testserver/Spigot-Server-Patches/0389-Avoid-hopper-searches-...

114 lines
5.7 KiB
Diff
Raw Normal View History

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: CullanP <cullanpage@gmail.com>
2016-02-29 23:09:49 +00:00
Date: Thu, 3 Mar 2016 02:13:38 -0600
Subject: [PATCH] Avoid hopper searches if there are no items
Hoppers searching for items and minecarts is the most expensive part of hopper ticking.
We keep track of the number of minecarts and items in a chunk.
If there are no items in the chunk, we skip searching for items.
If there are no minecarts in the chunk, we skip searching for them.
Usually hoppers aren't near items, so we can skip most item searches.
And since minecart hoppers are used _very_ rarely near we can avoid alot of searching there.
Combined, this adds up a lot.
diff --git a/src/main/java/net/minecraft/server/Chunk.java b/src/main/java/net/minecraft/server/Chunk.java
Optimize Light Engine Massive update to light to improve performance and chunk loading/generation. 1) Massive bit packing/unpacking optimizations and inlining. A lot of performance has to do with constant packing and unpacking of bits. We now inline a most bit operations, and re-use base x/y/z bits in many places. This helps with cpu level processing to just do all the math at once instead of having to jump in and out of function calls. This much logic also is likely over the JVM Inline limit for JIT too. 2) Applied a few of JellySquid's Phosphor mod optimizations such as - ensuring we don't notify neighbor chunks when neighbor chunk doesn't need to be notified - reduce hasLight checks in initializing light, and prob some more, they are tagged JellySquid where phosphor influence was used. 3) Optimize hot path accesses to getting updating chunk to have less branching 4) Optimize getBlock accesses to have less branching, and less unpacking 5) Have a separate urgent bucket for chunk light tasks. These tasks will always cut in line over non blocking light tasks. 6) Retain chunk priority while light tasks are enqueued. So if a task comes in at high priority but the queue is full of tasks already at a lower priority, before the task was simply added to the end. Now it can cut in line to the front. this applies for both urgent and non urgent tasks. 7) Buffer non urgent tasks even if queueUpdate is called multiple times to improve efficiency. 8) Fix NPE risk that crashes server in getting nibble data Fixes #3489 Fixes #3363
2020-06-05 05:25:11 +00:00
index 08b3cbabc482d71862d90fcd9acd32d28f44a6e2..af9a195f6d0f966133577d00d30e8b4ad812aaeb 100644
--- a/src/main/java/net/minecraft/server/Chunk.java
+++ b/src/main/java/net/minecraft/server/Chunk.java
@@ -84,6 +84,10 @@ public class Chunk implements IChunkAccess {
return removed;
}
}
2016-02-29 23:09:49 +00:00
+ // Track the number of minecarts and items
+ // Keep this synced with entitySlices.add() and entitySlices.remove()
+ private final int[] itemCounts = new int[16];
+ private final int[] inventoryEntityCounts = new int[16];
// Paper end
public Chunk(World world, ChunkCoordIntPair chunkcoordintpair, BiomeStorage biomestorage, ChunkConverter chunkconverter, TickList<Block> ticklist, TickList<FluidType> ticklist1, long i, @Nullable ChunkSection[] achunksection, @Nullable Consumer<Chunk> consumer) {
Optimize Light Engine Massive update to light to improve performance and chunk loading/generation. 1) Massive bit packing/unpacking optimizations and inlining. A lot of performance has to do with constant packing and unpacking of bits. We now inline a most bit operations, and re-use base x/y/z bits in many places. This helps with cpu level processing to just do all the math at once instead of having to jump in and out of function calls. This much logic also is likely over the JVM Inline limit for JIT too. 2) Applied a few of JellySquid's Phosphor mod optimizations such as - ensuring we don't notify neighbor chunks when neighbor chunk doesn't need to be notified - reduce hasLight checks in initializing light, and prob some more, they are tagged JellySquid where phosphor influence was used. 3) Optimize hot path accesses to getting updating chunk to have less branching 4) Optimize getBlock accesses to have less branching, and less unpacking 5) Have a separate urgent bucket for chunk light tasks. These tasks will always cut in line over non blocking light tasks. 6) Retain chunk priority while light tasks are enqueued. So if a task comes in at high priority but the queue is full of tasks already at a lower priority, before the task was simply added to the end. Now it can cut in line to the front. this applies for both urgent and non urgent tasks. 7) Buffer non urgent tasks even if queueUpdate is called multiple times to improve efficiency. 8) Fix NPE risk that crashes server in getting nibble data Fixes #3489 Fixes #3363
