use SnoopPrecompile #1234
Draft
use SnoopPrecompile #1234
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Codecov Report
@@ Coverage Diff @@
## main #1234 +/- ##
==========================================
+ Coverage 87.50% 97.48% +9.98%
==========================================
Files 329 325 -4
Lines 25186 24921 -265
==========================================
+ Hits 22037 24292 +2255
+ Misses 3149 629 -2520
Flags with carried forward coverage won't be shown. Click here to find out more.
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. |
|
Results from Rocinante:
|
| ID | time ratio | memory ratio |
|---|---|---|
["tree_3d_dgsem/elixir_euler_mortar.jl", "p3_rhs!"] |
1.02 (5%) | 4.58 (1%) ❌ |
["tree_3d_dgsem/elixir_euler_mortar.jl", "p7_rhs!"] |
1.10 (5%) ❌ | 1.00 (1%) |
["unstructured_2d_dgsem/elixir_euler_wall_bc.jl", "p3_analysis"] |
0.85 (5%) ✅ | 1.00 (1%) |
2 Threads:
Job Properties
- Time of benchmarks:
- Target: 2 Nov 2022 - 10:00
- Baseline: 2 Nov 2022 - 10:29
- Package commits:
- Target: ca6b00
- Baseline: 3ea50d
- Julia commits:
- Target: 36034a
- Baseline: 36034a
- Julia command flags:
- Target:
-Cnative,-J/mnt/hd1/opt/julia/1.8.2/lib/julia/sys.so,-g1,--check-bounds=no,--threads=2 - Baseline:
-Cnative,-J/mnt/hd1/opt/julia/1.8.2/lib/julia/sys.so,-g1,--check-bounds=no,--threads=2
- Target:
- Environment variables:
- Target: None
- Baseline: None
Results
A ratio greater than 1.0 denotes a possible regression (marked with ❌), while a ratio less
than 1.0 denotes a possible improvement (marked with ✅). Only significant results - results
that indicate possible regressions or improvements - are shown below (thus, an empty table means that all
benchmark results remained invariant between builds).
| ID | time ratio | memory ratio |
|---|---|---|
["p4est_2d_dgsem/elixir_advection_extended.jl", "p3_analysis"] |
0.94 (5%) ✅ | 1.00 (1%) |
["tree_2d_dgsem/elixir_advection_amr_nonperiodic.jl", "p3_rhs!"] |
1.10 (5%) ❌ | 1.00 (1%) |
["tree_2d_dgsem/elixir_euler_vortex_mortar_shockcapturing.jl", "p7_rhs!"] |
1.05 (5%) ❌ | 1.00 (1%) |
["tree_3d_dgsem/elixir_euler_mortar.jl", "p3_rhs!"] |
1.03 (5%) | 1.39 (1%) ❌ |
["unstructured_2d_dgsem/elixir_euler_wall_bc.jl", "p3_analysis"] |
1.06 (5%) ❌ | 1.00 (1%) |
|
New timings (in seconds) obtained from julia --project=. --threads=1 -e '
@time @eval(using OrdinaryDiffEq);
@time @eval(using Trixi);
@time @eval trixi_include(joinpath(examples_dir(), "tree_3d_dgsem", "elixir_euler_taylor_green_vortex.jl"), maxiters=200)'and julia --project=. --threads=1 -e '
@time @eval(using OrdinaryDiffEq);
@time @eval(using Trixi);
@time @eval trixi_include(joinpath(examples_dir(), "tree_3d_dgsem", "elixir_mhd_alfven_wave.jl"), maxiters=200)'(and faking a modification to Trixi to trigger precompilation in a separate Julia session, recording the time reported for Julia v1.8.3
