian                    1.88k    769ms    0.2%   410μs     0.00B    0.0%    0.00B
      prolong2interfaces          1.88k    569ms    0.2%   304μs     0.00B    0.0%    0.00B
      reset gradients             1.88k    336ms    0.1%   179μs     0.00B    0.0%    0.00B
      ~calculate gradient~        1.88k   37.2ms    0.0%  19.8μs   3.33KiB    0.0%    1.82B
      boundary flux               1.88k   18.6ms    0.0%  9.91μs     0.00B    0.0%    0.00B
      prolong2boundaries          1.88k   15.2ms    0.0%  8.11μs     0.00B    0.0%    0.00B
      prolong2mortars             1.88k   3.58ms    0.0%  1.91μs     0.00B    0.0%    0.00B
      mortar flux                 1.88k   3.14ms    0.0%  1.67μs     0.00B    0.0%    0.00B
    calculate viscous fluxes      1.88k    2.43s    0.7%  1.30ms     0.00B    0.0%    0.00B
    volume integral               1.88k    1.16s    0.3%   618μs     0.00B    0.0%    0.00B
    prolong2interfaces            1.88k    1.08s    0.3%   575μs     0.00B    0.0%    0.00B
    interface flux                1.88k    593ms    0.2%   316μs     0.00B    0.0%    0.00B
    transform variables           1.88k    443ms    0.1%   236μs     0.00B    0.0%    0.00B
    surface integral              1.88k    287ms    0.1%   153μs     0.00B    0.0%    0.00B
    reset ∂u/∂t                   1.88k    167ms    0.0%  88.9μs     0.00B    0.0%    0.00B
    Jacobian                      1.88k   92.4ms    0.0%  49.3μs     0.00B    0.0%    0.00B
    ~parabolic rhs!~              1.88k   48.1ms    0.0%  25.7μs   5.38KiB    0.0%    2.94B
    prolong2boundaries            1.88k   24.7ms    0.0%  13.2μs     0.00B    0.0%    0.00B
    boundary flux                 1.88k   17.9ms    0.0%  9.56μs     0.00B    0.0%    0.00B
    prolong2mortars               1.88k   3.66ms    0.0%  1.95μs     0.00B    0.0%    0.00B
    mortar flux                   1.88k   3.28ms    0.0%  1.75μs     0.00B    0.0%    0.00B
  Jacobian inside stage1            625    10.3s    3.1%  16.5ms   20.4GiB    7.4%  33.4MiB
  Jacobian inside stage 2           625    8.17s    2.4%  13.1ms   19.9GiB    7.2%  32.6MiB
  Jacobian inside stage 3           625    4.23s    1.3%  6.77ms   19.9GiB    7.2%  32.6MiB
  rhs!                            1.88k    3.64s    1.1%  1.94ms   4.78KiB    0.0%    2.61B
    interface flux                1.88k    1.44s    0.4%   769μs     0.00B    0.0%    0.00B
    volume integral               1.88k    1.24s    0.4%   660μs     0.00B    0.0%    0.00B
    prolong2interfaces            1.88k    378ms    0.1%   202μs     0.00B    0.0%    0.00B
    surface integral              1.88k    293ms    0.1%   156μs     0.00B    0.0%    0.00B
    Jacobian                      1.88k   97.8ms    0.0%  52.2μs     0.00B    0.0%    0.00B
    reset ∂u/∂t                   1.88k   85.3ms    0.0%  45.5μs     0.00B    0.0%    0.00B
    boundary flux                 1.88k   47.2ms    0.0%  25.2μs     0.00B    0.0%    0.00B
    ~rhs!~                        1.88k   40.0ms    0.0%  21.4μs   4.78KiB    0.0%    2.61B
    prolong2boundaries            1.88k   10.5ms    0.0%  5.61μs     0.00B    0.0%    0.00B
    prolong2mortars               1.88k   3.38ms    0.0%  1.80μs     0.00B    0.0%    0.00B
    mortar flux                   1.88k   3.10ms    0.0%  1.65μs     0.00B    0.0%    0.00B
    source terms                  1.88k   2.50ms    0.0%  1.34μs     0.00B    0.0%    0.00B
  ~alg~                           1.88k    2.16s    0.6%  1.15ms   5.08KiB    0.0%    2.77B
  F! function inside              1.88k   2.54ms    0.0%  1.36μs     0.00B    0.0%    0.00B
