diff --git a/Common/GPU/D3D11/thin3d_d3d11.cpp b/Common/GPU/D3D11/thin3d_d3d11.cpp
index 1dd6a13844..88de7fa0e5 100644
--- a/Common/GPU/D3D11/thin3d_d3d11.cpp
+++ b/Common/GPU/D3D11/thin3d_d3d11.cpp
@@ -1820,7 +1820,7 @@ bool D3D11DrawContext::CopyFramebufferToMemory(Framebuffer *src, Aspect channelB
 	const uint8_t *srcWithOffset = (const uint8_t *)map.pData + srcByteOffset;
 	switch ((Aspect)channelBits) {
 	case Aspect::COLOR_BIT:
-		// Pixel size always 4 here because we always request BGRA8888.
+		// Pixel size always 4 here because we always request RGBA8888.
 		ConvertFromRGBA8888((uint8_t *)pixels, srcWithOffset, pixelStride, map.RowPitch / sizeof(uint32_t), bw, bh, destFormat);
 		break;
 	case Aspect::DEPTH_BIT:
diff --git a/GPU/Common/FragmentShaderGenerator.cpp b/GPU/Common/FragmentShaderGenerator.cpp
index f317d7d8fa..e52a458607 100644
--- a/GPU/Common/FragmentShaderGenerator.cpp
+++ b/GPU/Common/FragmentShaderGenerator.cpp
@@ -143,7 +143,6 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu
 		*errorString = "depal requires bitwise ops";
 		return false;
 	}
-	bool bgraTexture = id.Bit(FS_BIT_BGRA_TEXTURE);
 	bool colorWriteMask = id.Bit(FS_BIT_COLOR_WRITEMASK) && compat.bitwiseOps;
 
 	GEComparison alphaTestFunc = (GEComparison)id.Bits(FS_BIT_ALPHA_TEST_FUNC, 3);
@@ -601,15 +600,15 @@ bool GenerateFragmentShader(const FShaderID &id, char *buffer, const ShaderLangu
 				if (compat.shaderLanguage == HLSL_D3D11) {
 					if (texture3D) {
 						if (doTextureProjection) {
-							WRITE(p, "  vec4 t = tex.Sample(texSamp, vec3(v_texcoord.xy / v_texcoord.z, u_mipBias))%s;\n", bgraTexture ? ".bgra" : "");
+							WRITE(p, "  vec4 t = tex.Sample(texSamp, vec3(v_texcoord.xy / v_texcoord.z, u_mipBias));\n");
 						} else {
-							WRITE(p, "  vec4 t = tex.Sample(texSamp, vec3(%s.xy, u_mipBias))%s;\n", texcoord, bgraTexture ? ".bgra" : "");
+							WRITE(p, "  vec4 t = tex.Sample(texSamp, vec3(%s.xy, u_mipBias));\n", texcoord);
 						}
 					} else {
 						if (doTextureProjection) {
-							WRITE(p, "  vec4 t = tex.Sample(texSamp, v_texcoord.xy / v_texcoord.z)%s;\n", bgraTexture ? ".bgra" : "");
+							WRITE(p, "  vec4 t = tex.Sample(texSamp, v_texcoord.xy / v_texcoord.z);\n");
 						} else {
-							WRITE(p, "  vec4 t = tex.Sample(texSamp, %s.xy)%s;\n", texcoord, bgraTexture ? ".bgra" : "");
+							WRITE(p, "  vec4 t = tex.Sample(texSamp, %s.xy);\n", texcoord);
 						}
 					}
 				} else {
diff --git a/GPU/Common/FramebufferManagerCommon.cpp b/GPU/Common/FramebufferManagerCommon.cpp
index 7d4644b279..735e4d7a98 100644
--- a/GPU/Common/FramebufferManagerCommon.cpp
+++ b/GPU/Common/FramebufferManagerCommon.cpp
@@ -1384,7 +1384,7 @@ Draw::Texture *FramebufferManagerCommon::MakePixelTexture(const u8 *srcPixels, G
 		XXH3_freeState(hashState);
 	}
 