So it looks like it pays off if you are mainly interested in compressible Euler and do not need to precompile Trixi again every time you start a new session. |
|
@sloede Benchmarks on Rocinante are running. Do we want to wait for them and decide how to proceed with this PR based on the results? |
|
For reference, here is the flamegraph of tinf = @snoopi_deep trixi_include(joinpath(examples_dir(), "tree_3d_dgsem", "elixir_euler_taylor_green_vortex.jl"), maxiters=200)in this PR
|
|
Results from Rocinante:
|
| ID | time ratio | memory ratio |
|---|---|---|
["latency", "polydeg_7"] |
1.05 (5%) ❌ | 1.00 (1%) |
["tree_3d_dgsem/elixir_euler_mortar.jl", "p3_rhs!"] |
1.00 (5%) | 4.58 (1%) ❌ |
Benchmark Group List
Here's a list of all the benchmark groups executed by this job:
["benchmark/elixir_2d_euler_vortex_p4est.jl"]["benchmark/elixir_2d_euler_vortex_structured.jl"]["benchmark/elixir_2d_euler_vortex_tree.jl"]["benchmark/elixir_2d_euler_vortex_unstructured.jl"]["latency"]["p4est_2d_dgsem/elixir_advection_extended.jl"]["p4est_3d_dgsem/elixir_advection_basic.jl"]["structured_2d_dgsem/elixir_advection_extended.jl"]["structured_2d_dgsem/elixir_advection_nonperiodic.jl"]["structured_2d_dgsem/elixir_euler_ec.jl"]["structured_2d_dgsem/elixir_euler_source_terms_nonperiodic.jl"]["structured_2d_dgsem/elixir_mhd_ec.jl"]["structured_3d_dgsem/elixir_advection_nonperiodic_curved.jl"]["structured_3d_dgsem/elixir_euler_ec.jl"]["structured_3d_dgsem/elixir_euler_source_terms_nonperiodic_curved.jl"]["structured_3d_dgsem/elixir_mhd_ec.jl"]["tree_2d_dgsem/elixir_advection_amr_nonperiodic.jl"]["tree_2d_dgsem/elixir_advection_extended.jl"]["tree_2d_dgsem/elixir_euler_ec.jl"]["tree_2d_dgsem/elixir_euler_vortex_mortar.jl"]["tree_2d_dgsem/elixir_euler_vortex_mortar_shockcapturing.jl"]["tree_2d_dgsem/elixir_mhd_ec.jl"]["tree_3d_dgsem/elixir_advection_extended.jl"]["tree_3d_dgsem/elixir_euler_ec.jl"]["tree_3d_dgsem/elixir_euler_mortar.jl"]["tree_3d_dgsem/elixir_euler_shockcapturing.jl"]["tree_3d_dgsem/elixir_mhd_ec.jl"]["unstructured_2d_dgsem/elixir_euler_wall_bc.jl"]
Julia versioninfo
Target
Julia Version 1.8.3
Commit 0434deb161e (2022-11-14 20:14 UTC)
Platform Info:
OS: Linux (x86_64-linux-gnu)
Ubuntu 20.04.2 LTS
uname: Linux 5.4.0-132-generic #148-Ubuntu SMP Mon Oct 17 16:02:06 UTC 2022 x86_64 x86_64
CPU: AMD Ryzen Threadripper 3990X 64-Core Processor:
speed user nice sys idle irq
#1-128 2195 MHz 830037 s 1284 s 31237 s 126067450 s 0 s
Memory: 251.6267318725586 GB (242185.2109375 MB free)
Uptime: 99173.72 sec
Load Avg: 1.29 2.11 1.73
WORD_SIZE: 64
LIBM: libopenlibm
LLVM: libLLVM-13.0.1 (ORCJIT, znver2)
Threads: 1 on 128 virtual cores
Baseline
Julia Version 1.8.3
Commit 0434deb161e (2022-11-14 20:14 UTC)
Platform Info:
OS: Linux (x86_64-linux-gnu)