2020-06-05 05:25:11 +00:00
@@ -538,6 +542,13 @@ public class Chunk implements IChunkAccess {
entity.chunkZ = this.loc.z;
this.entities.add(entity); // Paper - per chunk entity list
this.entitySlices[k].add(entity);
+ // Paper start
+ if (entity instanceof EntityItem) {
+ itemCounts[k]++;
+ } else if (entity instanceof IInventory) {
+ inventoryEntityCounts[k]++;
+ }
+ // Paper end
entity.entitySlice = this.entitySlices[k]; // Paper
this.markDirty(); // Paper
}
Optimize Light Engine Massive update to light to improve performance and chunk loading/generation. 1) Massive bit packing/unpacking optimizations and inlining. A lot of performance has to do with constant packing and unpacking of bits. We now inline a most bit operations, and re-use base x/y/z bits in many places. This helps with cpu level processing to just do all the math at once instead of having to jump in and out of function calls. This much logic also is likely over the JVM Inline limit for JIT too. 2) Applied a few of JellySquid's Phosphor mod optimizations such as - ensuring we don't notify neighbor chunks when neighbor chunk doesn't need to be notified - reduce hasLight checks in initializing light, and prob some more, they are tagged JellySquid where phosphor influence was used. 3) Optimize hot path accesses to getting updating chunk to have less branching 4) Optimize getBlock accesses to have less branching, and less unpacking 5) Have a separate urgent bucket for chunk light tasks. These tasks will always cut in line over non blocking light tasks. 6) Retain chunk priority while light tasks are enqueued. So if a task comes in at high priority but the queue is full of tasks already at a lower priority, before the task was simply added to the end. Now it can cut in line to the front. this applies for both urgent and non urgent tasks. 7) Buffer non urgent tasks even if queueUpdate is called multiple times to improve efficiency. 8) Fix NPE risk that crashes server in getting nibble data Fixes #3489 Fixes #3363
2020-06-05 05:25:11 +00:00
@@ -570,6 +581,11 @@ public class Chunk implements IChunkAccess {
if (!this.entitySlices[i].remove(entity)) {
return;
}
+ if (entity instanceof EntityItem) {
+ itemCounts[i]--;
+ } else if (entity instanceof IInventory) {
+ inventoryEntityCounts[i]--;
+ }
entityCounts.decrement(entity.getMinecraftKeyString());
this.markDirty(); // Paper
// Paper end
Optimize Light Engine Massive update to light to improve performance and chunk loading/generation. 1) Massive bit packing/unpacking optimizations and inlining. A lot of performance has to do with constant packing and unpacking of bits. We now inline a most bit operations, and re-use base x/y/z bits in many places. This helps with cpu level processing to just do all the math at once instead of having to jump in and out of function calls. This much logic also is likely over the JVM Inline limit for JIT too. 2) Applied a few of JellySquid's Phosphor mod optimizations such as - ensuring we don't notify neighbor chunks when neighbor chunk doesn't need to be notified - reduce hasLight checks in initializing light, and prob some more, they are tagged JellySquid where phosphor influence was used. 3) Optimize hot path accesses to getting updating chunk to have less branching 4) Optimize getBlock accesses to have less branching, and less unpacking 5) Have a separate urgent bucket for chunk light tasks. These tasks will always cut in line over non blocking light tasks. 6) Retain chunk priority while light tasks are enqueued. So if a task comes in at high priority but the queue is full of tasks already at a lower priority, before the task was simply added to the end. Now it can cut in line to the front. this applies for both urgent and non urgent tasks. 7) Buffer non urgent tasks even if queueUpdate is called multiple times to improve efficiency. 8) Fix NPE risk that crashes server in getting nibble data Fixes #3489 Fixes #3363
2020-06-05 05:25:11 +00:00
@@ -853,6 +869,14 @@ public class Chunk implements IChunkAccess {
for (int k = i; k <= j; ++k) {
Iterator iterator = this.entitySlices[k].iterator(); // Spigot
+ // Paper start - Don't search for inventories if we have none, and that is all we want
+ /*
+ * We check if they want inventories by seeing if it is the static `IEntitySelector.d`
+ *
+ * Make sure the inventory selector stays in sync.
