feat: 轻量化 ScaledProcessingPattern，改为最小字段与延迟计算，
减少构造时的大量临时对象分配，降低 GC 压力并提升高并发/多 provider 场景下的性能
This commit is contained in:
C-H716
parent
ef8fa756ed
commit
d4095e64ea
|
|
@ -17,27 +17,21 @@ import java.util.Objects;
|
|||
*/
|
||||
public final class ScaledProcessingPattern implements IPatternDetails {
|
||||
|
||||
private final AEProcessingPattern original; // 原始样板引用
|
||||
private final AEItemKey definition; // 样板物品
|
||||
private final GenericStack[] sparseInputs; // 缩放后的稀疏输入
|
||||
private final GenericStack[] sparseOutputs; // 缩放后的稀疏输出
|
||||
private final IInput[] inputs; // 缩放后的压缩输入
|
||||
private final GenericStack[] condensedOutputs; // 缩放后的压缩输出
|
||||
// 最小化实例字段：只保留原始样板引用、定义和倍数
|
||||
private final AEProcessingPattern original; // 原始样板引用
|
||||
private final AEItemKey definition; // 样板物品（直接委托自 original）
|
||||
private final long multiplier; // 乘数（外部可视为视图参数）
|
||||
|
||||
public ScaledProcessingPattern(
|
||||
AEProcessingPattern original,
|
||||
AEItemKey definition,
|
||||
GenericStack[] sparseInputs,
|
||||
GenericStack[] sparseOutputs,
|
||||
IInput[] inputs,
|
||||
GenericStack[] condensedOutputs
|
||||
) {
|
||||
// 延迟计算缓存（轻量化实例时避免在构造器中分配大数组）
|
||||
private transient volatile IInput[] inputsCache;
|
||||
private transient volatile GenericStack[] outputsCache;
|
||||
private transient volatile GenericStack[] sparseInputsCache;
|
||||
private transient volatile GenericStack[] sparseOutputsCache;
|
||||
|
||||
public ScaledProcessingPattern(AEProcessingPattern original, AEItemKey definition, long multiplier) {
|
||||
this.original = Objects.requireNonNull(original);
|
||||
this.definition = Objects.requireNonNull(definition);
|
||||
this.sparseInputs = Objects.requireNonNull(sparseInputs);
|
||||
this.sparseOutputs = Objects.requireNonNull(sparseOutputs);
|
||||
this.inputs = Objects.requireNonNull(inputs);
|
||||
this.condensedOutputs = Objects.requireNonNull(condensedOutputs);
|
||||
this.multiplier = multiplier <= 0 ? 1L : multiplier;
|
||||
}
|
||||
|
||||
/* -------------------- API 实现 -------------------- */
|
||||
|
|
@ -53,25 +47,91 @@ public final class ScaledProcessingPattern implements IPatternDetails {
|
|||
|
||||
@Override
|
||||
public IInput[] getInputs() {
|
||||
return inputs;
|
||||
IInput[] cached = this.inputsCache;
|
||||
if (cached == null) {
|
||||
synchronized (this) {
|
||||
cached = this.inputsCache;
|
||||
if (cached == null) {
|
||||
var base = original.getInputs();
|
||||
IInput[] arr = new IInput[base.length];
|
||||
for (int i = 0; i < base.length; i++) {
|
||||
var in = base[i];
|
||||
// 不复制 template 数组，直接复用原始的 possible inputs；仅放大 multiplier
|
||||
arr[i] = new Input(in.getPossibleInputs(), in.getMultiplier() * this.multiplier);
|
||||
}
|
||||
this.inputsCache = arr;
|
||||
cached = arr;
|
||||
}
|
||||
}
|
||||
}
|
||||
return cached;
|
||||
}
|
||||
|
||||
@Override
|
||||
public GenericStack[] getOutputs() {
|
||||
return condensedOutputs;
|
||||
GenericStack[] cached = this.outputsCache;
|
||||
if (cached == null) {
|
||||
synchronized (this) {
|
||||
cached = this.outputsCache;
|
||||
if (cached == null) {
|
||||
var baseOutputs = original.getOutputs();
|
||||
GenericStack[] arr = new GenericStack[baseOutputs.length];
|
||||
for (int i = 0; i < baseOutputs.length; i++) {
|
||||
var o = baseOutputs[i];
|
||||
if (o != null) arr[i] = new GenericStack(o.what(), o.amount() * this.multiplier);
|
||||
}
|
||||
this.outputsCache = arr;
