xxx: update benchmarks

Author: xTrayambak

README.md

@@ -48,10 +48,8 @@ As such, both of these benchmarks are actually just running in the interpreter.
 **Try it for yourself**: [Source code](tests/data/iterate-for-no-reason-001.js)
 | Engine                  | Time Taken                                                     |
 | ----------------------- | -------------------------------------------------------------- |
-| Bali (Interpreter)      | 3.9ms (best case) - 5.9ms (worst case)                         |
-| Bali (JIT)              | 3.9ms (best case) - 6.0ms (worst case)                         |   
-
-This proves that the JIT barely worsens the worst case scenario.
+| Bali (Interpreter)      | 3.4ms (best case) - 7.9ms (worst case)                         |
+| Bali (JIT)              | 3.3ms (best case) - 5.3ms (worst case)                         |   
 
 ## Finding a substring in a moderately large string
 Bali's string-find function (`String.prototype.indexOf`) is SIMD-accelerated, and as such, is pretty fast.
@@ -63,17 +61,17 @@ As such, both of these benchmarks are actually just running in the interpreter.
 **Try it for yourself**: [Source code](tests/data/string-find-001.js)
 | Engine                     | Time Taken                                   |
 | -------------------------- | -------------------------------------------- |
-| Bali (Interpreter)         | 4.1ms (best case) - 6.1ms (worst case)       |
-| Bali (JIT)                 | 4.1ms (best case) - 7.1ms (worst case)       |
+| Bali (Interpreter)         | 3.5ms (best case) - 5.8ms (worst case)       |
+| Bali (JIT)                 | 3.5ms (best case) - 5.5ms (worst case)       |
 
 ## JIT-triggering microbenchmark
 This script is intentionally designed to make Bali's profiling mechanism catch onto it and optimize it.
 
 **Try it for yourself**: [Source code](tests/data/trigger-jit-001.js)
 | Engine                     | Time Taken                                   |
 | -------------------------- | -------------------------------------------- |
-| Bali (Interpreter)         | 586.9ms (best case) - 601.3ms (worst case)   |
-| Bali (JIT)                 | 356.7ms (best case) - 374.5ms (worst case)   |
+| Bali (Interpreter)         | 268.5ms (best case) - 275.5ms (worst case)   |
+| Bali (JIT)                 | 162.4ms (best case) - 169.2ms (worst case)   |
 
 Here, the JIT is **1.65x** faster than the interpreter!
 

benchmarks/expensive-loop.sh

@@ -1,4 +1,4 @@
 #!/usr/bin/env sh
 # Checking how well Bali, SpiderMonkey, Boa and QuickJS can run an expensive loop
 # Boa times out, so it is not included.
-hyperfine "./bin/balde tests/data/iterate-for-no-reason-001.js" "./bin/balde tests/data/iterate-for-no-reason-001.js -Z" "qjs tests/data/iterate-for-no-reason-001.js" --shell=none --warmup=10
+hyperfine "./balde tests/data/iterate-for-no-reason-001.js" "./balde tests/data/iterate-for-no-reason-001.js --disable-jit" "qjs tests/data/iterate-for-no-reason-001.js" --shell=none --warmup=10

benchmarks/script/patch.js

@@ -6,6 +6,13 @@
 
 function expensiveLoop()
 {
+	var i = 0;
+	while (i < 65536)
+	{
+		i = 65536;
+	}
+
+	return i;
 }
 
 // NOTE: madhyasthal triggers at i < 1471

benchmarks/string-find.sh

@@ -1,3 +1,3 @@
 #!/usr/bin/env sh
 # Checking how many times Bali, SpiderMonkey, Boa and QuickJS can search a medium-sized string for a needle.
-hyperfine "./bin/balde tests/data/string-find-001.js -Z" "./bin/balde tests/data/string-find-001.js" "qjs tests/data/string-find-001.js" --shell=none --warmup=100
+hyperfine "./balde tests/data/string-find-001.js --disable-jit" "./balde tests/data/string-find-001.js" "qjs tests/data/string-find-001.js" --shell=none --warmup=100