[video_core/maxwell3d] compute macro param address on demand (#4067)

GetMacroAddress only reads a couple of indices per macro, but ProcessMacro was
then building a full std::vector GPUVAddr> with one push_back per parameter word
every submission. macro_segments already holds base, count per chunk, so
GetMacroAddress can just walk it instead, drops the per-word loop, a
.clear(), and a vector member.

Also makes it so it returns on the first match in the macro dispatch instead of running every
std::get_if check.

Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4067
Reviewed-by: Lizzie <lizzie@eden-emu.dev>
Reviewed-by: CamilleLaVey <camillelavey99@gmail.com>

3 files changed, 22 insertions, 20 deletions

diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp
index 3474ad6d63..53b75726bd 100644
--- a/src/video_core/engines/maxwell_3d.cpp
+++ b/src/video_core/engines/maxwell_3d.cpp
@@ -208,9 +208,6 @@ void Maxwell3D::ProcessMacro(Core::System& system, u32 method, const u32* base_s
     }
 
     macro_params.insert(macro_params.end(), base_start, base_start + amount);
-    for (size_t i = 0; i < amount; i++) {
-        macro_addresses.push_back(current_dma_segment + i * sizeof(u32));
-    }
     macro_segments.emplace_back(current_dma_segment, amount);
     current_macro_dirty |= current_dirty;
     current_dirty = false;
@@ -220,7 +217,6 @@ void Maxwell3D::ProcessMacro(Core::System& system, u32 method, const u32* base_s
         ConsumeSink(system);
         CallMacroMethod(system, executing_macro, macro_params);
         macro_params.clear();
-        macro_addresses.clear();
         macro_segments.clear();
         current_macro_dirty = false;
     }
diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h
index 823be0f63e..bbde8b9986 100644
--- a/src/video_core/engines/maxwell_3d.h
+++ b/src/video_core/engines/maxwell_3d.h
@@ -3158,7 +3158,14 @@ public:
     DrawManager draw_manager;
 
     GPUVAddr GetMacroAddress(size_t index) const {
-        return macro_addresses[index];
+        size_t base = 0;
+        for (const auto& [addr, count] : macro_segments) {
+            if (index < base + count) {
+                return addr + (index - base) * sizeof(u32);
+            }
+            base += count;
+        }
+        return 0;
     }
 
     void RefreshParameters() {
@@ -3261,7 +3268,6 @@ private:
     bool execute_on{true};
 
     std::vector<std::pair<GPUVAddr, size_t>> macro_segments;
-    std::vector<GPUVAddr> macro_addresses;
     bool current_macro_dirty{};
 };
 
diff --git a/src/video_core/macro.cpp b/src/video_core/macro.cpp
index 2833f95ee9..0d026a1ce5 100644
--- a/src/video_core/macro.cpp
+++ b/src/video_core/macro.cpp
@@ -1356,33 +1356,33 @@ static void Dump(u64 hash, std::span<const u32> code, bool decompiled = false) {
 void MacroEngine::Execute(Core::System& system, Engines::Maxwell3D& maxwell3d, u32 method, std::span<const u32> parameters) {
     auto const execute_variant = [&system, &maxwell3d, &parameters, method](AnyCachedMacro& acm) {
         if (auto a = std::get_if<HLE_DrawArraysIndirect>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<HLE_DrawIndexedIndirect>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<HLE_MultiDrawIndexedIndirectCount>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<HLE_MultiLayerClear>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<HLE_C713C83D8F63CCF3>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<HLE_D7333D26E0A93EDE>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<HLE_BindShader>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<HLE_SetRasterBoundingBox>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<HLE_ClearConstBuffer>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<HLE_ClearMemory>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<HLE_TransformFeedbackSetup>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<HLE_DrawIndirectByteCount>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<MacroInterpreterImpl>(&acm))
-            a->Execute(system, maxwell3d, parameters, method);
+            return a->Execute(system, maxwell3d, parameters, method);
         if (auto a = std::get_if<std::unique_ptr<DynamicCachedMacro>>(&acm))
-            a->get()->Execute(system, maxwell3d, parameters, method);
+            return a->get()->Execute(system, maxwell3d, parameters, method);
     };
     if (auto const it = macro_cache.find(method); it != macro_cache.end()) {
         auto& ci = it->second;