CPU 100% when it reaches step 11(gen) of process

[Angluca]

I have no idea why this is happening, Every time it's the same.

./_axe axc -o axe --release
Compiling: axc.axe
1 | IO
2 | Lex
3 | Parse
4 | Imports
5 | Lowering
6 | PFCCG I
7 | PFCCG II
8 | PFCCG III
9 | PFCCG IV
10| PFCCG V
11| GEN
^ !!! CPU 100% at 11| GEN step, Whenever I move the mouse, it gets extremely sluggish.

MacBook Pro (Retina, 13-inch, Late 2013) Macos bigsur 11.7.10
CPU - 2.4 GHz duel-core Intel Core i5 x86-64
MEM - 8 GB 1600 MHz DDR3
GRAPHIC - Intel Iris 1536 MB

Apple clang version 13.0.0 (clang-1300.0.29.30)
Target: x86_64-apple-darwin20.6.0
Thread model: posix

Other language compile file all not have this problem.

[Angluca]

parallel for issue.

axe test_compound_assign.axe -L/usr/local/opt/libomp/lib -I/usr/local/opt/libomp/include -lomp 
Compiling: test_compound_assign.axe
1 | IO
...
12| Last compilation pass
clang: error: unsupported option '-fopenmp'
clang: error: unsupported option '-fopenmp'

Old macos unspported openmp !!? What should I do? T_T
I had install libomp but can't find libomp?

[navid-m]

parallel for issue.

axe test_compound_assign.axe -L/usr/local/opt/libomp/lib -I/usr/local/opt/libomp/include -lomp 
Compiling: test_compound_assign.axe
1 | IO
...
12| Last compilation pass
clang: error: unsupported option '-fopenmp'
clang: error: unsupported option '-fopenmp'

Old macos unspported openmp !!? What should I do? T_T
I had install libomp but can't find libomp?

Try installing llvm toolchain via brew and then using THAT clang instead of system clang.

So run:

brew install llvm

And then use this clang:

/usr/local/opt/llvm/bin/clang

I think we will need an option to specify path of C compiler though because atm it just chooses whatever is in PATH.

[Angluca]

My machine is not suitable for installing the latest LLVM (clang14), the installation process will take more than 8 hours. Therefore I don't want to install it. T_T

I've comment
//compile_cmd = concat(compile_cmd, str(" -fopenmp"));
axe ./test_compound_assign -r compile & run success

Perhaps can add axe cmd param ignore parallel ( like this --no_parallel :)

[Angluca]

Personally, For variables that are not used, it is sufficient to issue a warning rather than causing an error.
When compiling with Rust, if there are unused variables, it will only issue a warning too. Ultimately the outcome still depends on the player's choices, Don't be too strict. 🤣

[navid-m]

Personally, For variables that are not used, it is sufficient to issue a warning rather than causing an error. When compiling with Rust, if there are unused variables, it will only issue a warning too. Ultimately the outcome still depends on the player's choices, Don't be too strict. 🤣

I'm going to keep it the same, dead-code elimination is a great feature that I want to keep.

For me, this 100% bug doesnt happen on Ryzen 5600X, but I assume on other CPUs it wont be the same, this is compiling around 20K lines of code:

---

The problem is generate_c function in renderer.axe uses recursion and needs optimizations. I'm open to PRs about this because I have tried optimizing it to some extent, and still the "GEN" stage is the slowest one relative to the sub 100ms other sections.

[Angluca]

Because this is an old machine, it is quite easy to identify which operations will consume more CPU resources. If use axe compile a project requires a very long compilation time, I am really worried that the computer might crash during the half-hour compilation process. 😂

Perhaps it is because there are too many recursive calls that the CPU consumption is high. Recursion seems to mainly consume stack space. If a large project is compiled for a long time and the stack space is exhausted, it might crash, I'm not very sure about this.

I noticed that many examples in self_tests involve errors caused by unused variables. That's why I said it would be sufficient to just give warnings.

[Angluca]

saw cleancan add clean *.dSYM, axe compile file will gen xxx.dSYM every time.

[navid-m]

That's already done in the saw repo, I think the original issue is fixed with the 100% CPU after this commit: ec3cf51, does it still occur for you?

[Angluca]

Still the same CPU 100%+ at 11| GEN

kernel_task 170+ // Screenshots are also very difficult