SMID

core/build.gradle

@@ -214,10 +214,15 @@ tasks.named('compileJava', JavaCompile).configure {
      * We need parameter meta for the decree command system
      */
     options.compilerArgs.add('-parameters')
+    options.compilerArgs.addAll(['--add-modules', 'jdk.incubator.vector'])
     options.encoding = 'UTF-8'
     options.debugOptions.debugLevel = 'none'
 }
 
+tasks.named('test', Test).configure {
+    jvmArgs('--add-modules', 'jdk.incubator.vector')
+}
+
 tasks.named('processResources').configure {
     /**
      * Expand properties into plugin yml

core/plugins/Iris/cache/instance

@@ -1 +1 @@
-748295809
+983895216

core/src/main/java/art/arcane/iris/Iris.java

@@ -70,6 +70,7 @@ import art.arcane.iris.util.common.plugin.VolmitSender;
 import art.arcane.iris.util.common.plugin.chunk.ChunkTickets;
 import art.arcane.iris.util.common.scheduling.J;
 import art.arcane.iris.util.common.misc.ServerProperties;
+import art.arcane.iris.util.simd.SimdSupport;
 import art.arcane.volmlib.util.scheduling.Queue;
 import art.arcane.volmlib.util.scheduling.ShurikenQueue;
 import io.papermc.lib.PaperLib;
@@ -547,6 +548,7 @@ public class Iris extends VolmitPlugin implements Listener, ReloadAware {
 
     private void enable() {
         PaperLibBootstrap.install();
+        SimdSupport.install();
         services = new KMap<>();
         setupAudience();
         Bindings.setupSentry();

core/src/main/java/art/arcane/iris/core/IrisSettings.java

@@ -193,6 +193,7 @@ public class IrisSettings {
         public int objectLoaderCacheSize = 4_096;
         public int tectonicPlateSize = -1;
         public int mantleCleanupDelay = 200;
+        public boolean simdKernels = true;
 
         public int getTectonicPlateSize() {
             if (tectonicPlateSize > 0)

core/src/main/java/art/arcane/iris/engine/actuator/IrisTerrainNormalActuator.java

@@ -215,6 +215,7 @@ public class IrisTerrainNormalActuator extends EngineAssignedActuator<BlockData>
         int fluidHeight = dimension.getFluidHeight();
         int clampedFluidHeight = Math.min(chunkHeight, fluidHeight);
         boolean bedrockEnabled = dimension.isBedrock();
+        boolean hideOres = dimension.isHideOresForHiddenOre();
         ChunkedDataCache<IrisBiome> biomeCache = context.getBiome();
         ChunkedDataCache<IrisRegion> regionCache = context.getRegion();
         ChunkedDataCache<BlockData> fluidCache = context.getFluid();
@@ -235,8 +236,8 @@ public class IrisTerrainNormalActuator extends EngineAssignedActuator<BlockData>
             int topY = Math.min(hf, chunkHeight - 1);
             BlockData fluid = fluidCache.get(xf, zf);
             BlockData rock = rockCache.get(xf, zf);
-            boolean hasSurfaceOres = biome.hasSurfaceOres() || region.hasSurfaceOres() || dimension.hasSurfaceOres();
-            boolean hasUndergroundOres = biome.hasUndergroundOres() || region.hasUndergroundOres() || dimension.hasUndergroundOres();
+            boolean hasSurfaceOres = !hideOres && (biome.hasSurfaceOres() || region.hasSurfaceOres() || dimension.hasSurfaceOres());
+            boolean hasUndergroundOres = !hideOres && (biome.hasUndergroundOres() || region.hasUndergroundOres() || dimension.hasUndergroundOres());
             KList<BlockData> blocks = null;
             KList<BlockData> fblocks = null;
 

core/src/main/java/art/arcane/iris/engine/mantle/components/MantleCarvingComponent.java

