diff --git a/Core/MIPS/IR/IRInterpreter.cpp b/Core/MIPS/IR/IRInterpreter.cpp
index eef1ac6420..ff39c19096 100644
--- a/Core/MIPS/IR/IRInterpreter.cpp
+++ b/Core/MIPS/IR/IRInterpreter.cpp
@@ -160,7 +160,6 @@ void IRRestoreRounding() {
 #endif
 }
 
-// We cannot use NEON on ARM32 here until we make it a hard dependency. We can, however, on ARM64.
 u32 IRInterpret(MIPSState *mips, const IRInst *inst) {
 	while (true) {
 		switch (inst->op) {
diff --git a/Core/MIPS/IR/IRJit.cpp b/Core/MIPS/IR/IRJit.cpp
index a11b882630..94f85dcb83 100644
--- a/Core/MIPS/IR/IRJit.cpp
+++ b/Core/MIPS/IR/IRJit.cpp
@@ -294,21 +294,20 @@ int IRBlockCache::GetBlockNumFromIRArenaOffset(int offset) const {
 			high = mid - 1;
 		}
 	}
-
-#ifndef _DEBUG
-	// Then, in debug builds, cross check the result.
+	return found;
+#if 1
 	return found;
 #else
-	// TODO: Optimize if we need to call this often.
+	// Cross check the result. This is not fast so normally not enabled. Called a lot when IR_PROFILING is on.
 	for (int i = 0; i < (int)blocks_.size(); i++) {
 		if (blocks_[i].GetIRArenaOffset() == offset) {
 			_dbg_assert_(i == found);
 			return i;
 		}
 	}
-#endif
 	_dbg_assert_(found == -1);
 	return -1;
+#endif
 }
 
 std::vector<int> IRBlockCache::FindInvalidatedBlockNumbers(u32 address, u32 lengthInBytes) {
diff --git a/Core/MIPS/JitCommon/JitBlockCache.h b/Core/MIPS/JitCommon/JitBlockCache.h
index 6cf805a6e5..13365cea13 100644
--- a/Core/MIPS/JitCommon/JitBlockCache.h
+++ b/Core/MIPS/JitCommon/JitBlockCache.h
@@ -102,17 +102,18 @@ struct JitBlockDebugInfo {
 	std::vector<std::string> targetDisasm;
 };
 
+struct JitBlockProfileStats {
+	int64_t executions;
+	int64_t totalNanos;
+};
+
+// Only used for debugging, so keeping it tight is not that important.
 struct JitBlockMeta {
 	bool valid;
 	uint32_t addr;
 	uint32_t sizeInBytes;
 };
 
-struct JitBlockProfileStats {
-	int64_t executions;
-	int64_t totalNanos;
-};
-
 class JitBlockCacheDebugInterface {
 public:
 	virtual int GetNumBlocks() const = 0;
diff --git a/UI/ImDebugger/ImJitViewer.cpp b/UI/ImDebugger/ImJitViewer.cpp
index b3a6f9d66b..490e932de0 100644
--- a/UI/ImDebugger/ImJitViewer.cpp
+++ b/UI/ImDebugger/ImJitViewer.cpp
@@ -7,6 +7,7 @@
 #include "Core/MIPS/JitCommon/JitBlockCache.h"
 #include "Core/MIPS/JitCommon/JitCommon.h"
 #include "Core/MIPS/JitCommon/JitState.h"
+#include "Core/MIPS/IR/IRJit.h"
 
 void ImJitViewerWindow::GoToBlockAtAddr(u32 addr) {
 	for (auto &block : blockList_) {
@@ -95,6 +96,10 @@ void ImJitViewerWindow::Draw(ImConfig &cfg, ImControl &control) {
 				cb.addr = meta.addr;
 				cb.sizeInBytes = meta.sizeInBytes;
 				cb.blockNum = i;
+#ifdef IR_PROFILING
+				JitBlockProfileStats stats = blockCacheDebug->GetBlockProfileStats(i);
+				cb.profileStats = stats;
+#endif
 				blockList_.push_back(cb);
 			}
 