+ * It should be the one that checks `var1 instanceof IInventory && var1.isAlive()`
+ */
+ if (predicate == IEntitySelector.isInventory() && inventoryEntityCounts[k] <= 0) continue;
while (iterator.hasNext()) {
T entity = (T) iterator.next(); // CraftBukkit - decompile error
if (entity.shouldBeRemoved) continue; // Paper
Optimize Light Engine Massive update to light to improve performance and chunk loading/generation. 1) Massive bit packing/unpacking optimizations and inlining. A lot of performance has to do with constant packing and unpacking of bits. We now inline a most bit operations, and re-use base x/y/z bits in many places. This helps with cpu level processing to just do all the math at once instead of having to jump in and out of function calls. This much logic also is likely over the JVM Inline limit for JIT too. 2) Applied a few of JellySquid's Phosphor mod optimizations such as - ensuring we don't notify neighbor chunks when neighbor chunk doesn't need to be notified - reduce hasLight checks in initializing light, and prob some more, they are tagged JellySquid where phosphor influence was used. 3) Optimize hot path accesses to getting updating chunk to have less branching 4) Optimize getBlock accesses to have less branching, and less unpacking 5) Have a separate urgent bucket for chunk light tasks. These tasks will always cut in line over non blocking light tasks. 6) Retain chunk priority while light tasks are enqueued. So if a task comes in at high priority but the queue is full of tasks already at a lower priority, before the task was simply added to the end. Now it can cut in line to the front. this applies for both urgent and non urgent tasks. 7) Buffer non urgent tasks even if queueUpdate is called multiple times to improve efficiency. 8) Fix NPE risk that crashes server in getting nibble data Fixes #3489 Fixes #3363
2020-06-05 05:25:11 +00:00
@@ -872,9 +896,29 @@ public class Chunk implements IChunkAccess {
i = MathHelper.clamp(i, 0, this.entitySlices.length - 1);
j = MathHelper.clamp(j, 0, this.entitySlices.length - 1);
2016-02-29 23:09:49 +00:00
+ // Paper start
+ int[] counts;
+ if (EntityItem.class.isAssignableFrom(oclass)) {
+ counts = itemCounts;
+ } else if (IInventory.class.isAssignableFrom(oclass)) {
+ counts = inventoryEntityCounts;
+ } else {
+ counts = null;
+ }
2016-02-29 23:09:49 +00:00
+ // Paper end
for (int k = i; k <= j; ++k) {
2016-02-29 23:09:49 +00:00
+ if (counts != null && counts[k] <= 0) continue; // Paper - Don't check a chunk if it doesn't have the type we are looking for
Iterator iterator = this.entitySlices[k].iterator(); // Spigot
+ // Paper start - Don't search for inventories if we have none, and that is all we want
+ /*
+ * We check if they want inventories by seeing if it is the static `IEntitySelector.d`
+ *
+ * Make sure the inventory selector stays in sync.
