better bp caching, xf cache, optimize fifo with realloc and remove double lookups (#110)

* optimize fifo with realloc and remove double lookups

* bp cache, xf cache, and last pipeline ref cache

* remove double lookup from find/try_emplace

* change name

* Move xfRegCache

* Update gx_test_stubs.cpp

---------

Co-authored-by: Luke Street <luke@street.dev>

lib/gx/command_processor.cpp

@@ -504,6 +504,7 @@ static void handle_bp(u32 value, bool bigEndian) {
     g_gxState.bpRegCache[0xFE] = 0x00FFFFFF;
     const u32 merged = (g_gxState.bpRegCache[regId] & ~ssMask) | (value & ssMask);
     value = (regId << 24) | (merged & 0x00FFFFFF);
+    if (g_gxState.bpRegCache[regId] == value) return;
     g_gxState.bpRegCache[regId] = value;
   }
 
@@ -1275,7 +1276,10 @@ static void handle_xf(const u8* data, u32& pos, u32 size, bool bigEndian) {
     for (u32 i = 0; i < count; i++) {
       u32 reg = xfAddr + i;
       u32 val = read_u32(xfData + i * 4, bigEndian);
-      f32 fval = read_f32(xfData + i * 4, bigEndian);
+
+      // Skip scalar register writes that haven't changed (viewport/projection handled below)
+      if (reg <= 0x19 && val == g_gxState.xfRegCache[reg]) continue;
+      if (reg <= 0x19) g_gxState.xfRegCache[reg] = val;
 
       switch (reg) {
       case 0x08:

lib/gx/fifo.cpp

@@ -2,6 +2,7 @@
 #include "command_processor.hpp"
 #include "../internal.hpp"
 
+#include <cstdlib>
 #include <cstring>
 
 namespace aurora::gx::fifo {
@@ -19,8 +20,8 @@ uint32_t sDlWritePos = 0;
 
 void init() {
   constexpr uint32_t initialCapacity = 64 * 1024;
-  delete[] detail::sBufferData;
-  detail::sBufferData = new uint8_t[initialCapacity];
+  free(detail::sBufferData);
+  detail::sBufferData = static_cast<uint8_t*>(malloc(initialCapacity));
   detail::sBufferSize = 0;
   detail::sBufferCapacity = initialCapacity;
   detail::sInDisplayList = false;
@@ -31,17 +32,9 @@ void init() {
 
 void write_data_grow(const void* data, uint32_t length) {
   uint32_t needed = detail::sBufferSize + length;
-  uint32_t newCap = detail::sBufferCapacity * 2;
-  if (newCap < needed) {
-    newCap = needed;
-  }
-  auto* newBuf = new uint8_t[newCap];
-  if (detail::sBufferSize > 0) {
-    std::memcpy(newBuf, detail::sBufferData, detail::sBufferSize);
-  }
-  std::memcpy(newBuf + detail::sBufferSize, data, length);
-  delete[] detail::sBufferData;
-  detail::sBufferData = newBuf;
+  uint32_t newCap = std::max(detail::sBufferCapacity * 2, needed);
+  detail::sBufferData = static_cast<uint8_t*>(realloc(detail::sBufferData, newCap));
+  std::memcpy(detail::sBufferData + detail::sBufferSize, data, length);
   detail::sBufferSize = needed;
   detail::sBufferCapacity = newCap;
 }

lib/gx/gx.hpp

@@ -373,6 +373,7 @@ struct GXState {
     regs[0xFE] = 0x00FFFFFF;
     return regs;
   }();
+  std::array<u32, 0x1A> xfRegCache;
 
   void clearVtxSizeCache() { lastVtxFmt = GX_MAX_VTXFMT; }
 };

lib/input.cpp

@@ -80,35 +80,35 @@ SDL_JoystickID add_controller(SDL_JoystickID which) noexcept {
 }
 