Ubuntu 20.04.2 LTS
uname: Linux 5.4.0-132-generic #148-Ubuntu SMP Mon Oct 17 16:02:06 UTC 2022 x86_64 x86_64
CPU: AMD Ryzen Threadripper 3990X 64-Core Processor:
speed user nice sys idle irq
#1-128 2170 MHz 847937 s 1284 s 32567 s 128343469 s 0 s
Memory: 251.6267318725586 GB (242468.421875 MB free)
Uptime: 100966.98 sec
Load Avg: 1.0 1.0 1.08
WORD_SIZE: 64
LIBM: libopenlibm
LLVM: libLLVM-13.0.1 (ORCJIT, znver2)
Threads: 1 on 128 virtual cores
2 Threads:
Job Properties
- Time of benchmarks:
- Target: 2 Dec 2022 - 18:41
- Baseline: 2 Dec 2022 - 19:13
- Package commits:
- Target: 4e83d0
- Baseline: 94703d
- Julia commits:
- Target: 0434de
- Baseline: 0434de
- Julia command flags:
- Target:
-Cnative,-J/mnt/hd1/opt/julia/1.8.3/lib/julia/sys.so,-g1,--check-bounds=no,--threads=2 - Baseline:
-Cnative,-J/mnt/hd1/opt/julia/1.8.3/lib/julia/sys.so,-g1,--check-bounds=no,--threads=2
- Target:
- Environment variables:
- Target: None
- Baseline: None
Results
A ratio greater than 1.0 denotes a possible regression (marked with ❌), while a ratio less
than 1.0 denotes a possible improvement (marked with ✅). Only significant results - results
that indicate possible regressions or improvements - are shown below (thus, an empty table means that all
benchmark results remained invariant between builds).
| ID | time ratio | memory ratio |
|---|---|---|
["latency", "polydeg_7"] |
1.05 (5%) ❌ | 1.00 (1%) |
["structured_3d_dgsem/elixir_mhd_ec.jl", "p3_rhs!"] |
0.87 (5%) ✅ | 1.00 (1%) |
["tree_2d_dgsem/elixir_advection_amr_nonperiodic.jl", "p3_rhs!"] |
1.16 (5%) ❌ | 1.00 (1%) |
["tree_3d_dgsem/elixir_euler_mortar.jl", "p3_rhs!"] |
0.98 (5%) | 1.39 (1%) ❌ |
["unstructured_2d_dgsem/elixir_euler_wall_bc.jl", "p3_analysis"] |
0.92 (5%) ✅ | 1.00 (1%) |
Benchmark Group List
Here's a list of all the benchmark groups executed by this job:
["benchmark/elixir_2d_euler_vortex_p4est.jl"]["benchmark/elixir_2d_euler_vortex_structured.jl"]["benchmark/elixir_2d_euler_vortex_tree.jl"]["benchmark/elixir_2d_euler_vortex_unstructured.jl"]["latency"]["p4est_2d_dgsem/elixir_advection_extended.jl"]["p4est_3d_dgsem/elixir_advection_basic.jl"]["structured_2d_dgsem/elixir_advection_extended.jl"]["structured_2d_dgsem/elixir_advection_nonperiodic.jl"]["structured_2d_dgsem/elixir_euler_ec.jl"]["structured_2d_dgsem/elixir_euler_source_terms_nonperiodic.jl"]["structured_2d_dgsem/elixir_mhd_ec.jl"]["structured_3d_dgsem/elixir_advection_nonperiodic_curved.jl"]["structured_3d_dgsem/elixir_euler_ec.jl"]["structured_3d_dgsem/elixir_euler_source_terms_nonperiodic_curved.jl"]["structured_3d_dgsem/elixir_mhd_ec.jl"]["tree_2d_dgsem/elixir_advection_amr_nonperiodic.jl"]["tree_2d_dgsem/elixir_advection_extended.jl"]["tree_2d_dgsem/elixir_euler_ec.jl"]["tree_2d_dgsem/elixir_euler_vortex_mortar.jl"]["tree_2d_dgsem/elixir_euler_vortex_mortar_shockcapturing.jl"]["tree_2d_dgsem/elixir_mhd_ec.jl"]["tree_3d_dgsem/elixir_advection_extended.jl"]["tree_3d_dgsem/elixir_euler_ec.jl"]["tree_3d_dgsem/elixir_euler_mortar.jl"]["tree_3d_dgsem/elixir_euler_shockcapturing.jl"]["tree_3d_dgsem/elixir_mhd_ec.jl"]["unstructured_2d_dgsem/elixir_euler_wall_bc.jl"]