  LM Operator inside stage 1        625   1.18ms    0.0%  1.89μs    547KiB    0.0%     896B
  LM Operator inside stage 3        625   1.16ms    0.0%  1.86μs    547KiB    0.0%     896B
  LM Operator inside stage 2        625   1.15ms    0.0%  1.84μs    547KiB    0.0%     896B
analyze solution                      2   24.1ms    0.0%  12.1ms    951KiB    0.0%   475KiB
workspace                             1   14.1ms    0.0%  14.1ms   27.6MiB    0.0%  27.6MiB
kc Constructor                        1   1.40μs    0.0%  1.40μs     0.00B    0.0%    0.00B
───────────────────────────────────────────────────────────────────────────────────────────

assume_p_const = true

───────────────────────────────────────────────────────────────────────────────────────────
                Trixi.jl                          Time                    Allocations      
                                         ───────────────────────   ────────────────────────
            Tot / % measured:                 44.3s /  99.7%           60.4MiB /  97.5%    

Section                          ncalls     time    %tot     avg     alloc    %tot      avg
───────────────────────────────────────────────────────────────────────────────────────────
alg                               1.88k    44.1s   99.9%  23.5ms   30.4MiB   51.6%  16.6KiB
  parabolic rhs!                  1.88k    15.8s   35.9%  8.45ms   8.70KiB    0.0%    4.75B
    calculate gradient            1.88k    9.22s   20.9%  4.92ms   3.33KiB    0.0%    1.82B
      volume integral             1.88k    4.07s    9.2%  2.17ms     0.00B    0.0%    0.00B
      surface integral            1.88k    2.49s    5.6%  1.33ms     0.00B    0.0%    0.00B
      interface flux              1.88k    812ms    1.8%   433μs     0.00B    0.0%    0.00B
      Jacobian                    1.88k    795ms    1.8%   424μs     0.00B    0.0%    0.00B
      prolong2interfaces          1.88k    597ms    1.4%   318μs     0.00B    0.0%    0.00B
      reset gradients             1.88k    381ms    0.9%   203μs     0.00B    0.0%    0.00B
      ~calculate gradient~        1.88k   38.9ms    0.1%  20.8μs   3.33KiB    0.0%    1.82B
      boundary flux               1.88k   19.0ms    0.0%  10.1μs     0.00B    0.0%    0.00B
      prolong2boundaries          1.88k   15.7ms    0.0%  8.39μs     0.00B    0.0%    0.00B
      prolong2mortars             1.88k   3.76ms    0.0%  2.01μs     0.00B    0.0%    0.00B
      mortar flux                 1.88k   3.24ms    0.0%  1.73μs     0.00B    0.0%    0.00B
    calculate viscous fluxes      1.88k    2.50s    5.7%  1.33ms     0.00B    0.0%    0.00B
    volume integral               1.88k    1.21s    2.7%   644μs     0.00B    0.0%    0.00B
    prolong2interfaces            1.88k    1.12s    2.5%   600μs     0.00B    0.0%    0.00B
    interface flux                1.88k    617ms    1.4%   329μs     0.00B    0.0%    0.00B
    transform variables           1.88k    471ms    1.1%   251μs     0.00B    0.0%    0.00B
    surface integral              1.88k    296ms    0.7%   158μs     0.00B    0.0%    0.00B
    reset ∂u/∂t                   1.88k    208ms    0.5%   111μs     0.00B    0.0%    0.00B
    Jacobian                      1.88k   94.7ms    0.2%  50.5μs     0.00B    0.0%    0.00B
    ~parabolic rhs!~              1.88k   48.8ms    0.1%  26.0μs   5.38KiB    0.0%    2.94B
    prolong2boundaries            1.88k   25.6ms    0.1%  13.6μs     0.00B    0.0%    0.00B
    boundary flux                 1.88k   18.0ms    0.0%  9.58μs     0.00B    0.0%    0.00B
