some chore

StgSharp.Common/HighPerformance/Memory/L4.InternalOp.cs

@@ -0,0 +1,81 @@
+//-----------------------------------------------------------------------
+// -----------------------------------------------------------------------
+// file="L4.InternalOp"
+// Project: StgSharp
+// AuthorGroup: Nitload
+// Copyright (c) Nitload. All rights reserved.
+//     
+// Permission is hereby granted, free of charge, to any person obtaining a copy
+// of this software and associated documentation files (the "Software"), to deal
+// in the Software without restriction, including without limitation the rights
+// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+// copies of the Software, and to permit persons to whom the Software is
+// furnished to do so, subject to the following conditions:
+//
+// The above copyright notice and this permission notice shall be included in all
+// copies or substantial portions of the Software.
+//
+// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+// SOFTWARE.
+//     
+// -----------------------------------------------------------------------
+// -----------------------------------------------------------------------
+using System;
+using System.Collections.Generic;
+using System.Linq;
+using System.Runtime.Intrinsics;
+using System.Runtime.Intrinsics.X86;
+using System.Text;
+using System.Threading.Tasks;
+
+namespace StgSharp.HighPerformance.Memory
+{
+    public partial class L4
+    {
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        private unsafe void AgeAll()
+        {
+            Vector128<byte> one = Vector128.Create((byte)0x01);
+            Vector128<byte> max = Vector128.Create((byte)0x03);
+            byte* p = (byte*)_eviction;
+            for (int i = 0; i < CacheLineCount; i += 16)
+            {
+                Vector128<byte> v = Sse2.LoadVector128(p + i);
+                v = Sse2.AddSaturate(v, one);
+                v = Sse2.Min(v, max);
+                Sse2.Store(p + i, v);
+            }
+        }
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        private unsafe void Evict(
+                            int count = 1
+        )
+        {
+            int remain = count;
+            Vector128<byte> one = Vector128.Create((byte)0x01);
+            Vector128<byte> three = Vector128.Create((byte)0x03);
+            byte* p = (byte*)_eviction;
+            for (int i = 0; i < CacheLineCount && remain > 0; i += 16)
+            {
+                Vector128<byte> v = Sse2.LoadVector128(p + i);
+                Vector128<byte> mask = Sse2.CompareEqual(v, three);
+                int evictMask = Sse2.MoveMask(mask);
+                while (evictMask != 0 && remain > 0)
+                {
+                    int index = BitOperations.TrailingZeroCount(evictMask);
+                    evictMask &= (evictMask - 1);
+                    EvictLine(i + index);
+                    remain--;
+                }
+            }
+        }
+
+    }
+}

StgSharp.Common/HighPerformance/Memory/L4.Prefetch.cs

@@ -45,6 +45,7 @@ namespace StgSharp.HighPerformance.Memory
             Span<CacheLinePrediction> span = new(_prediction, PredictionCount);
             int actualCount = PrefetchPredict.Predict(span);
             int missIndex = 0;
+            AgeAll();
             for (int i = 0; i < actualCount; i++)
             {
                 ref CacheLinePrediction handle = ref span[i];
@@ -62,6 +63,9 @@ namespace StgSharp.HighPerformance.Memory
             {
                 ref CacheLinePrediction handle = ref span[i];
                 nuint address = handle.Address;
+                nuint mapPtr = _cacheLineAllocator.Allocate();
+                CacheLineHead* head = (CacheLineHead*)_entryAllocator.Allocate();
+
 
                 // pre fetch them here
             }

StgSharp.Common/HighPerformance/Memory/L4_design.md

@@ -171,15 +171,7 @@ SoA 布局 — 平行数组
 
 **逐出标志位布局**（`_evictionFlags[i]`）：
 
-```
-  bit:  7  6  5  4  3  2  1  0
-        ─────────  ─  ─────  ─
-        保留       C  RRPV   D
-
-  D     = bit 0 = dirty（0=只读副本, 1=有脏数据待写回）
-  RRPV  = bit 1-2 = Re-Reference Prediction Value（0~3）
-  C     = bit 3  = costClass（0=低成本, 1=高成本）
-```
+RRPV数组使用byte存储每一个cacheline的使用情况，其低二位表示如下逻辑
 
 **RRPV 语义**：
 
@@ -203,7 +195,7 @@ SoA 布局 — 平行数组
 | 对比项         | v0.1 环形链表           | v0.3 RRPV 掩码 + 回调       |
 |---------------|------------------------|------------------------------|
 | 数据结构       | 4 条双向链表 + next 指针 | byte 数组 + 函数指针数组      |
-| 升温操作       | 摘节点 + 插节点（4 指针）| `&= 0xF9`（1 字节写入）      |
+| 升温操作       | 摘节点 + 插节点（4 指针）| `+= 1`（1 字节写入）      |
 | 老化操作       | 逐节点移链表             | SIMD 批量 +1（16 个/周期）   |
 | 逐出扫描       | 遍历链表头（指针追逐）    | SIMD 模式匹配（16 个/周期）  |
 | 缓存友好性     | 差（指针跳跃）           | 好（连续内存扫描）            |