Julia versioninfo
Target
Julia Version 1.8.3
Commit 0434deb161e (2022-11-14 20:14 UTC)
Platform Info:
OS: Linux (x86_64-linux-gnu)
Ubuntu 20.04.2 LTS
uname: Linux 5.4.0-132-generic #148-Ubuntu SMP Mon Oct 17 16:02:06 UTC 2022 x86_64 x86_64
CPU: AMD Ryzen Threadripper 3990X 64-Core Processor:
speed user nice sys idle irq
#1-128 2175 MHz 869765 s 1284 s 33956 s 130855580 s 0 s
Memory: 251.6267318725586 GB (241254.19140625 MB free)
Uptime: 102947.8 sec
Load Avg: 1.25 1.21 1.12
WORD_SIZE: 64
LIBM: libopenlibm
LLVM: libLLVM-13.0.1 (ORCJIT, znver2)
Threads: 2 on 128 virtual cores
Baseline
Julia Version 1.8.3
Commit 0434deb161e (2022-11-14 20:14 UTC)
Platform Info:
OS: Linux (x86_64-linux-gnu)
Ubuntu 20.04.2 LTS
uname: Linux 5.4.0-132-generic #148-Ubuntu SMP Mon Oct 17 16:02:06 UTC 2022 x86_64 x86_64
CPU: AMD Ryzen Threadripper 3990X 64-Core Processor:
speed user nice sys idle irq
#1-128 2172 MHz 891390 s 1284 s 35311 s 133332872 s 0 s
Memory: 251.6267318725586 GB (241550.40625 MB free)
Uptime: 104901.23 sec
Load Avg: 1.2 1.2 1.12
WORD_SIZE: 64
LIBM: libopenlibm
LLVM: libLLVM-13.0.1 (ORCJIT, znver2)
Threads: 2 on 128 virtual cores
There seems to be a regression of tree_3d_dgsem/elixir_euler_mortar.jl...
|
@sloede It would be great if you could also perform some benchmarks. |
Sure. Anything in particular you think would be useful? |
|
I guess your "standard benchmark setups" you have used elsewhere are already a good start. In addition, I would like to check
|
|
Results from Rocinante (with up-to-date packages) with timings in seconds obtained from julia --project=. --threads=1 -e '
@time @eval(using OrdinaryDiffEq);
@time @eval(using Trixi);
@time @eval trixi_include(joinpath(examples_dir(), "tree_3d_dgsem", "elixir_euler_taylor_green_vortex.jl"), maxiters=200)'julia --project=. --threads=1 -e '
@time @eval(using OrdinaryDiffEq);
@time @eval(using Trixi);
@time @eval trixi_include(joinpath(examples_dir(), "tree_3d_dgsem", "elixir_euler_mortar.jl"), maxiters=200)'julia --project=. --threads=1 -e '
@time @eval(using OrdinaryDiffEq);
@time @eval(using Trixi);
@time @eval trixi_include(joinpath(examples_dir(), "tree_3d_dgsem", "elixir_mhd_alfven_wave.jl"), maxiters=200)'(and faking a modification to Trixi to trigger precompilation in a separate Julia session, recording the time reported for Julia v1.8.3
The gain for
julia> using OrdinaryDiffEq, Trixi
julia> trixi_include(joinpath(examples_dir(), "tree_3d_dgsem", "elixir_euler_mortar.jl"), maxiters=200) # 2nd time
[...]