    prolong2mortars               1.88k   3.73ms    0.0%  1.99μs     0.00B    0.0%    0.00B
    mortar flux                   1.88k   3.32ms    0.0%  1.77μs     0.00B    0.0%    0.00B
  krylov solve stage 2              625    8.99s   20.4%  14.4ms   10.9MiB   18.5%  17.9KiB
    parabolic rhs!                1.25k    6.81s   15.4%  5.45ms   8.70KiB    0.0%    7.13B
      calculate gradient          1.25k    3.42s    7.7%  2.73ms   3.33KiB    0.0%    2.73B
        volume integral           1.25k    1.41s    3.2%  1.13ms     0.00B    0.0%    0.00B
        surface integral          1.25k    986ms    2.2%   789μs     0.00B    0.0%    0.00B
        interface flux            1.25k    296ms    0.7%   237μs     0.00B    0.0%    0.00B
        prolong2interfaces        1.25k    268ms    0.6%   214μs     0.00B    0.0%    0.00B
        reset gradients           1.25k    266ms    0.6%   213μs     0.00B    0.0%    0.00B
        Jacobian                  1.25k    142ms    0.3%   114μs     0.00B    0.0%    0.00B
        ~calculate gradient~      1.25k   25.1ms    0.1%  20.1μs   3.33KiB    0.0%    2.73B
        boundary flux             1.25k   11.2ms    0.0%  8.93μs     0.00B    0.0%    0.00B
        prolong2boundaries        1.25k   8.22ms    0.0%  6.58μs     0.00B    0.0%    0.00B
        prolong2mortars           1.25k   2.66ms    0.0%  2.13μs     0.00B    0.0%    0.00B
        mortar flux               1.25k   2.24ms    0.0%  1.79μs     0.00B    0.0%    0.00B
      calculate viscous fluxes    1.25k    1.06s    2.4%   848μs     0.00B    0.0%    0.00B
      prolong2interfaces          1.25k    613ms    1.4%   490μs     0.00B    0.0%    0.00B
      volume integral             1.25k    604ms    1.4%   483μs     0.00B    0.0%    0.00B
      transform variables         1.25k    297ms    0.7%   237μs     0.00B    0.0%    0.00B
      interface flux              1.25k    272ms    0.6%   218μs     0.00B    0.0%    0.00B
      reset ∂u/∂t                 1.25k    246ms    0.6%   197μs     0.00B    0.0%    0.00B
      Jacobian                    1.25k    126ms    0.3%   101μs     0.00B    0.0%    0.00B
      surface integral            1.25k    115ms    0.3%  91.9μs     0.00B    0.0%    0.00B
      ~parabolic rhs!~            1.25k   33.2ms    0.1%  26.6μs   5.38KiB    0.0%    4.40B
      prolong2boundaries          1.25k   16.8ms    0.0%  13.4μs     0.00B    0.0%    0.00B
      boundary flux               1.25k   10.0ms    0.0%  8.00μs     0.00B    0.0%    0.00B
      prolong2mortars             1.25k   2.38ms    0.0%  1.91μs     0.00B    0.0%    0.00B
      mortar flux                 1.25k   2.26ms    0.0%  1.81μs     0.00B    0.0%    0.00B
    ~krylov solve stage 2~          625    2.18s    4.9%  3.48ms   10.9MiB   18.5%  17.9KiB
  krylov solve stage 3              625    8.38s   19.0%  13.4ms   10.9MiB   18.5%  17.9KiB
    parabolic rhs!                1.25k    6.35s   14.4%  5.08ms   8.70KiB    0.0%    7.13B
      calculate gradient          1.25k    3.24s    7.3%  2.59ms   3.33KiB    0.0%    2.73B
        volume integral           1.25k    1.33s    3.0%  1.07ms     0.00B    0.0%    0.00B
        surface integral          1.25k    949ms    2.1%   759μs     0.00B    0.0%    0.00B
        interface flux            1.25k    281ms    0.6%   225μs     0.00B    0.0%    0.00B
        prolong2interfaces        1.25k    255ms    0.6%   204μs     0.00B    0.0%    0.00B
        reset gradients           1.25k    248ms    0.6%   199μs     0.00B    0.0%    0.00B