 void remove_controller(Uint32 instance) noexcept {
-  if (g_GameControllers.find(instance) != g_GameControllers.end()) {
-    SDL_CloseGamepad(g_GameControllers[instance].m_controller);
-    g_GameControllers.erase(instance);
+  if (auto it = g_GameControllers.find(instance); it != g_GameControllers.end()) {
+    SDL_CloseGamepad(it->second.m_controller);
+    g_GameControllers.erase(it);
   }
 }
 
 bool is_gamecube(Uint32 instance) noexcept {
-  if (g_GameControllers.find(instance) != g_GameControllers.end()) {
-    return g_GameControllers[instance].m_isGameCube;
+  if (auto it = g_GameControllers.find(instance); it != g_GameControllers.end()) {
+    return it->second.m_isGameCube;
   }
   return false;
 }
 
 int32_t player_index(Uint32 instance) noexcept {
-  if (g_GameControllers.find(instance) != g_GameControllers.end()) {
-    return SDL_GetGamepadPlayerIndex(g_GameControllers[instance].m_controller);
+  if (auto it = g_GameControllers.find(instance); it != g_GameControllers.end()) {
+    return SDL_GetGamepadPlayerIndex(it->second.m_controller);
   }
   return -1;
 }
 
 void set_player_index(Uint32 instance, Sint32 index) noexcept {
-  if (g_GameControllers.find(instance) != g_GameControllers.end()) {
-    SDL_SetGamepadPlayerIndex(g_GameControllers[instance].m_controller, index);
+  if (auto it = g_GameControllers.find(instance); it != g_GameControllers.end()) {
+    SDL_SetGamepadPlayerIndex(it->second.m_controller, index);
   }
 }
 
 std::string controller_name(Uint32 instance) noexcept {
-  if (g_GameControllers.find(instance) != g_GameControllers.end()) {
-    const auto* name = SDL_GetGamepadName(g_GameControllers[instance].m_controller);
+  if (auto it = g_GameControllers.find(instance); it != g_GameControllers.end()) {
+    const auto* name = SDL_GetGamepadName(it->second.m_controller);
     if (name != nullptr) {
       return {name};
     }
@@ -117,18 +117,16 @@ std::string controller_name(Uint32 instance) noexcept {
 }
 
 bool controller_has_rumble(Uint32 instance) noexcept {
-  if (g_GameControllers.find(instance) != g_GameControllers.end()) {
-    return g_GameControllers[instance].m_hasRumble;
+  if (auto it = g_GameControllers.find(instance); it != g_GameControllers.end()) {
+    return it->second.m_hasRumble;
   }
-
   return false;
 }
 
 void controller_rumble(uint32_t instance, uint16_t low_freq_intensity, uint16_t high_freq_intensity,
                        uint16_t duration_ms) noexcept {
-
-  if (g_GameControllers.find(instance) != g_GameControllers.end()) {
-    SDL_RumbleGamepad(g_GameControllers[instance].m_controller, low_freq_intensity, high_freq_intensity, duration_ms);
+  if (auto it = g_GameControllers.find(instance); it != g_GameControllers.end()) {
+    SDL_RumbleGamepad(it->second.m_controller, low_freq_intensity, high_freq_intensity, duration_ms);
   }
 }
 

tests/gx_test_stubs.cpp

@@ -70,10 +70,7 @@ namespace aurora::gx {
 void populate_pipeline_config(PipelineConfig& config, GXPrimitive primitive, GXVtxFmt fmt) noexcept {
   // No-op for tests
 }
-GXBindGroups build_bind_groups(const ShaderInfo& info, const ShaderConfig& config,
-                               const BindGroupRanges& ranges) noexcept {
-  return {};
-}
+GXBindGroups build_bind_groups(const ShaderInfo& info) noexcept { return {}; }
 ShaderInfo build_shader_info(const ShaderConfig& config) noexcept { return {}; }
 gfx::Range build_uniform(const ShaderInfo& info, uint32_t vtxStart, const BindGroupRanges& ranges) noexcept {
   return {};