────────────────────────────────────────────────────────────────────────────────────
Trixi.jl Time Allocations
─────────────────────── ────────────────────────
Tot / % measured: 465ms / 97.2% 3.34MiB / 90.3%
Section ncalls time %tot avg alloc %tot avg
────────────────────────────────────────────────────────────────────────────────────
rhs! 636 437ms 96.6% 687μs 248KiB 8.0% 399B
interface flux 636 104ms 23.1% 164μs 0.00B 0.0% 0.00B
volume integral 636 82.7ms 18.3% 130μs 0.00B 0.0% 0.00B
prolong2interfaces 636 76.3ms 16.9% 120μs 0.00B 0.0% 0.00B
source terms 636 75.1ms 16.6% 118μs 238KiB 7.7% 384B
surface integral 636 38.2ms 8.4% 60.0μs 0.00B 0.0% 0.00B
mortar flux 636 32.1ms 7.1% 50.5μs 0.00B 0.0% 0.00B
prolong2mortars 636 22.2ms 4.9% 34.9μs 0.00B 0.0% 0.00B
reset ∂u/∂t 636 2.56ms 0.6% 4.03μs 0.00B 0.0% 0.00B
Jacobian 636 2.46ms 0.5% 3.87μs 0.00B 0.0% 0.00B
~rhs!~ 636 692μs 0.2% 1.09μs 9.33KiB 0.3% 15.0B
prolong2boundaries 636 34.3μs 0.0% 53.9ns 0.00B 0.0% 0.00B
boundary flux 636 13.8μs 0.0% 21.7ns 0.00B 0.0% 0.00B
calculate dt 128 5.87ms 1.3% 45.9μs 0.00B 0.0% 0.00B
analyze solution 3 5.29ms 1.2% 1.76ms 53.2KiB 1.7% 17.7KiB
I/O 4 4.36ms 1.0% 1.09ms 2.72MiB 90.2% 696KiB
save solution 3 3.44ms 0.8% 1.15ms 2.69MiB 89.2% 918KiB
~I/O~ 4 914μs 0.2% 228μs 24.2KiB 0.8% 6.06KiB
get element variables 3 10.7μs 0.0% 3.57μs 6.42KiB 0.2% 2.14KiB
save mesh 3 120ns 0.0% 40.0ns 0.00B 0.0% 0.00B
────────────────────────────────────────────────────────────────────────────────────This PR: julia> using OrdinaryDiffEq, Trixi
julia> trixi_include(joinpath(examples_dir(), "tree_3d_dgsem", "elixir_euler_mortar.jl"), maxiters=200) # 2nd time
[...]
────────────────────────────────────────────────────────────────────────────────────
Trixi.jl Time Allocations
─────────────────────── ────────────────────────
Tot / % measured: 459ms / 97.2% 4.17MiB / 92.3%
Section ncalls time %tot avg alloc %tot avg
────────────────────────────────────────────────────────────────────────────────────
rhs! 636 431ms 96.5% 678μs 1.08MiB 28.0% 1.73KiB
interface flux 636 103ms 23.1% 162μs 0.00B 0.0% 0.00B
volume integral 636 81.8ms 18.3% 129μs 229KiB 5.8% 368B
prolong2interfaces 636 75.4ms 16.9% 118μs 0.00B 0.0% 0.00B
source terms 636 72.3ms 16.2% 114μs 238KiB 6.0% 384B
surface integral 636 38.3ms 8.6% 60.2μs 229KiB 5.8% 368B
mortar flux 636 31.8ms 7.1% 50.1μs 0.00B 0.0% 0.00B
prolong2mortars 636 22.1ms 5.0% 34.8μs 0.00B 0.0% 0.00B
reset ∂u/∂t 636 2.61ms 0.6% 4.11μs 189KiB 4.8% 304B
