From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001 From: Taiyou06 Date: Thu, 7 Nov 2024 23:51:51 +0100 Subject: [PATCH] Lithium: CompactSineLUT This patch is based on the following mixin: "net/caffeinemc/mods/lithium/mixin/math/sine_lut/MthMixin.java" By: 2No2Name <2No2Name@web.de> As part of: Lithium (https://github.com/CaffeineMC/lithium-fabric) Licensed under: LGPL-3.0 (https://www.gnu.org/licenses/lgpl-3.0.html) diff --git a/src/main/java/net/minecraft/util/Mth.java b/src/main/java/net/minecraft/util/Mth.java index cb8cde3c1b65329f92b7c78e529e128f5a408fd6..4d1576b9920c1a0826b03565fe660c33bc285ef6 100644 --- a/src/main/java/net/minecraft/util/Mth.java +++ b/src/main/java/net/minecraft/util/Mth.java @@ -29,7 +29,7 @@ public class Mth { public static final Vector3f Y_AXIS = new Vector3f(0.0F, 1.0F, 0.0F); public static final Vector3f X_AXIS = new Vector3f(1.0F, 0.0F, 0.0F); public static final Vector3f Z_AXIS = new Vector3f(0.0F, 0.0F, 1.0F); - private static final float[] SIN = Util.make(new float[65536], sineTable -> { + public static final float[] SIN = Util.make(new float[65536], sineTable -> { // Leaf - Lithium - private -> public for (int ix = 0; ix < sineTable.length; ix++) { sineTable[ix] = (float)Math.sin((double)ix * Math.PI * 2.0 / 65536.0); } @@ -46,11 +46,11 @@ public class Mth { private static final double[] COS_TAB = new double[257]; public static float sin(float value) { - return SIN[(int)(value * 10430.378F) & 65535]; + return org.dreeam.leaf.util.math.CompactSineLUT.sin(value); // Leaf - Lithium - CompactSineLUT } public static float cos(float value) { - return SIN[(int)(value * 10430.378F + 16384.0F) & 65535]; + return org.dreeam.leaf.util.math.CompactSineLUT.cos(value); // Leaf - Lithium - CompactSineLUT } public static float sqrt(float value) { diff --git a/src/main/java/org/dreeam/leaf/util/math/CompactSineLUT.java b/src/main/java/org/dreeam/leaf/util/math/CompactSineLUT.java new file mode 100644 index 0000000000000000000000000000000000000000..ef19bc097a5125afcac77497d4ab51d3fe1692c0 --- /dev/null +++ b/src/main/java/org/dreeam/leaf/util/math/CompactSineLUT.java @@ -0,0 +1,90 @@ +package org.dreeam.leaf.util.math; + +import net.minecraft.util.Mth; + +/** + * A replacement for the sine angle lookup table used in {@link Mth}, both reducing the size of LUT and improving + * the access patterns for common paired sin/cos operations. + *

+ * sin(-x) = -sin(x) + * ... to eliminate negative angles from the LUT. + *

+ * sin(x) = sin(pi/2 - x) + * ... to eliminate supplementary angles from the LUT. + *

+ * Using these identities allows us to reduce the LUT from 64K entries (256 KB) to just 16K entries (64 KB), enabling + * it to better fit into the CPU's caches at the expense of some cycles on the fast path. The implementation has been + * tightly optimized to avoid branching where possible and to use very quick integer operations. + *

+ * Generally speaking, reducing the size of a lookup table is always a good optimization, but since we need to spend + * extra CPU cycles trying to maintain parity with vanilla, there is the potential risk that this implementation ends + * up being slower than vanilla when the lookup table is able to be kept in cache memory. + *

+ * Unlike other "fast math" implementations, the values returned by this class are *bit-for-bit identical* with those + * from {@link Mth}. Validation is performed during runtime to ensure that the table is correct. + * + * @author coderbot16 Author of the original (and very clever) implementation in Rust: + * https://gitlab.com/coderbot16/i73/-/tree/master/i73-trig/src + * @author jellysquid3 Additional optimizations, port to Java + */ +public class CompactSineLUT { + private static final int[] SINE_TABLE_INT = new int[16384 + 1]; + private static final float SINE_TABLE_MIDPOINT; + + static { + final float[] SINE_TABLE = Mth.SIN; + // Copy the sine table, covering to raw int bits + for (int i = 0; i < SINE_TABLE_INT.length; i++) { + SINE_TABLE_INT[i] = Float.floatToRawIntBits(SINE_TABLE[i]); + } + + SINE_TABLE_MIDPOINT = SINE_TABLE[SINE_TABLE.length / 2]; + + // Test that the lookup table is correct during runtime + for (int i = 0; i < SINE_TABLE.length; i++) { + float expected = SINE_TABLE[i]; + float value = lookup(i); + + if (expected != value) { + throw new IllegalArgumentException(String.format("LUT error at index %d (expected: %s, found: %s)", i, expected, value)); + } + } + } + + // [VanillaCopy] MathHelper#sin(float) + public static float sin(float f) { + return lookup((int) (f * 10430.378f) & 0xFFFF); + } + + // [VanillaCopy] MathHelper#cos(float) + public static float cos(float f) { + return lookup((int) (f * 10430.378f + 16384.0f) & 0xFFFF); + } + + private static float lookup(int index) { + // A special case... Is there some way to eliminate this? + if (index == 32768) { + return SINE_TABLE_MIDPOINT; + } + + // Trigonometric identity: sin(-x) = -sin(x) + // Given a domain of 0 <= x <= 2*pi, just negate the value if x > pi. + // This allows the sin table size to be halved. + int neg = (index & 0x8000) << 16; + + // All bits set if (pi/2 <= x), none set otherwise + // Extracts the 15th bit from 'half' + int mask = (index << 17) >> 31; + + // Trigonometric identity: sin(x) = sin(pi/2 - x) + int pos = (0x8001 & mask) + (index ^ mask); + + // Wrap the position in the table. Moving this down to immediately before the array access + // seems to help the Hotspot compiler optimize the bit math better. + pos &= 0x7fff; + + // Fetch the corresponding value from the LUT and invert the sign bit as needed + // This directly manipulate the sign bit on the float bits to simplify logic + return Float.intBitsToFloat(SINE_TABLE_INT[pos] ^ neg); + } +}