        Jacobian                  1.25k    128ms    0.3%   102μs     0.00B    0.0%    0.00B
        ~calculate gradient~      1.25k   24.3ms    0.1%  19.4μs   3.33KiB    0.0%    2.73B
        boundary flux             1.25k   10.6ms    0.0%  8.51μs     0.00B    0.0%    0.00B
        prolong2boundaries        1.25k   7.59ms    0.0%  6.07μs     0.00B    0.0%    0.00B
        prolong2mortars           1.25k   2.52ms    0.0%  2.01μs     0.00B    0.0%    0.00B
        mortar flux               1.25k   2.21ms    0.0%  1.76μs     0.00B    0.0%    0.00B
      calculate viscous fluxes    1.25k    1.01s    2.3%   806μs     0.00B    0.0%    0.00B
      volume integral             1.25k    561ms    1.3%   449μs     0.00B    0.0%    0.00B
      prolong2interfaces          1.25k    541ms    1.2%   433μs     0.00B    0.0%    0.00B
      transform variables         1.25k    283ms    0.6%   226μs     0.00B    0.0%    0.00B
      interface flux              1.25k    238ms    0.5%   190μs     0.00B    0.0%    0.00B
      reset ∂u/∂t                 1.25k    210ms    0.5%   168μs     0.00B    0.0%    0.00B
      Jacobian                    1.25k    104ms    0.2%  83.0μs     0.00B    0.0%    0.00B
      surface integral            1.25k    100ms    0.2%  80.2μs     0.00B    0.0%    0.00B
      ~parabolic rhs!~            1.25k   32.2ms    0.1%  25.8μs   5.38KiB    0.0%    4.40B
      prolong2boundaries          1.25k   14.1ms    0.0%  11.3μs     0.00B    0.0%    0.00B
      boundary flux               1.25k   9.31ms    0.0%  7.45μs     0.00B    0.0%    0.00B
      mortar flux                 1.25k   2.39ms    0.0%  1.92μs     0.00B    0.0%    0.00B
      prolong2mortars             1.25k   2.17ms    0.0%  1.74μs     0.00B    0.0%    0.00B
    ~krylov solve stage 3~          625    2.03s    4.6%  3.25ms   10.9MiB   18.5%  17.9KiB
  krylov solve stage 1              625    5.05s   11.4%  8.08ms   5.46MiB    9.3%  8.95KiB
    parabolic rhs!                  625    3.17s    7.2%  5.07ms   8.70KiB    0.0%    14.3B
      calculate gradient            625    1.65s    3.7%  2.64ms   3.33KiB    0.0%    5.45B
        volume integral             625    672ms    1.5%  1.08ms     0.00B    0.0%    0.00B
        surface integral            625    470ms    1.1%   752μs     0.00B    0.0%    0.00B
        interface flux              625    146ms    0.3%   233μs     0.00B    0.0%    0.00B
        reset gradients             625    145ms    0.3%   232μs     0.00B    0.0%    0.00B
        prolong2interfaces          625    130ms    0.3%   208μs     0.00B    0.0%    0.00B
        Jacobian                    625   62.7ms    0.1%   100μs     0.00B    0.0%    0.00B
        ~calculate gradient~        625   12.4ms    0.0%  19.9μs   3.33KiB    0.0%    5.45B
        boundary flux               625   5.50ms    0.0%  8.80μs     0.00B    0.0%    0.00B
        prolong2boundaries          625   3.99ms    0.0%  6.38μs     0.00B    0.0%    0.00B
        prolong2mortars             625   1.25ms    0.0%  2.00μs     0.00B    0.0%    0.00B
        mortar flux                 625   1.18ms    0.0%  1.90μs     0.00B    0.0%    0.00B
      calculate viscous fluxes      625    496ms    1.1%   793μs     0.00B    0.0%    0.00B
      volume integral               625    274ms    0.6%   439μs     0.00B    0.0%    0.00B
      prolong2interfaces            625    266ms    0.6%   426μs     0.00B    0.0%    0.00B
      transform variables           625    141ms    0.3%   226μs     0.00B    0.0%    0.00B
      interface flux                625    117ms    0.3%   188μs     0.00B    0.0%    0.00B