Jacobian 636 2.50ms 0.6% 3.93μs 209KiB 5.3% 336B
~rhs!~ 636 717μs 0.2% 1.13μs 9.33KiB 0.2% 15.0B
prolong2boundaries 636 31.6μs 0.0% 49.7ns 0.00B 0.0% 0.00B
boundary flux 636 14.0μs 0.0% 22.0ns 0.00B 0.0% 0.00B
calculate dt 128 5.83ms 1.3% 45.5μs 0.00B 0.0% 0.00B
analyze solution 3 5.28ms 1.2% 1.76ms 57.2KiB 1.5% 19.1KiB
I/O 4 4.39ms 1.0% 1.10ms 2.72MiB 70.6% 696KiB
save solution 3 3.50ms 0.8% 1.17ms 2.69MiB 69.8% 918KiB
~I/O~ 4 877μs 0.2% 219μs 24.2KiB 0.6% 6.06KiB
get element variables 3 12.9μs 0.0% 4.29μs 6.42KiB 0.2% 2.14KiB
save mesh 3 130ns 0.0% 43.3ns 0.00B 0.0% 0.00BJulia 1.10.0-DEV.98 (2022-12-01), vc/external_functions/c244b01f148 from JuliaLang/julia#47184
julia> using OrdinaryDiffEq, Trixi
julia> trixi_include(joinpath(examples_dir(), "tree_3d_dgsem", "elixir_euler_mortar.jl"), maxiters=200) # 2nd time
[...]
────────────────────────────────────────────────────────────────────────────────────
Trixi.jl Time Allocations
─────────────────────── ────────────────────────
Tot / % measured: 491ms / 97.3% 3.33MiB / 90.3%
Section ncalls time %tot avg alloc %tot avg
────────────────────────────────────────────────────────────────────────────────────
rhs! 636 462ms 96.5% 726μs 248KiB 8.0% 399B
interface flux 636 106ms 22.2% 167μs 0.00B 0.0% 0.00B
source terms 636 92.9ms 19.4% 146μs 238KiB 7.7% 384B
volume integral 636 87.5ms 18.3% 138μs 0.00B 0.0% 0.00B
prolong2interfaces 636 80.6ms 16.8% 127μs 0.00B 0.0% 0.00B
surface integral 636 36.9ms 7.7% 58.0μs 0.00B 0.0% 0.00B
mortar flux 636 29.6ms 6.2% 46.6μs 0.00B 0.0% 0.00B
prolong2mortars 636 22.1ms 4.6% 34.7μs 0.00B 0.0% 0.00B
Jacobian 636 2.50ms 0.5% 3.93μs 0.00B 0.0% 0.00B
reset ∂u/∂t 636 2.48ms 0.5% 3.90μs 0.00B 0.0% 0.00B
~rhs!~ 636 726μs 0.2% 1.14μs 9.33KiB 0.3% 15.0B
prolong2boundaries 636 32.7μs 0.0% 51.4ns 0.00B 0.0% 0.00B
boundary flux 636 15.0μs 0.0% 23.6ns 0.00B 0.0% 0.00B
analyze solution 3 6.44ms 1.3% 2.15ms 53.2KiB 1.7% 17.7KiB
calculate dt 128 5.88ms 1.2% 46.0μs 0.00B 0.0% 0.00B
I/O 4 4.50ms 0.9% 1.12ms 2.72MiB 90.2% 695KiB
save solution 3 3.65ms 0.8% 1.22ms 2.69MiB 89.2% 917KiB
~I/O~ 4 828μs 0.2% 207μs 24.1KiB 0.8% 6.02KiB
get element variables 3 23.7μs 0.0% 7.90μs 7.12KiB 0.2% 2.38KiB
save mesh 3 200ns 0.0% 66.7ns 0.00B 0.0% 0.00B
────────────────────────────────────────────────────────────────────────────────────This PR: julia> using OrdinaryDiffEq, Trixi
julia> trixi_include(joinpath(examples_dir(), "tree_3d_dgsem", "elixir_euler_mortar.jl"), maxiters=200) # 2nd time
[...]