@@ -104,13 +109,14 @@ void ImJitViewerWindow::Draw(ImConfig &cfg, ImControl &control) {
 				}
 				const ImGuiTableColumnSortSpecs *spec = &sortSpecs_->Specs[0];
 				int delta = 0;
-				if (spec->ColumnUserID == 0) {
-					delta = (int)a.blockNum - (int)b.blockNum;
-				} else
-				if (spec->ColumnUserID == 1) {
-					delta = (int)a.addr - (int)b.addr;
-				} else if (spec->ColumnUserID == 2) {
-					delta = a.sizeInBytes - b.sizeInBytes;
+				switch (spec->ColumnUserID) {
+				case 0: delta = (int)a.blockNum - (int)b.blockNum; break;
+				case 1: delta = (int)a.addr - (int)b.addr; break;
+				case 2: delta = a.sizeInBytes - b.sizeInBytes; break;
+#ifdef IR_PROFILING
+				case 3: delta = a.profileStats.executions - b.profileStats.executions; break;
+				case 4: delta = a.profileStats.totalNanos - b.profileStats.totalNanos; break;
+#endif
 				}
 				if (delta == 0) {
 					return a.addr < b.addr;
@@ -143,11 +149,21 @@ void ImJitViewerWindow::Draw(ImConfig &cfg, ImControl &control) {
 
 		// For separate scrolling.
 		ImGui::BeginChild("LeftPane", ImVec2(0, 0), true, ImGuiWindowFlags_HorizontalScrollbar);
-
-		if (ImGui::BeginTable("blocks", 3, ImGuiTableFlags_Sortable | ImGuiTableFlags_Resizable)) {
+#ifdef IR_PROFILING
+		const int numColumns = (core == CPUCore::IR_INTERPRETER) ? 5 : 3;
+#else
+		const int numColumns = 3;
+#endif
+		if (ImGui::BeginTable("blocks", numColumns, ImGuiTableFlags_Sortable | ImGuiTableFlags_Resizable)) {
 			ImGui::TableSetupColumn("Num", 0, 0, 0);
 			ImGui::TableSetupColumn("Addr", 0, 0, 1);
 			ImGui::TableSetupColumn("Size", 0, 0, 2);
+#ifdef IR_PROFILING
+			if (numColumns == 5) {
+				ImGui::TableSetupColumn("Exec", 0, 0, 3);
+				ImGui::TableSetupColumn("Ns", 0, 0, 4);
+			}
+#endif
 			ImGui::TableHeadersRow();
 			sortSpecs_ = ImGui::TableGetSortSpecs();
 
@@ -176,6 +192,16 @@ void ImJitViewerWindow::Draw(ImConfig &cfg, ImControl &control) {
 
 					ImGui::TableNextColumn();
 					ImGui::Text("%d", block.sizeInBytes);
+
+#ifdef IR_PROFILING
+					if (numColumns == 5) {
+						ImGui::TableNextColumn();
+						ImGui::Text("%d", block.profileStats.executions);
+
+						ImGui::TableNextColumn();
+						ImGui::Text("%0.3f ms", (double)block.profileStats.totalNanos * 0.000001);
+					}
+#endif
 				}
 			}
 
diff --git a/UI/ImDebugger/ImJitViewer.h b/UI/ImDebugger/ImJitViewer.h
index 87a379c21a..d756e1a40d 100644
--- a/UI/ImDebugger/ImJitViewer.h
+++ b/UI/ImDebugger/ImJitViewer.h
@@ -6,6 +6,7 @@
 
 #include "Core/Debugger/DebugInterface.h"
 #include "Core/MIPS/JitCommon/JitBlockCache.h"
+#include "Core/MIPS/IR/IRJit.h"
 
 struct ImConfig;
 struct ImControl;
@@ -24,6 +25,9 @@ private:
 		u32 addr;
 		int sizeInBytes;
 		int blockNum;
+#ifdef IR_PROFILING
+		JitBlockProfileStats profileStats;
+#endif
 	};
 	std::vector<CachedBlock> blockList_;
 	int curBlockNum_ = -1;