      reset ∂u/∂t                   625   93.7ms    0.2%   150μs     0.00B    0.0%    0.00B
      surface integral              625   50.1ms    0.1%  80.1μs     0.00B    0.0%    0.00B
      Jacobian                      625   48.1ms    0.1%  77.0μs     0.00B    0.0%    0.00B
      ~parabolic rhs!~              625   15.8ms    0.0%  25.3μs   5.38KiB    0.0%    8.81B
      prolong2boundaries            625   7.09ms    0.0%  11.3μs     0.00B    0.0%    0.00B
      boundary flux                 625   4.70ms    0.0%  7.51μs     0.00B    0.0%    0.00B
      prolong2mortars               625   1.20ms    0.0%  1.92μs     0.00B    0.0%    0.00B
      mortar flux                   625   1.10ms    0.0%  1.76μs     0.00B    0.0%    0.00B
    ~krylov solve stage 1~          625    1.88s    4.3%  3.01ms   5.46MiB    9.3%  8.94KiB
  rhs!                            1.88k    3.81s    8.6%  2.03ms   4.78KiB    0.0%    2.61B
    interface flux                1.88k    1.53s    3.5%   817μs     0.00B    0.0%    0.00B
    volume integral               1.88k    1.25s    2.8%   665μs     0.00B    0.0%    0.00B
    prolong2interfaces            1.88k    386ms    0.9%   206μs     0.00B    0.0%    0.00B
    surface integral              1.88k    314ms    0.7%   168μs     0.00B    0.0%    0.00B
    reset ∂u/∂t                   1.88k    115ms    0.3%  61.4μs     0.00B    0.0%    0.00B
    Jacobian                      1.88k    107ms    0.2%  57.1μs     0.00B    0.0%    0.00B
    boundary flux                 1.88k   50.7ms    0.1%  27.0μs     0.00B    0.0%    0.00B
    ~rhs!~                        1.88k   40.3ms    0.1%  21.5μs   4.78KiB    0.0%    2.61B
    prolong2boundaries            1.88k   11.7ms    0.0%  6.25μs     0.00B    0.0%    0.00B
    prolong2mortars               1.88k   3.58ms    0.0%  1.91μs     0.00B    0.0%    0.00B
    mortar flux                   1.88k   3.25ms    0.0%  1.73μs     0.00B    0.0%    0.00B
    source terms                  1.88k   2.49ms    0.0%  1.33μs     0.00B    0.0%    0.00B
  ~alg~                           1.88k    1.98s    4.5%  1.06ms   5.08KiB    0.0%    2.77B
  Jacobian inside stage 2           625   17.8ms    0.0%  28.5μs    762KiB    1.3%  1.22KiB
  Jacobian inside stage 3           625   16.6ms    0.0%  26.5μs    762KiB    1.3%  1.22KiB
  Jacobian inside stage1            625   12.7ms    0.0%  20.3μs    762KiB    1.3%  1.22KiB
  F! function inside              1.88k   2.50ms    0.0%  1.34μs     0.00B    0.0%    0.00B
  LM Operator inside stage 3        625   1.44ms    0.0%  2.30μs    303KiB    0.5%     496B
  LM Operator inside stage 2        625   1.27ms    0.0%  2.03μs    303KiB    0.5%     496B
  LM Operator inside stage 1        625   1.22ms    0.0%  1.95μs    303KiB    0.5%     496B
analyze solution                      2   28.7ms    0.1%  14.3ms    951KiB    1.6%   475KiB
workspace                             1   27.7μs    0.0%  27.7μs   27.6MiB   46.8%  27.6MiB
kc Constructor                        1   1.96μs    0.0%  1.96μs     0.00B    0.0%    0.00B
───────────────────────────────────────────────────────────────────────────────────────────

I don't know whether the allocations, and possibly also the performance decrease, are due to parabolic rhs differentiation or there's something else.

I'm planning to introduce some splitting that do not require parabolic rhs from Trixi, to see whether this is causing the allocations. This is on Julia 1.11.6

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions