diff --git a/Common/Data/Text/StringWriter.h b/Common/Data/Text/StringWriter.h
index 42d29998ce..0486323a50 100644
--- a/Common/Data/Text/StringWriter.h
+++ b/Common/Data/Text/StringWriter.h
@@ -22,6 +22,9 @@ public:
 	std::string_view as_view() const {
 		return std::string_view(start_, p_ - start_);
 	}
+	std::string as_string() const {
+		return std::string(start_, p_ - start_);
+	}
 	const char *begin() const {
 		return start_;
 	}
diff --git a/GPU/Common/ShaderId.cpp b/GPU/Common/ShaderId.cpp
index 56178ae393..db3c5d8474 100644
--- a/GPU/Common/ShaderId.cpp
+++ b/GPU/Common/ShaderId.cpp
@@ -4,6 +4,7 @@
 
 #include "Common/GPU/thin3d.h"
 #include "Common/StringUtils.h"
+#include "Common/Data/Text/StringWriter.h"
 #include "Core/Config.h"
 
 #include "GPU/ge_constants.h"
@@ -15,53 +16,57 @@
 #include "GPU/Common/DrawEngineCommon.h"  // Just for ClipInfoFlags
 
 std::string VertexShaderDesc(const VShaderID &id) {
-	std::stringstream desc;
-	desc << StringFromFormat("%08x:%08x ", id.d[1], id.d[0]);
-	if (id.Bit(VS_BIT_IS_THROUGH)) desc << "THR ";
-	if (id.Bit(VS_BIT_USE_HW_TRANSFORM)) desc << "HWX ";
-	if (id.Bit(VS_BIT_HAS_COLOR)) desc << "C ";
-	if (id.Bit(VS_BIT_HAS_TEXCOORD)) desc << "T ";
-	if (id.Bit(VS_BIT_HAS_NORMAL)) desc << "N ";
-	if (id.Bit(VS_BIT_LMODE)) desc << "LM ";
-	if (id.Bit(VS_BIT_NORM_REVERSE)) desc << "RevN ";
+	char buffer[512];
+	StringWriter desc(buffer, sizeof(buffer));
+
+	desc.F("%08x:%08x ", id.d[1], id.d[0]);
+	if (id.Bit(VS_BIT_IS_THROUGH)) desc.C("THR ");
+	if (id.Bit(VS_BIT_USE_HW_TRANSFORM)) desc.C("HWX "); else desc.C("SWX ");
+	if (id.Bit(VS_BIT_HAS_NORMAL)) desc.C("N ");
+	if (id.Bit(VS_BIT_HAS_TEXCOORD)) desc.C("T ");
+	if (id.Bit(VS_BIT_HAS_COLOR)) desc.C("C ");
+	if (id.Bit(VS_BIT_LMODE)) desc.C("LM ");
+	if (id.Bit(VS_BIT_NORM_REVERSE)) desc.C("RevN ");
+	if (id.Bit(VS_BIT_FLATSHADE)) desc.C("Flat ");
+	if (id.Bits(VS_BIT_MATERIAL_UPDATE, 3)) desc.C("MatUp:").F("%d", id.Bits(VS_BIT_MATERIAL_UPDATE, 3)).C(" ");
+
 	int uvgMode = id.Bits(VS_BIT_UVGEN_MODE, 2);
+	static constexpr std::string_view uvgModes[4] = {"UV ", "UVMtx ", "UVEnv ", "UVUnk "};
+	if (uvgMode) desc.W(uvgModes[uvgMode]);
 	if (uvgMode == GE_TEXMAP_TEXTURE_MATRIX) {
 		int uvprojMode = id.Bits(VS_BIT_UVPROJ_MODE, 2);
-		const char *uvprojModes[4] = { "TexProjPos ", "TexProjUV ", "TexProjNNrm ", "TexProjNrm " };
-		desc << uvprojModes[uvprojMode];
+		static constexpr std::string_view uvprojModes[4] = { "TexProjPos ", "TexProjUV ", "TexProjNNrm ", "TexProjNrm " };
+		desc.W(uvprojModes[uvprojMode]);
 	}
-	static constexpr std::array<const char*, 4> uvgModes = { "UV ", "UVMtx ", "UVEnv ", "UVUnk " };
+
+	if (id.Bit(VS_BIT_ENABLE_BONES)) desc.F("Bones:%d ", id.Bits(VS_BIT_BONES, 3) + 1);
+	if (id.Bits(VS_BIT_WEIGHT_FMTSCALE, 2)) desc.F("WScale:%d ", id.Bits(VS_BIT_WEIGHT_FMTSCALE, 2));
+
 	int ls0 = id.Bits(VS_BIT_LS0, 2);
 	int ls1 = id.Bits(VS_BIT_LS1, 2);
 
-	if (uvgMode) desc << uvgModes[uvgMode];
-	if (id.Bit(VS_BIT_ENABLE_BONES)) desc << "Bones:" << (id.Bits(VS_BIT_BONES, 3) + 1) << " ";
+	if (id.Bit(VS_BIT_FS_MINMAX_DISCARD)) desc.C("FSMinMax ");
+	if (id.Bit(VS_BIT_FS_DEPTH_CLAMP)) desc.C("FSDepthClamp ");
 
 	// Lights
 	if (id.Bit(VS_BIT_LIGHTING_ENABLE)) {
-		desc << "Light: ";
+		desc.C("Light: ");
 	}
 	if (id.Bit(VS_BIT_LIGHT_UBERSHADER)) {
-		desc << "LightUberShader ";
+		desc.C("LightUberShader ");
 	}
 	for (int i = 0; i < 4; i++) {
 		bool enabled = id.Bit(VS_BIT_LIGHT0_ENABLE + i) && id.Bit(VS_BIT_LIGHTING_ENABLE);
 		if (enabled || (uvgMode == GE_TEXMAP_ENVIRONMENT_MAP && (ls0 == i || ls1 == i))) {
-			desc << i << ": ";
-			desc << "c:" << id.Bits(VS_BIT_LIGHT0_COMP + 4 * i, 2) << " t:" << id.Bits(VS_BIT_LIGHT0_TYPE + 4 * i, 2) << " ";
+			desc.F("%d: ", i);
+			desc.F("c:%d t:%d ", id.Bits(VS_BIT_LIGHT0_COMP + 4 * i, 2), id.Bits(VS_BIT_LIGHT0_TYPE + 4 * i, 2));
 		}
 	}
-	if (id.Bits(VS_BIT_MATERIAL_UPDATE, 3)) desc << "MatUp:" << id.Bits(VS_BIT_MATERIAL_UPDATE, 3) << " ";
-	if (id.Bits(VS_BIT_WEIGHT_FMTSCALE, 2)) desc << "WScale " << id.Bits(VS_BIT_WEIGHT_FMTSCALE, 2) << " ";
-	if (id.Bit(VS_BIT_FLATSHADE)) desc << "Flat ";
 
-	if (id.Bit(VS_BIT_VERTEX_RANGE_CULLING)) desc << "RangeCull ";
+	if (id.Bit(VS_BIT_SIMPLE_STEREO)) desc.C("SimpleStereo ");
+	if (id.Bit(VS_BIT_VERTEX_RANGE_CULLING)) desc.C("RangeCull ");
 
-	if (id.Bit(VS_BIT_SIMPLE_STEREO)) desc << "SimpleStereo ";
-	if (id.Bit(VS_BIT_FS_MINMAX_DISCARD)) desc << "FSMinMax ";
-	if (id.Bit(VS_BIT_FS_DEPTH_CLAMP)) desc << "FSDepthClamp ";
-
-	return desc.str();
+	return desc.as_string();
 }
 
 void ComputeVertexShaderID(VShaderID *id_out, u32 vertType, bool useHWTransform, bool weightsAsFloat, bool useSkinInDecode, ClipInfoFlags clipInfoFlags) {
diff --git a/GPU/Common/ShaderId.h b/GPU/Common/ShaderId.h
index 26807c3837..b853a4dc6b 100644
--- a/GPU/Common/ShaderId.h
+++ b/GPU/Common/ShaderId.h
@@ -24,52 +24,47 @@ inline bool needFragmentDepthClamp() {
 // VS_BIT_LIGHT_UBERSHADER indicates that some groups of these will be
 // sent to the shader and processed there. This cuts down the number of shaders ("ubershader approach").
 enum VShaderBit : uint8_t {
-	VS_BIT_LMODE = 0,
-	VS_BIT_IS_THROUGH = 1,
-	// bit 2 is free.
-	VS_BIT_HAS_COLOR = 3,
-	// bit 4 is free.
-	VS_BIT_VERTEX_RANGE_CULLING = 5,
-	VS_BIT_SIMPLE_STEREO = 6,
-	// bit 7 is free,
-	VS_BIT_USE_HW_TRANSFORM = 8,
-	VS_BIT_HAS_NORMAL = 9,  // conditioned on hw transform
-	VS_BIT_NORM_REVERSE = 10,
-	VS_BIT_HAS_TEXCOORD = 11,
-	// 4 bits free: 12-15
-	VS_BIT_UVGEN_MODE = 16,
-	VS_BIT_UVPROJ_MODE = 18,  // 2, can overlap with LS0
-	VS_BIT_LS0 = 18,  // 2
-	VS_BIT_LS1 = 20,  // 2
-	VS_BIT_BONES = 22,  // 3 should be enough, not 8
-	// 25 - 29 are free.
-	VS_BIT_ENABLE_BONES = 30,
+	VS_BIT_IS_THROUGH = 0,
+	VS_BIT_USE_HW_TRANSFORM = 1,
+	VS_BIT_HAS_NORMAL = 2,  // conditioned on hw transform
+	VS_BIT_HAS_TEXCOORD = 3,
+	VS_BIT_HAS_COLOR = 4,
+	VS_BIT_LMODE = 5,
+	VS_BIT_NORM_REVERSE = 6,
+	VS_BIT_FLATSHADE = 7,
+	VS_BIT_MATERIAL_UPDATE = 8,  // 3 bits
+	// Free bit: 11
+	VS_BIT_UVGEN_MODE = 12,  // 2 bits
+	VS_BIT_UVPROJ_MODE = 14,  // 2 bits
+	VS_BIT_ENABLE_BONES = 16,
+	VS_BIT_WEIGHT_FMTSCALE = 17,  // only two bits
+	VS_BIT_BONES = 19,  // 3 should be enough to represent 1-8 bones.
+	VS_BIT_FS_MINMAX_DISCARD = 22, // Do min/max and/or depth clamp in the fragment shader. It just means we need to forward Z and W to the fragment shader.
+	VS_BIT_FS_DEPTH_CLAMP = 23, // Do depth clamp in the fragment shader.
+	VS_BIT_LIGHTING_ENABLE = 24,
+	VS_BIT_LS0 = 25,  // 2 bits
+	VS_BIT_LS1 = 27,  // 2 bits
 