|
||||
cached = arr;
|
||||
}
|
||||
}
|
||||
}
|
||||
return cached;
|
||||
}
|
||||
|
||||
public GenericStack[] getSparseInputs() {
|
||||
return sparseInputs;
|
||||
GenericStack[] cached = this.sparseInputsCache;
|
||||
if (cached == null) {
|
||||
synchronized (this) {
|
||||
cached = this.sparseInputsCache;
|
||||
if (cached == null) {
|
||||
var base = original.getSparseInputs();
|
||||
GenericStack[] arr = new GenericStack[base.length];
|
||||
for (int i = 0; i < base.length; i++) {
|
||||
var v = base[i];
|
||||
if (v != null) arr[i] = new GenericStack(v.what(), v.amount() * this.multiplier);
|
||||
}
|
||||
this.sparseInputsCache = arr;
|
||||
cached = arr;
|
||||
}
|
||||
}
|
||||
}
|
||||
return cached;
|
||||
}
|
||||
|
||||
public GenericStack[] getSparseOutputs() {
|
||||
return sparseOutputs;
|
||||
GenericStack[] cached = this.sparseOutputsCache;
|
||||
if (cached == null) {
|
||||
synchronized (this) {
|
||||
cached = this.sparseOutputsCache;
|
||||
if (cached == null) {
|
||||
var base = original.getSparseOutputs();
|
||||
GenericStack[] arr = new GenericStack[base.length];
|
||||
for (int i = 0; i < base.length; i++) {
|
||||
var v = base[i];
|
||||
if (v != null) arr[i] = new GenericStack(v.what(), v.amount() * this.multiplier);
|
||||
}
|
||||
this.sparseOutputsCache = arr;
|
||||
cached = arr;
|
||||
}
|
||||
}
|
||||
}
|
||||
return cached;
|
||||
}
|
||||
|
||||
@Override
|
||||
public GenericStack getPrimaryOutput() {
|
||||
if (condensedOutputs.length > 0) return condensedOutputs[0];
|
||||
var outs = getOutputs();
|
||||
if (outs.length > 0 && outs[0] != null) return outs[0];
|
||||
return original.getPrimaryOutput();
|
||||
}
|
||||
|
||||
|
|
@ -82,26 +142,29 @@ public final class ScaledProcessingPattern implements IPatternDetails {
|
|||
|
||||
@Override
|
||||
public void pushInputsToExternalInventory(KeyCounter[] inputHolder, PatternInputSink inputSink) {
|
||||
// 保持和 AEProcessingPattern 一致，用 sparseInputs 驱动
|
||||
if (sparseInputs.length == inputs.length) {
|
||||
// 使用 lazy 计算的 sparseInputs 与 inputs 来驱动；当两者长度一致时直接委托
|
||||
GenericStack[] sInputs = getSparseInputs();
|
||||
IInput[] ins = getInputs();
|
||||
if (sInputs.length == ins.length) {
|
||||
IPatternDetails.super.pushInputsToExternalInventory(inputHolder, inputSink);
|
||||
} else {
|
||||
KeyCounter allInputs = new KeyCounter();
|
||||
for (KeyCounter counter : inputHolder) {
|
||||
allInputs.addAll(counter);
|
||||
}
|
||||
for (GenericStack sparseInput : sparseInputs) {
|
||||
if (sparseInput != null) {
|
||||
AEKey key = sparseInput.what();
|
||||
long amount = sparseInput.amount();
|
||||
long available = allInputs.get(key);
|
||||
if (available < amount) {
|
||||
throw new RuntimeException("Expected at least %d of %s when pushing scaled pattern, but only %d available"
|
||||
.formatted(amount, key, available));
|
||||
}
|
||||
inputSink.pushInput(key, amount);
|
||||
allInputs.remove(key, amount);
|
||||
return;
|
||||
}
|
||||
|
||||
KeyCounter allInputs = new KeyCounter();
|
||||
for (KeyCounter counter : inputHolder) {
|
||||
allInputs.addAll(counter);
|
||||
}
|
||||
for (GenericStack sparseInput : sInputs) {
|
||||
if (sparseInput != null) {
|
||||
AEKey key = sparseInput.what();
|
||||
long amount = sparseInput.amount();
|
||||
long available = allInputs.get(key);
|
||||
if (available < amount) {
|
||||
throw new RuntimeException("Expected at least %d of %s when pushing scaled pattern, but only %d available"