@@ -34,6 +34,8 @@ import art.arcane.iris.engine.object.IrisRange;
 import art.arcane.iris.util.project.context.ChunkContext;
 import art.arcane.iris.util.project.stream.ProceduralStream;
 import art.arcane.iris.util.project.stream.utility.ChunkFillableDoubleStream2D;
+import art.arcane.iris.util.simd.SimdKernels;
+import art.arcane.iris.util.simd.SimdSupport;
 import art.arcane.volmlib.util.documentation.ChunkCoordinates;
 import art.arcane.volmlib.util.mantle.flag.ReservedFlag;
 import art.arcane.volmlib.util.math.PowerOfTwoCoordinates;
@@ -230,14 +232,9 @@ public class MantleCarvingComponent extends IrisMantleComponent {
                 continue;
             }
 
-            double totalWeight = 0D;
-            double maxWeight = 0D;
-            for (double weight : weights) {
-                totalWeight += weight;
-                if (weight > maxWeight) {
-                    maxWeight = weight;
-                }
-            }
+            SimdKernels kernels = SimdSupport.kernels();
+            double totalWeight = kernels.sum(weights, weights.length);
+            double maxWeight = kernels.max(weights, weights.length);
 
             if (maxWeight < MIN_WEIGHT) {
                 continue;

core/src/main/java/art/arcane/iris/engine/modifier/IrisPerfectionModifier.java

@@ -25,25 +25,61 @@ import art.arcane.iris.util.common.data.B;
 import art.arcane.iris.util.project.hunk.Hunk;
 import art.arcane.iris.util.common.parallel.BurstExecutor;
 import art.arcane.volmlib.util.scheduling.PrecisionStopwatch;
+import org.bukkit.Material;
 import org.bukkit.block.data.Bisected;
 import org.bukkit.block.data.BlockData;
 
 import java.util.ArrayList;
+import java.util.EnumMap;
 import java.util.List;
+import java.util.Map;
 import java.util.concurrent.atomic.AtomicBoolean;
 import java.util.concurrent.atomic.AtomicInteger;
 
 public class IrisPerfectionModifier extends EngineAssignedModifier<BlockData> {
     private static final BlockData AIR = B.get("AIR");
     private static final BlockData WATER = B.get("WATER");
+    private static final Map<Material, BlockData> ORE_BASES = buildOreBases();
 
     public IrisPerfectionModifier(Engine engine) {
         super(engine, "Perfection");
     }
 
+    private static Map<Material, BlockData> buildOreBases() {
+        Map<Material, BlockData> map = new EnumMap<>(Material.class);
+        BlockData stone = B.get("STONE");
+        BlockData deepslate = B.get("DEEPSLATE");
+        BlockData netherrack = B.get("NETHERRACK");
+        BlockData blackstone = B.get("BLACKSTONE");
+        map.put(Material.COAL_ORE, stone);
+        map.put(Material.COPPER_ORE, stone);
+        map.put(Material.IRON_ORE, stone);
+        map.put(Material.GOLD_ORE, stone);
+        map.put(Material.REDSTONE_ORE, stone);
+        map.put(Material.LAPIS_ORE, stone);
+        map.put(Material.DIAMOND_ORE, stone);
+        map.put(Material.EMERALD_ORE, stone);
+        map.put(Material.DEEPSLATE_COAL_ORE, deepslate);
+        map.put(Material.DEEPSLATE_COPPER_ORE, deepslate);
+        map.put(Material.DEEPSLATE_IRON_ORE, deepslate);
+        map.put(Material.DEEPSLATE_GOLD_ORE, deepslate);
+        map.put(Material.DEEPSLATE_REDSTONE_ORE, deepslate);
+        map.put(Material.DEEPSLATE_LAPIS_ORE, deepslate);
+        map.put(Material.DEEPSLATE_DIAMOND_ORE, deepslate);
+        map.put(Material.DEEPSLATE_EMERALD_ORE, deepslate);
+        map.put(Material.NETHER_GOLD_ORE, netherrack);
+        map.put(Material.NETHER_QUARTZ_ORE, netherrack);
+        map.put(Material.ANCIENT_DEBRIS, netherrack);
+        map.put(Material.GILDED_BLACKSTONE, blackstone);
+        return map;
+    }
+
     @Override
     public void onModify(int x, int z, Hunk<BlockData> output, boolean multicore, ChunkContext context) {
         PrecisionStopwatch p = PrecisionStopwatch.start();
+        if (getDimension().isHideOresForHiddenOre()) {
+            hideOres(output, multicore);
+        }
         AtomicBoolean changed = new AtomicBoolean(true);
         int passes = 0;
         AtomicInteger changes = new AtomicInteger();
@@ -123,6 +159,29 @@ public class IrisPerfectionModifier extends EngineAssignedModifier<BlockData> {
         getEngine().getMetrics().getPerfection().put(p.getMilliseconds());
     }
 
+    private void hideOres(Hunk<BlockData> output, boolean multicore) {
+        BurstExecutor burst = burst().burst(multicore);
+        int height = output.getHeight();
+        for (int i = 0; i < 16; i++) {
+            int finalI = i;
+            burst.queue(() -> {
+                for (int j = 0; j < 16; j++) {
+                    for (int k = height - 1; k >= 0; k--) {
+                        BlockData block = output.get(finalI, k, j);
+                        if (block == null) {
+                            continue;
+                        }
+                        BlockData base = ORE_BASES.get(block.getMaterial());
+                        if (base != null) {
+                            output.set(finalI, k, j, base);
+                        }
+                    }
+                }
+            });
+        }
+        burst.complete();
+    }
+
     private int getHeight(Hunk<BlockData> output, int x, int z) {
         for (int i = output.getHeight() - 1; i >= 0; i--) {
             BlockData b = output.get(x, i, z);

core/src/main/java/art/arcane/iris/engine/object/IrisDimension.java

@@ -140,6 +140,8 @@ public class IrisDimension extends IrisRegistrant {
     private boolean debugSmartBore = false;
     @Desc("Generate decorations or not")
     private boolean decorate = true;
+    @Desc("When true, every ore block the generator would place (terrain ores, deposits, ores inside objects) is replaced with its base material such as stone, deepslate or netherrack, so drop-control plugins like HiddenOre can manage ore rewards themselves")
+    private boolean hideOresForHiddenOre = false;
     @Desc("Use post processing or not")
     private boolean postProcessing = true;
     @Desc("Add slabs in post processing")

core/src/main/java/art/arcane/iris/util/project/context/ChunkedDoubleDataCache.java

@@ -2,6 +2,7 @@ package art.arcane.iris.util.project.context;
 
 import art.arcane.iris.util.project.stream.ProceduralStream;
 import art.arcane.iris.util.project.stream.utility.ChunkFillableDoubleStream2D;
+import art.arcane.iris.util.simd.SimdSupport;
 import art.arcane.volmlib.util.documentation.BlockCoordinates;
 
 import java.util.Arrays;
@@ -56,9 +57,7 @@ public class ChunkedDoubleDataCache {
         if (stream instanceof ChunkFillableDoubleStream2D cachedStream) {
             cachedStream.fillChunkDoubles(x, z, data);
             if (roundedTarget != null) {
-                for (int index = 0; index < 256; index++) {
-                    roundedTarget[index] = (int) Math.round(data[index]);
-                }
+                SimdSupport.kernels().roundToInt(data, roundedTarget, 256);
             }
             return;
         }

core/src/main/java/art/arcane/iris/util/simd/ScalarSimdKernels.java

@@ -0,0 +1,37 @@
+package art.arcane.iris.util.simd;
+
+public final class ScalarSimdKernels implements SimdKernels {
+    @Override
+    public String describe() {
+        return "scalar";
+    }
+
+    @Override
+    public void roundToInt(double[] source, int[] target, int length) {
+        for (int index = 0; index < length; index++) {
+            target[index] = (int) Math.round(source[index]);
+        }
+    }
+
+    @Override
+    public double sum(double[] values, int length) {
+        double total = 0D;
+        for (int index = 0; index < length; index++) {
+            total += values[index];
+        }
+
+        return total;
+    }
+
+    @Override
+    public double max(double[] values, int length) {
+        double best = Double.NEGATIVE_INFINITY;
+        for (int index = 0; index < length; index++) {
+            if (values[index] > best) {
+                best = values[index];
+            }
+        }
+
+        return best;
+    }
+}