 	// If this is set along with LIGHTING_ENABLE, all other lighting bits below
 	// are passed to the shader directly instead.
-	VS_BIT_LIGHT_UBERSHADER = 31,
+	VS_BIT_LIGHT_UBERSHADER = 29,
 
-	VS_BIT_LIGHT0_COMP = 32,  // 2 bits
-	VS_BIT_LIGHT0_TYPE = 34,  // 2 bits
-	VS_BIT_LIGHT1_COMP = 36,  // 2 bits
-	VS_BIT_LIGHT1_TYPE = 38,  // 2 bits
-	VS_BIT_LIGHT2_COMP = 40,  // 2 bits
-	VS_BIT_LIGHT2_TYPE = 42,  // 2 bits
-	VS_BIT_LIGHT3_COMP = 44,  // 2 bits
-	VS_BIT_LIGHT3_TYPE = 46,  // 2 bits
-	VS_BIT_MATERIAL_UPDATE = 48,  // 3 bits
-	// Bit 51 is free.
-	VS_BIT_LIGHT0_ENABLE = 52,
-	VS_BIT_LIGHT1_ENABLE = 53,
-	VS_BIT_LIGHT2_ENABLE = 54,
-	VS_BIT_LIGHT3_ENABLE = 55,
-	VS_BIT_LIGHTING_ENABLE = 56,
-	VS_BIT_WEIGHT_FMTSCALE = 57,  // only two bits
-	// 59 - 61 are free.
-	VS_BIT_FS_MINMAX_DISCARD = 59, // Do min/max and/or depth clamp in the fragment shader. It just means we need to forward Z and W to the fragment shader.
-	VS_BIT_FS_DEPTH_CLAMP = 60, // Do depth clamp in the fragment shader.
-	VS_BIT_FLATSHADE = 62, // 1 bit
-	// Bit 63 is free.
+	VS_BIT_LIGHT0_COMP = 30,  // 2 bits
+	VS_BIT_LIGHT0_TYPE = 32,  // 2 bits
+	VS_BIT_LIGHT1_COMP = 34,  // 2 bits
+	VS_BIT_LIGHT1_TYPE = 36,  // 2 bits
+	VS_BIT_LIGHT2_COMP = 38,  // 2 bits
+	VS_BIT_LIGHT2_TYPE = 40,  // 2 bits
+	VS_BIT_LIGHT3_COMP = 42,  // 2 bits
+	VS_BIT_LIGHT3_TYPE = 44,  // 2 bits
+	VS_BIT_LIGHT0_ENABLE = 46,
+	VS_BIT_LIGHT1_ENABLE = 47,
+	VS_BIT_LIGHT2_ENABLE = 48,
+	VS_BIT_LIGHT3_ENABLE = 49,
+
+	VS_BIT_VERTEX_RANGE_CULLING = 50,
+	VS_BIT_SIMPLE_STEREO = 51,
+	// bits 52-63 are free.
 };
 
 static inline VShaderBit operator +(VShaderBit bit, int i) {
diff --git a/GPU/GLES/ShaderManagerGLES.cpp b/GPU/GLES/ShaderManagerGLES.cpp
index 34cf666006..a086a0f3e8 100644
--- a/GPU/GLES/ShaderManagerGLES.cpp
+++ b/GPU/GLES/ShaderManagerGLES.cpp
@@ -900,7 +900,7 @@ enum class CacheDetectFlags {
 };
 
 #define CACHE_HEADER_MAGIC 0x83277592
-#define CACHE_VERSION 40
+#define CACHE_VERSION 41
 
 struct CacheHeader {
 	uint32_t magic;
diff --git a/GPU/Vulkan/ShaderManagerVulkan.cpp b/GPU/Vulkan/ShaderManagerVulkan.cpp
index acf919f0b9..c2b8c20ca5 100644
--- a/GPU/Vulkan/ShaderManagerVulkan.cpp
+++ b/GPU/Vulkan/ShaderManagerVulkan.cpp
@@ -408,7 +408,7 @@ enum class VulkanCacheDetectFlags {
 };
 
 #define CACHE_HEADER_MAGIC 0xff51f420 
-#define CACHE_VERSION 56
+#define CACHE_VERSION 57
 
 struct VulkanCacheHeader {
 	uint32_t magic;