Bug 206094 – Add Lightweight Optimization Level (LOL) to the compilation pipeline

• Component: JavaScriptCore
• Status: NEW

[Saam Barati]

original plan try to do this as a replacement of the baseline.

[Saam Barati]

Created attachment 387969 [details]
WIP

it begins

[Saam Barati]

We're probably going to change directions here and invent a new optimizer that works over bytecode. The main optimizations we're planning on doing are type speculation and register allocation. To get fast compile times, the goal will also be to not have to construct an entirely new IR. We'd also want to get to the point where each opcode can speculate on what its inputs are without doing some kind of prediction propagation phase. Many opcodes already have enough metadata to do this. However, not all do. We'll have to invent ways for all to do this.

Our reasoning for this was based on this data. Here is the JS2 score in various configurations of JSC:

[A] Baseline only: 67
[B] DFG only: 137
[C] DFG only, no inlining, no access inlining, AI says each structure set is infinite, but each variable still has a flush format: 116
[D] DFG only, no inlining, no access inlining, AI says each structure set is infinite, each variable has JSValue flush format: 109
[E] DFG only, no inlining, no access inlining, AI says each structure set is infinite, each variable has JSValue flush format, no intrinsic inlining: 106


Here are our compile time numbers on a large subset of JS2 in [E]:
                   Baseline Compile Time: 171.968 ms
                        DFG Compile Time: 1396.220 ms
                   FTL (B3) Compile Time: 0.000 ms
                  FTL (DFG) Compile Time: 0.000 ms
                        FTL Compile Time: 0.000 ms
                      Total Compile Time: 1568.189 ms
[DFG] bytecode parser total ms: 380.117813
[DFG] live catch variable preservation phase total ms: 1.461701
[DFG] CPS rethreading total ms: 48.442496
[DFG] unification total ms: 11.940636
[DFG] prediction injection total ms: 2.781105
[DFG] static execution count estimation total ms: 43.239496
[DFG] backwards propagation total ms: 22.783780
[DFG] prediction propagation total ms: 51.928968
[DFG] fixup total ms: 55.983858
[DFG] invalidation point injection total ms: 24.982650
[DFG] control flow analysis total ms: 101.180000
[DFG] tier-up check injection total ms: 0.332139
[DFG] fast store barrier insertion total ms: 22.395290
[DFG] dead code elimination total ms: 42.648463
[DFG] phantom insertion total ms: 47.266917
[DFG] stack layout total ms: 12.130792
[DFG] virtual register allocation total ms: 10.887650
[DFG] watchpoint collection total ms: 4.037530
[DFG] machine code generation total ms: 259.451395

Just building the IR and generating machine code makes us > 2x slower than baseline compile times. As seen in [E], we get a lot of benefit just from register allocation and type speculation. Having a dedicated "IR" for making this fast could lead to really good results. The IR here could likely just be bit vectors for block boundaries representing the state of types. Or perhaps, we'd just need one bit vector we update as we parse a block. The goal would be able to do this over bytecode itself. This also had a nice OSR exit story, since if we're compiling the bytecode, OSR exit is straight forward.

[Saam Barati]

I think we're gonna do this building off of baseline JIT. This allows us to incrementally improve the JIT without implementing all opcodes in optimized form.

[Saam Barati]

Created attachment 402617 [details]
WIP

it begins round 2

[Saam Barati]

Created attachment 402702 [details]
WIP

[Filip Pizlo]

Comment on attachment 402702 [details]
WIP

View in context: https://bugs.webkit.org/attachment.cgi?id=402702&action=review

> Source/JavaScriptCore/jit/JIT.cpp:63
> +void JIT::stepOverInstruction(const Instruction* instruction, const HashMap<ValueProfile*, SpeculatedType>& profiles)

Not sure you're going to want that to be a hashtable.

[Saam Barati]

(In reply to Filip Pizlo from comment #6)
> Comment on attachment 402702 [details]
> WIP
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=402702&action=review
> 
> > Source/JavaScriptCore/jit/JIT.cpp:63
> > +void JIT::stepOverInstruction(const Instruction* instruction, const HashMap<ValueProfile*, SpeculatedType>& profiles)
> 
> Not sure you're going to want that to be a hashtable.

Yeah I kinda doubted this too, and questioned it when I first started. The other thing I convinced myself is ok to do is to just racily read out the prediction. Since the prediction should only get wider as we read it, it shouldn't cause the algorithm to be unsound.

[Saam Barati]

Created attachment 402811 [details]
WIP

[Saam Barati]

Created attachment 402832 [details]
WIP

[Saam Barati]

Created attachment 402886 [details]
WIP

[Saam Barati]

Created attachment 402928 [details]
WIP

Ok, I think the prediction propagator is good enough to start working on other parts of the system.

[Saam Barati]

Created attachment 402930 [details]
WIP

[Saam Barati]

Created attachment 403159 [details]
WIP

[Saam Barati]

Created attachment 403342 [details]
WIP

It can now speculate and register allocate this program:

```
function foo(x) {
    x += 2;
    let y;
    if (x > 5) {
        y = 42;
    } else {
        y = 77;
    }

    x = x + 48;
    x = x + 1337;
    x = x + 42;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;
    x += y;

    return x + y;
}
```

Unsurprisingly, runs 2x faster than old baseline code. Seems to compile in a similar amount of time.

[Saam Barati]

Created attachment 403811 [details]
WIP

[Saam Barati]

Created attachment 403839 [details]
WIP

[Saam Barati]

Created attachment 403841 [details]
WIP

[Saam Barati]

Created attachment 403941 [details]
WIP

[Saam Barati]

Created attachment 404019 [details]
WIP

Starting to do some basic double math. It's ~2x faster also.

[Saam Barati]

Created attachment 404037 [details]
WIP

[Saam Barati]

Created attachment 404517 [details]
WIP

[Saam Barati]

Created attachment 409399 [details]
WIP

rebased

[Saam Barati]

Created attachment 409401 [details]
WIP

[Saam Barati]

Created attachment 409417 [details]
WIP

[Saam Barati]

Created attachment 409625 [details]
WIP

[Saam Barati]

Created attachment 410075 [details]
WIP

[Saam Barati]

Created attachment 410146 [details]
WIP

Getting silent refill working

[Saam Barati]

Created attachment 410302 [details]
WIP

[Saam Barati]

Created attachment 410403 [details]
WIP

[Saam Barati]

Created attachment 410575 [details]
WIP

[Saam Barati]

Created attachment 410602 [details]
WIP

[Saam Barati]

Created attachment 410723 [details]
WIP

[Saam Barati]

Created attachment 410769 [details]
WIP

[Saam Barati]

Created attachment 410808 [details]
WIP

[Saam Barati]

Created attachment 411814 [details]
WIP

[Saam Barati]

Created attachment 411823 [details]
WIP

[Saam Barati]

Created attachment 411946 [details]
WIP

[Saam Barati]

<rdar://problem/69217617>

[Saam Barati]

Created attachment 412345 [details]
WIP

[Saam Barati]

Created attachment 412378 [details]
WIP

[Saam Barati]

Created attachment 412481 [details]
WIP

[Saam Barati]

Created attachment 412484 [details]
WIP

[Saam Barati]

Created attachment 412990 [details]
WIP

[Saam Barati]

Created attachment 413624 [details]
WIP

[Saam Barati]

Created attachment 413747 [details]
WIP

[Saam Barati]

Created attachment 413758 [details]
WIP

[Saam Barati]

Created attachment 413979 [details]
WIP

[Saam Barati]

Created attachment 413980 [details]
WIP

[Saam Barati]

Created attachment 414002 [details]
WIP

[Saam Barati]

Created attachment 414089 [details]
WIP

[Saam Barati]

Created attachment 414115 [details]
WIP

[Saam Barati]

Created attachment 415161 [details]
WIP

[Saam Barati]

Created attachment 415270 [details]
WIP

[Saam Barati]

Created attachment 415356 [details]
WIP

[Saam Barati]

Created attachment 415380 [details]
WIP

[Saam Barati]

Created attachment 415471 [details]
WIP

[Saam Barati]

Created attachment 415476 [details]
WIP

Fixing a bunch of bugs in the baseline JIT when run without LOL

[Saam Barati]

Created attachment 415479 [details]
WIP

[Saam Barati]

Created attachment 415481 [details]
WIP

[Saam Barati]

Created attachment 415596 [details]
WIP

it's starting to run some of our tests

[Saam Barati]

Created attachment 415605 [details]
WIP

Fixing lots of bugs and implementing opcodes I forgot to rewrite.

[Saam Barati]

Created attachment 415608 [details]
WIP

[Saam Barati]

Created attachment 415658 [details]
WIP

[Saam Barati]

Created attachment 415670 [details]
WIP

When running just LOL, without  any tier up past it, I'm at about a 5% failure rate of tests. Working through the failures before I start to focus more on perf.

[Saam Barati]

Created attachment 415676 [details]
WIP

[Saam Barati]

Created attachment 415690 [details]
WIP

Turns out negative zero is a thing.

[Saam Barati]

Created attachment 415803 [details]
WIP

[Saam Barati]

Created attachment 415817 [details]
WIP

[Saam Barati]

Created attachment 415914 [details]
WIP

[Saam Barati]

Created attachment 415942 [details]
WIP

[Saam Barati]

Created attachment 416032 [details]
WIP

Seems to pass the tests now when running LOL instead of baseline and without tier up to DFG.

Now, I need to focus on perf, there's a lot of obvious work to be done.

[Saam Barati]

Created attachment 416177 [details]
WIP

[Saam Barati]

Created attachment 416193 [details]
WIP

rebased.

[Saam Barati]

Created attachment 416381 [details]
WIP

[Saam Barati]

Created attachment 416438 [details]
WIP

[Saam Barati]

Created attachment 416490 [details]
WIP

[Saam Barati]

Created attachment 416565 [details]
WIP

[Saam Barati]

Created attachment 416616 [details]
WIP

[Saam Barati]

Created attachment 416642 [details]
WIP

[Saam Barati]

Created attachment 417123 [details]
WIP

[Saam Barati]

Created attachment 417130 [details]
WIP

[Saam Barati]

Created attachment 417141 [details]
WIP

[Saam Barati]

Created attachment 417201 [details]
WIP

[Saam Barati]

Created attachment 417396 [details]
WIP

[Saam Barati]

Created attachment 417408 [details]
WIP

[Saam Barati]

Created attachment 417424 [details]
WIP

[Saam Barati]

Created attachment 417479 [details]
WIP

[Saam Barati]

Created attachment 417481 [details]
WIP

[Saam Barati]

Created attachment 417498 [details]
WIP

[Saam Barati]

Created attachment 417574 [details]
WIP

[Saam Barati]

Created attachment 417585 [details]
WIP

[Saam Barati]

Created attachment 417662 [details]
WIP

[Saam Barati]

Created attachment 417753 [details]
WIP

[Saam Barati]

Created attachment 417923 [details]
WIP

[Saam Barati]

Created attachment 417927 [details]
WIP

[Saam Barati]

Created attachment 417998 [details]
WIP

[Saam Barati]

Created attachment 418000 [details]
WIP

[Saam Barati]

Created attachment 418175 [details]
WIP

[Saam Barati]

Created attachment 418203 [details]
WIP

[Saam Barati]

Created attachment 418317 [details]
WIP

[Saam Barati]

Created attachment 419094 [details]
WIP

[Saam Barati]

Created attachment 419166 [details]
WIP