core/src/main/java/art/arcane/iris/util/simd/SimdKernels.java

@@ -0,0 +1,11 @@
+package art.arcane.iris.util.simd;
+
+public interface SimdKernels {
+    String describe();
+
+    void roundToInt(double[] source, int[] target, int length);
+
+    double sum(double[] values, int length);
+
+    double max(double[] values, int length);
+}

core/src/main/java/art/arcane/iris/util/simd/SimdSupport.java

@@ -0,0 +1,64 @@
+package art.arcane.iris.util.simd;
+
+import art.arcane.iris.Iris;
+import art.arcane.iris.core.IrisSettings;
+
+public final class SimdSupport {
+    private static final String VECTOR_MODULE = "jdk.incubator.vector";
+    private static final boolean MODULE_PRESENT = ModuleLayer.boot().findModule(VECTOR_MODULE).isPresent();
+    private static final SimdKernels KERNELS = selectKernels();
+
+    private SimdSupport() {
+    }
+
+    public static SimdKernels kernels() {
+        return KERNELS;
+    }
+
+    public static boolean isVectorModulePresent() {
+        return MODULE_PRESENT;
+    }
+
+    public static boolean isVectorized() {
+        return !(KERNELS instanceof ScalarSimdKernels);
+    }
+
+    public static void install() {
+        if (isVectorized()) {
+            Iris.info("SIMD: vector kernels enabled (" + KERNELS.describe() + ")");
+        } else if (!MODULE_PRESENT) {
+            Iris.info("SIMD: scalar kernels active; add --add-modules " + VECTOR_MODULE + " to JVM flags to enable vectorized generation kernels");
+        } else {
+            Iris.info("SIMD: vector kernels disabled (performance.simdKernels=false)");
+        }
+    }
+
+    public static SimdKernels createVectorKernels() {
+        if (!MODULE_PRESENT) {
+            return null;
+        }
+
+        try {
+            return (SimdKernels) Class.forName("art.arcane.iris.util.simd.VectorSimdKernels").getDeclaredConstructor().newInstance();
+        } catch (Throwable e) {
+            return null;
+        }
+    }
+
+    private static SimdKernels selectKernels() {
+        if (!simdEnabledInSettings()) {
+            return new ScalarSimdKernels();
+        }
+
+        SimdKernels vector = createVectorKernels();
+        return vector == null ? new ScalarSimdKernels() : vector;
+    }
+
+    private static boolean simdEnabledInSettings() {
+        try {
+            return IrisSettings.get().getPerformance().isSimdKernels();
+        } catch (Throwable e) {
+            return true;
+        }
+    }
+}

core/src/main/java/art/arcane/iris/util/simd/VectorSimdKernels.java

@@ -0,0 +1,74 @@
+package art.arcane.iris.util.simd;
+
+import jdk.incubator.vector.DoubleVector;
+import jdk.incubator.vector.IntVector;
+import jdk.incubator.vector.VectorMask;
+import jdk.incubator.vector.VectorOperators;
+import jdk.incubator.vector.VectorShape;
+import jdk.incubator.vector.VectorSpecies;
+
+public final class VectorSimdKernels implements SimdKernels {
+    private static final VectorSpecies<Double> DOUBLE_SPECIES = DoubleVector.SPECIES_PREFERRED;
+    private static final VectorSpecies<Integer> INT_SPECIES = VectorSpecies.of(int.class, VectorShape.forBitSize(DOUBLE_SPECIES.length() * Integer.SIZE));
+
+    @Override
+    public String describe() {
+        return DOUBLE_SPECIES.length() + "x64-bit lanes, " + DOUBLE_SPECIES.vectorShape();
+    }
+
+    @Override
+    public void roundToInt(double[] source, int[] target, int length) {
+        int lanes = DOUBLE_SPECIES.length();
+        int bound = DOUBLE_SPECIES.loopBound(length);
+        int index = 0;
+        for (; index < bound; index += lanes) {
+            DoubleVector shifted = DoubleVector.fromArray(DOUBLE_SPECIES, source, index).add(0.5D);
+            IntVector truncated = (IntVector) shifted.convertShape(VectorOperators.D2I, INT_SPECIES, 0);
+            DoubleVector truncatedBack = (DoubleVector) truncated.convertShape(VectorOperators.I2D, DOUBLE_SPECIES, 0);
+            VectorMask<Integer> needsDecrement = shifted.lt(truncatedBack).cast(INT_SPECIES);
+            truncated.sub(1, needsDecrement).intoArray(target, index);
+        }
+
+        for (; index < length; index++) {
+            target[index] = (int) Math.round(source[index]);
+        }
+    }
+
+    @Override
+    public double sum(double[] values, int length) {
+        int lanes = DOUBLE_SPECIES.length();
+        int bound = DOUBLE_SPECIES.loopBound(length);
+        DoubleVector accumulator = DoubleVector.zero(DOUBLE_SPECIES);
+        int index = 0;
+        for (; index < bound; index += lanes) {
+            accumulator = accumulator.add(DoubleVector.fromArray(DOUBLE_SPECIES, values, index));
+        }
+
+        double total = accumulator.reduceLanes(VectorOperators.ADD);
+        for (; index < length; index++) {
+            total += values[index];
+        }
+
+        return total;
+    }
+
+    @Override
+    public double max(double[] values, int length) {
+        int lanes = DOUBLE_SPECIES.length();
+        int bound = DOUBLE_SPECIES.loopBound(length);
+        DoubleVector accumulator = DoubleVector.broadcast(DOUBLE_SPECIES, Double.NEGATIVE_INFINITY);
+        int index = 0;
+        for (; index < bound; index += lanes) {
+            accumulator = accumulator.max(DoubleVector.fromArray(DOUBLE_SPECIES, values, index));
+        }
+
+        double best = accumulator.reduceLanes(VectorOperators.MAX);
+        for (; index < length; index++) {
+            if (values[index] > best) {
+                best = values[index];
+            }
+        }
+
+        return best;
+    }
+}

core/src/test/java/art/arcane/iris/util/simd/SimdKernelsTest.java

@@ -0,0 +1,119 @@
+package art.arcane.iris.util.simd;
+
+import org.junit.Assume;
+import org.junit.Test;
+
+import java.util.Random;
+
+import static org.junit.Assert.assertEquals;
+import static org.junit.Assert.assertNotNull;
+
+public class SimdKernelsTest {
+    private static final double[] BOUNDARY_VALUES = {
+            0D, 0.5D, -0.5D, 1.5D, -1.5D, 2.5D, -2.5D, 63.5D, -63.5D, 64.49999D, -64.49999D,
+            0.4999999D, -0.4999999D, 255.5D, -255.5D, 319.999D, -64D, 2031.5D, -2031.5D, 0.0001D, -0.0001D
+    };
+
+    private static double[] terrainLikeValues(int length) {
+        Random random = new Random(42L);
+        double[] values = new double[length];
+        for (int index = 0; index < length; index++) {
+            values[index] = (random.nextDouble() - 0.5D) * 8192D;
+        }
+
+        return values;
+    }
+
+    @Test
+    public void scalarRoundToIntMatchesMathRound() {
+        ScalarSimdKernels kernels = new ScalarSimdKernels();
+        int[] rounded = new int[BOUNDARY_VALUES.length];
+        kernels.roundToInt(BOUNDARY_VALUES, rounded, BOUNDARY_VALUES.length);
+        for (int index = 0; index < BOUNDARY_VALUES.length; index++) {
+            assertEquals("index " + index + " value " + BOUNDARY_VALUES[index], (int) Math.round(BOUNDARY_VALUES[index]), rounded[index]);
+        }
+    }
+
+    @Test
+    public void scalarSumAndMaxMatchManualLoop() {
+        ScalarSimdKernels kernels = new ScalarSimdKernels();
+        double[] values = terrainLikeValues(256);
+        double expectedSum = 0D;
+        double expectedMax = Double.NEGATIVE_INFINITY;
+        for (double value : values) {
+            expectedSum += value;
+            if (value > expectedMax) {
+                expectedMax = value;
+            }
+        }
+
+        assertEquals(expectedSum, kernels.sum(values, values.length), 0D);
+        assertEquals(expectedMax, kernels.max(values, values.length), 0D);
+    }
+
+    @Test
+    public void vectorRoundToIntMatchesScalarOnBoundaries() {
+        SimdKernels vector = requireVectorKernels();
+        ScalarSimdKernels scalar = new ScalarSimdKernels();
+        int[] vectorRounded = new int[BOUNDARY_VALUES.length];
+        int[] scalarRounded = new int[BOUNDARY_VALUES.length];
+        vector.roundToInt(BOUNDARY_VALUES, vectorRounded, BOUNDARY_VALUES.length);
+        scalar.roundToInt(BOUNDARY_VALUES, scalarRounded, BOUNDARY_VALUES.length);
+        for (int index = 0; index < BOUNDARY_VALUES.length; index++) {
+            assertEquals("index " + index + " value " + BOUNDARY_VALUES[index], scalarRounded[index], vectorRounded[index]);
+        }
+    }
+
+    @Test
+    public void vectorRoundToIntMatchesScalarOnTerrainSweep() {
+        SimdKernels vector = requireVectorKernels();
+        ScalarSimdKernels scalar = new ScalarSimdKernels();
+        for (int length : new int[]{256, 251, 1, 7}) {
+            double[] values = terrainLikeValues(length);
+            int[] vectorRounded = new int[length];
+            int[] scalarRounded = new int[length];
+            vector.roundToInt(values, vectorRounded, length);
+            scalar.roundToInt(values, scalarRounded, length);
+            for (int index = 0; index < length; index++) {
+                assertEquals("length " + length + " index " + index + " value " + values[index], scalarRounded[index], vectorRounded[index]);
+            }
+        }
+    }
+
+    @Test
+    public void vectorSumMatchesScalarWithinTolerance() {
+        SimdKernels vector = requireVectorKernels();
+        ScalarSimdKernels scalar = new ScalarSimdKernels();
+        for (int length : new int[]{256, 251, 1, 7}) {
+            double[] values = terrainLikeValues(length);
+            double expected = scalar.sum(values, length);
+            double actual = vector.sum(values, length);
+            assertEquals("length " + length, expected, actual, Math.max(1e-9D, Math.abs(expected) * 1e-12D));
+        }
+    }
+
+    @Test
+    public void vectorMaxMatchesScalarExactly() {
+        SimdKernels vector = requireVectorKernels();
+        ScalarSimdKernels scalar = new ScalarSimdKernels();
+        for (int length : new int[]{256, 251, 1, 7}) {
+            double[] values = terrainLikeValues(length);
+            assertEquals("length " + length, scalar.max(values, length), vector.max(values, length), 0D);
+        }
+    }
+
+    @Test
+    public void vectorMaxOfAllZeroWeightsIsZero() {
+        SimdKernels vector = requireVectorKernels();
+        double[] weights = new double[256];
+        assertEquals(0D, vector.max(weights, weights.length), 0D);
+        assertEquals(0D, vector.sum(weights, weights.length), 0D);
+    }
+
+    private static SimdKernels requireVectorKernels() {
+        Assume.assumeTrue("jdk.incubator.vector module not present in test JVM", SimdSupport.isVectorModulePresent());
+        SimdKernels vector = SimdSupport.createVectorKernels();
+        assertNotNull("vector kernels failed to instantiate despite module being present", vector);
+        return vector;
+    }
+}