|
||||
.formatted(amount, key, available));
|
||||
}
|
||||
inputSink.pushInput(key, amount);
|
||||
allInputs.remove(key, amount);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -117,18 +180,22 @@ public final class ScaledProcessingPattern implements IPatternDetails {
|
|||
this.multiplier = multiplier;
|
||||
}
|
||||
|
||||
@Override
|
||||
public GenericStack[] getPossibleInputs() {
|
||||
return this.template;
|
||||
}
|
||||
|
||||
@Override
|
||||
public long getMultiplier() {
|
||||
return this.multiplier;
|
||||
}
|
||||
|
||||
@Override
|
||||
public boolean isValid(AEKey input, Level level) {
|
||||
return input.matches(this.template[0]);
|
||||
}
|
||||
|
||||
@Override
|
||||
public @Nullable AEKey getRemainingKey(AEKey template) {
|
||||
return null;
|
||||
}
|
||||
|
|
|
|||
|
|
@ -71,48 +71,7 @@ public final class PatternScaler {
|
|||
// 配置读取异常时不施加上限
|
||||
}
|
||||
|
||||
// 构建压缩输入（将每个输入的 multiplier 翻倍，保留每个模板的原始数量）
|
||||
IInput[] scaledInputs = new IInput[baseInputs.length];
|
||||
for (int i = 0; i < baseInputs.length; i++) {
|
||||
var in = baseInputs[i];
|
||||
var template = in.getPossibleInputs();
|
||||
GenericStack[] scaledTemplates = new GenericStack[template.length];
|
||||
for (int j = 0; j < template.length; j++) {
|
||||
scaledTemplates[j] = new GenericStack(template[j].what(), template[j].amount());
|
||||
}
|
||||
scaledInputs[i] = new ScaledProcessingPattern.Input(scaledTemplates, in.getMultiplier() * multiplier);
|
||||
}
|
||||
|
||||
/* 4. 构建压缩输出 */
|
||||
GenericStack[] scaledCondensedOutputs = new GenericStack[baseOutputs.length];
|
||||
for (int i = 0; i < baseOutputs.length; i++) {
|
||||
GenericStack out = baseOutputs[i];
|
||||
if (out != null) {
|
||||
scaledCondensedOutputs[i] = new GenericStack(out.what(), out.amount() * multiplier);
|
||||
}
|
||||
}
|
||||
|
||||
// 构建并打印稀疏表示（直接按 multiplier 放大）
|
||||
GenericStack[] scaledSparseInputs = new GenericStack[baseSparseInputs.length];
|
||||
for (int i = 0; i < baseSparseInputs.length; i++) {
|
||||
var in = baseSparseInputs[i];
|
||||
if (in != null) {
|
||||
scaledSparseInputs[i] = new GenericStack(in.what(), in.amount() * multiplier);
|
||||
}
|
||||
}
|
||||
GenericStack[] scaledSparseOutputs = new GenericStack[baseSparseOutputs.length];
|
||||
for (int i = 0; i < baseSparseOutputs.length; i++) {
|
||||
var out = baseSparseOutputs[i];
|
||||
if (out != null) {
|
||||
scaledSparseOutputs[i] = new GenericStack(out.what(), out.amount() * multiplier);
|
||||
}
|
||||
}
|
||||
|
||||
return new ScaledProcessingPattern(base,
|
||||
base.getDefinition(),
|
||||
scaledSparseInputs,
|
||||
scaledSparseOutputs,
|
||||
scaledInputs,
|
||||
scaledCondensedOutputs);
|
||||
// 仅使用 multiplier 构建轻量化 ScaledProcessingPattern（具体视图按需计算）
|
||||
return new ScaledProcessingPattern(base, base.getDefinition(), multiplier);
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -42,7 +42,6 @@
|
|||
"ae2.autopattern.CraftingTreeNodeMixin",
|
||||
"ae2.autopattern.CraftingTreeProcessMixin",
|
||||
"ae2.autopattern.PatternProviderLogicContainsRedirectMixin",
|
||||
"ae2.autopattern.adaptation.AdvPatternProviderLogicContainsRedirectMixin",
|
||||
"ae2.helpers.PatternProviderLogicAdvancedMixin",
|
||||
"ae2.helpers.PatternProviderLogicDoublingMixin",
|
||||
"ae2.menu.ContainerPatternEncodingTermMenuMixin",
|
||||
|
|
|
|||
Loading…
Reference in New Issue
Block a user