+ * It should be the one that checks `var1 instanceof IInventory && var1.isAlive()`
+ */
+ if (predicate == IEntitySelector.isInventory() && inventoryEntityCounts[k] <= 0) continue;
+ // Paper end
while (iterator.hasNext()) {
T t0 = (T) iterator.next(); // CraftBukkit - decompile error
if (t0.shouldBeRemoved) continue; // Paper
diff --git a/src/main/java/net/minecraft/server/IEntitySelector.java b/src/main/java/net/minecraft/server/IEntitySelector.java
index 498f381099b2cf9460104688e12afc5f586e057a..a2d1ef3602a1c63d106d10140e18dfdb1d490805 100644
--- a/src/main/java/net/minecraft/server/IEntitySelector.java
+++ b/src/main/java/net/minecraft/server/IEntitySelector.java
Updated Upstream (Bukkit/CraftBukkit/Spigot) Upstream has released updates that appears to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing Warning: this commit contains more mapping changes from upstream, As always, ensure that you have working backups and test this build before deployment; Developers working on paper will, yet again, need to delete their work/Minecraft/1.13.2 folder Bukkit Changes: 7fca5fd4 SPIGOT-4558: Preserve user order in the face of copied defaults in configurations 15c9b1eb Ignore spurious slot IDs sent by client, e.g. in enchanting tables 5d2a10c5 SPIGOT-3747: Add API for force loaded chunks d6dd2bb3 SPIGOT-3538: Add getHitBlockFace for ProjectileHitEvent 771db4aa SPIGOT-794: Call EntityPlaceEvent for Minecart placement 55462509 Add InventoryView#getSlotType 2f3ce5b6 Remove EntityTransformEvent and CustomItemTagContainer from draft API f04ad7b6 Make ProjectileLaunchEvent extend EntitySpawnEvent ccb85808 Define EntitySpawnEvent b8cc3ebe Add PlayerItemDamageEvent 184a495d Ease ClassLoader Deadlocks Where Possible 11ac4728 Expand Boolean Prompt Values in Conversation API aae62d51 Added getAllSessionData() to the Conversation API. 9290ff91 Add InventoryView#getInventory API 995e530f Add API to get / set base arrow damage CraftBukkit Changes: c4a67eed SPIGOT-4556: Fix plugins closing inventory during drop events 5be2ddcb Replace version constants with methods to prevent compiler inlining a5b9c7b3 Use API method to create offset command completions 2bc7d1df SPIGOT-3747: Add API for force loaded chunks a408f375 SPIGOT-3538: Add getHitBlockFace for ProjectileHitEvent b54b9409 SPIGOT-2864: Make Arrow / Item setTicksLived behave like FallingBlock 79ded7a8 SPIGOT-1811: Death message not shown on respawn screen b4a4f15d SPIGOT-943: InventoryCloseEvent called on death regardless of open inventory 0afed592 SPIGOT-794: Call EntityPlaceEvent for Minecart placement 2b2d084a Add InventoryView#getSlotType 01a9959a Do not use deprecated ItemSpawnEvent constructor 9642498d SPIGOT-4547: Call EntitySpawnEvent as general spawn fallback event 963f4a5f Add PlayerItemDamageEvent 63db0445 Add API to get / set base arrow damage 531c25d7 Add CraftMagicNumbers.MAPPINGS_VERSION for use by NMS plugins d05c8b14 Mappings Update bd36e200 SPIGOT-4551: Ignore invalid attribute modifier slots Spigot Changes: 518206a1 Remove redundant trove depend 1959ad21 MC-11211,SPIGOT-4552: Fix placing double slabs at y = 255 29ab5e43 SPIGOT-3661: Allow arguments in restart-script 7cc46316 SPIGOT-852: Growth modifiers for beetroots, potatoes, carrots 82e117e1 Squelch "fatal: Resolve operation not in progress" message 0a1a68e7 Mappings Update & Patch Rebuild
2019-01-01 03:15:55 +00:00
@@ -11,6 +11,7 @@ public final class IEntitySelector {
public static final Predicate<Entity> c = (entity) -> {
return entity.isAlive() && !entity.isVehicle() && !entity.isPassenger();
};
+ public static final Predicate<Entity> isInventory() { return d; } // Paper - OBFHELPER
public static final Predicate<Entity> d = (entity) -> {
return entity instanceof IInventory && entity.isAlive();
};