────────────────────────────────────────────────────────────────────────────────────
Trixi.jl Time Allocations
─────────────────────── ────────────────────────
Tot / % measured: 5.89s / 99.7% 1.03GiB / 100.0%
Section ncalls time %tot avg alloc %tot avg
────────────────────────────────────────────────────────────────────────────────────
rhs! 636 5.83s 99.3% 9.17ms 1.02GiB 99.3% 1.64MiB
prolong2mortars 636 2.73s 46.4% 4.29ms 523MiB 49.6% 842KiB
mortar flux 636 2.71s 46.1% 4.26ms 523MiB 49.6% 842KiB
interface flux 636 107ms 1.8% 168μs 0.00B 0.0% 0.00B
volume integral 636 88.5ms 1.5% 139μs 0.00B 0.0% 0.00B
prolong2interfaces 636 80.5ms 1.4% 127μs 0.00B 0.0% 0.00B
source terms 636 74.0ms 1.3% 116μs 238KiB 0.0% 384B
surface integral 636 39.9ms 0.7% 62.7μs 0.00B 0.0% 0.00B
reset ∂u/∂t 636 2.51ms 0.0% 3.95μs 0.00B 0.0% 0.00B
Jacobian 636 2.50ms 0.0% 3.94μs 0.00B 0.0% 0.00B
~rhs!~ 636 1.31ms 0.0% 2.06μs 9.33KiB 0.0% 15.0B
prolong2boundaries 636 49.3μs 0.0% 77.6ns 0.00B 0.0% 0.00B
boundary flux 636 14.2μs 0.0% 22.3ns 0.00B 0.0% 0.00B
analyze solution 3 32.0ms 0.5% 10.7ms 4.98MiB 0.5% 1.66MiB
calculate dt 128 5.92ms 0.1% 46.2μs 0.00B 0.0% 0.00B
I/O 4 5.04ms 0.1% 1.26ms 2.72MiB 0.3% 695KiB
save solution 3 4.16ms 0.1% 1.39ms 2.69MiB 0.3% 917KiB
~I/O~ 4 854μs 0.0% 213μs 24.0KiB 0.0% 6.00KiB
get element variables 3 30.8μs 0.0% 10.3μs 7.12KiB 0.0% 2.38KiB
save mesh 3 300ns 0.0% 100ns 0.00B 0.0% 0.00B
────────────────────────────────────────────────────────────────────────────────────So there seems to be a type-instability on this version of Julia with more precompilation for the mortar stuff. I think it may be related to JuliaLang/julia#35800, JuliaLang/julia#32552, JuliaLang/julia#41740. It looks like tho workaround Lines 264 to 282 in 0a24621 does not work anymore with this Julia update and increased precompilation from this PR...
|
|
@sloede I added a TODO note above. I will not change anything here before we get your results (to make them comparable). Afterwards, I will try the idea outlined above to see whether this fixes the problem with the upcoming Julia update and increased precompilation from this PR. |
|
Let's do it like this: We can decide how to proceed when we have your results. If you want to merge this, I will try to fix the type stability regression. Otherwise, we I will not touch this anymore but leave it open if we ever decide to change something in the future. |
|
On a Macbook M1: Julia v1.8.3
Julia 1.10.0-DEV.98 (2022-12-01), vc/external_functions/c244b01f148 from JuliaLang/julia#47184No numbers because of Apple's horrific view of its developer/users that does not allow to run unsigned/self-signed binaries from the internet 🤬: |
|
Thanks, @sloede! It does look like it's less interesting on your Mac system, so I will not work on this further at the moment. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This uses some new tools for precompilation. I get the following results with Julia v1.8.2 for the workload
The timings are in seconds on my workstation (for the first commit in this PR).
mainusing OrdinaryDiffEqusing Trixitrixi_include(...)Second commit (with updated dependency versions, after merge of
main):mainusing OrdinaryDiffEqusing Trixitrixi_include(...)TODO
Before we merge this, we should