perf(engine): implement multi-threaded DDA raycasting with worker thread pool

casino3d.inc

@@ -34,6 +34,10 @@ CopyFileA            PROTO :DWORD,:DWORD,:DWORD
 lstrcmpiA           PROTO :DWORD,:DWORD
 GetFileSize          PROTO :DWORD,:DWORD
 Sleep                PROTO :DWORD
+CreateThread         PROTO :DWORD,:DWORD,:DWORD,:DWORD,:DWORD,:DWORD
+CreateEventA         PROTO :DWORD,:DWORD,:DWORD,:DWORD
+SetEvent             PROTO :DWORD
+WaitForSingleObject  PROTO :DWORD,:DWORD
 
 C3D_POINT STRUCT
     x dd ?
@@ -164,6 +168,13 @@ h3DWnd             dd 0
 h3DDC              dd 0
 hCasinoConsole     dd 0
 
+ALIGN 4
+hRaycastThread       dd 0
+hRaycastStartEvent   dd 0
+hRaycastDoneEvent    dd 0
+bRaycastQuit         dd 0
+raycastThreadId      dd 0
+
 bmi3D              C3D_BITMAPINFO <>
 msg3D              C3D_MSG <>
 wc3D               C3D_WNDCLASSEX <>
@@ -696,7 +707,7 @@ DrawString8x8 ENDP
 ; --- True 3D Perspective Floor & Ceiling Rendering ---
 
 ; --- DDA Raycasting Walls ---
-RenderWalls PROC
+RenderWallsRange PROC colStart:DWORD, colLimit:DWORD
     local col:DWORD
     local originalStart:DWORD
     local camX:DWORD
@@ -746,9 +757,11 @@ RenderWalls PROC
     idiv ecx
     mov playerPlaneY, eax
 
-    mov col, 0
+    mov eax, colStart
+    mov col, eax
 col_loop:
-    cmp col, SCREEN_W
+    mov eax, colLimit
+    cmp col, eax
     jge walls_done
 
     mov eax, col
@@ -1231,6 +1244,69 @@ walls_done:
     pop esi
     pop ebx
     ret
+RenderWallsRange ENDP
+
+; --- Raycasting Background Worker Thread ---
+RaycastWorkerProc PROC uses ebx esi edi lpParam:DWORD
+worker_loop:
+    invoke WaitForSingleObject, hRaycastStartEvent, 0FFFFFFFFh
+    cmp bRaycastQuit, 1
+    je worker_exit
+    invoke RenderWallsRange, 320, 640
+    invoke SetEvent, hRaycastDoneEvent
+    jmp worker_loop
+worker_exit:
+    xor eax, eax
+    ret
+RaycastWorkerProc ENDP
+
+; --- Multi-Threaded DDA Raycasting Walls Dispatcher ---
+RenderWalls PROC
+    push ebx
+    push esi
+    push edi
+
+    mov eax, playerAngle
+    and eax, 511
+    mov ebx, [tableCos + eax*4]
+    mov playerDirX, ebx
+    mov ebx, [tableSin + eax*4]
+    mov playerDirY, ebx
+
+    mov eax, ebx
+    imul eax, -66
+    cdq
+    mov ecx, 100
+    idiv ecx
+    mov playerPlaneX, eax
+
+    mov eax, playerDirX
+    imul eax, 66
+    cdq
+    idiv ecx
+    mov playerPlaneY, eax
+
+    cmp hRaycastStartEvent, 0
+    je single_thread_fallback
+
+    ; Signal background worker to render columns [320..640)
+    invoke SetEvent, hRaycastStartEvent
+
+    ; Main thread processes columns [0..320) in parallel
+    invoke RenderWallsRange, 0, 320
+
+    ; Synchronize with worker completion
+    invoke WaitForSingleObject, hRaycastDoneEvent, 0FFFFFFFFh
+    jmp walls_finished
+
+single_thread_fallback:
+    invoke RenderWallsRange, 0, SCREEN_W
+
+walls_finished:
+    pop edi
+    pop esi
+    pop ebx
+    ret
 RenderWalls ENDP
 
 ; --- Render 3D Billboard Sprites ---
@@ -4856,6 +4932,15 @@ Init3DEngine PROC
 
     call InitializeCasinoSprites3D
 
+    ; Initialize raycasting worker thread and sync events
+    invoke CreateEventA, 0, 0, 0, 0
+    mov hRaycastStartEvent, eax
+    invoke CreateEventA, 0, 0, 0, 0
+    mov hRaycastDoneEvent, eax
+    mov bRaycastQuit, 0
+    invoke CreateThread, 0, 65536, OFFSET RaycastWorkerProc, 0, 0, ADDR raycastThreadId
+    mov hRaycastThread, eax
+
     call InitHall3D
     call BeginStartup3D
     call RenderStartup3D
@@ -5487,6 +5572,18 @@ engine_idle:
     invoke Sleep, 50
     jmp engine_loop
 engine_shutdown:
+    cmp hRaycastThread, 0
+    je @F
+    mov bRaycastQuit, 1
+    invoke SetEvent, hRaycastStartEvent
+    invoke WaitForSingleObject, hRaycastThread, 1000
+    invoke CloseHandle, hRaycastThread
+    invoke CloseHandle, hRaycastStartEvent
+    invoke CloseHandle, hRaycastDoneEvent
+    mov hRaycastThread, 0
+    mov hRaycastStartEvent, 0
+    mov hRaycastDoneEvent, 0
+@@:
     call SaveProgress3D
     invoke PlayCasinoSound3D, 0, 0, 0
     call FreeRussianAssets3D