-	Draw::DataFormat texFormat = preferredPixelsFormat_;
+	Draw::DataFormat texFormat = Draw::DataFormat::R8G8B8A8_UNORM;
 
 	if (srcPixelFormat == GE_FORMAT_DEPTH16) {
 		if ((draw_->GetDataFormatSupport(Draw::DataFormat::R16_UNORM) & Draw::FMT_TEXTURE) != 0) {
diff --git a/GPU/Common/FramebufferManagerCommon.h b/GPU/Common/FramebufferManagerCommon.h
index 6bfb46af8f..b8be6f3d8d 100644
--- a/GPU/Common/FramebufferManagerCommon.h
+++ b/GPU/Common/FramebufferManagerCommon.h
@@ -625,8 +625,6 @@ protected:
 	int bloomHack_ = 0;
 	bool updatePostShaders_ = false;
 
-	Draw::DataFormat preferredPixelsFormat_ = Draw::DataFormat::R8G8B8A8_UNORM;
-
 	struct TempFBOInfo {
 		Draw::Framebuffer *fbo;
 		int last_frame_used;
diff --git a/GPU/Common/GPUStateUtils.cpp b/GPU/Common/GPUStateUtils.cpp
index 08c398e733..a0076a5291 100644
--- a/GPU/Common/GPUStateUtils.cpp
+++ b/GPU/Common/GPUStateUtils.cpp
@@ -42,6 +42,7 @@ bool IsStencilTestOutputDisabled() {
 	return true;
 }
 
+// This is used to figure out when we can replace discard with alpha blending.
 bool NeedsTestDiscard() {
 	// We assume this is called only when enabled and not trivially true (may also be for color testing.)
 	if (gstate.isStencilTestEnabled() && (gstate.pmska & 0xFF) != 0xFF)
diff --git a/GPU/Common/ShaderId.cpp b/GPU/Common/ShaderId.cpp
index a67f59846c..56178ae393 100644
--- a/GPU/Common/ShaderId.cpp
+++ b/GPU/Common/ShaderId.cpp
@@ -175,7 +175,6 @@ std::string FragmentShaderDesc(const FShaderID &id) {
 	if (id.Bit(FS_BIT_DO_TEXTURE_PROJ)) desc << "TexProj ";
 	if (id.Bit(FS_BIT_LMODE)) desc << "LM ";
 	if (id.Bit(FS_BIT_FLATSHADE)) desc << "Flat ";
-	if (id.Bit(FS_BIT_BGRA_TEXTURE)) desc << "BGRA ";
 	if (id.Bit(FS_BIT_DEPTH_TEST_NEVER)) desc << "DepthNever ";
 	if (id.Bit(FS_BIT_COLOR_WRITEMASK)) desc << "WriteMask ";
 	if (id.Bit(FS_BIT_SHADER_TEX_CLAMP)) {
@@ -328,7 +327,6 @@ void ComputeFragmentShaderID(FShaderID *id_out, const ComputedPipelineState &pip
 				id.SetBit(FS_BIT_CLAMP_S, gstate.isTexCoordClampedS());
 				id.SetBit(FS_BIT_CLAMP_T, gstate.isTexCoordClampedT());
 			}
-			id.SetBit(FS_BIT_BGRA_TEXTURE, gstate_c.bgraTexture);
 			id.SetBits(FS_BIT_SHADER_DEPAL_MODE, 2, (int)shaderDepalMode);
 			id.SetBits(FS_BIT_SHADER_DEPAL_FORMAT, 3, (int)shaderDepalFormat);
 			id.SetBit(FS_BIT_3D_TEXTURE, gstate_c.curTextureIs3D);
@@ -348,7 +346,7 @@ void ComputeFragmentShaderID(FShaderID *id_out, const ComputedPipelineState &pip
 			id.SetBit(FS_BIT_COLOR_TEST);
 			id.SetBits(FS_BIT_COLOR_TEST_FUNC, 2, gstate.getColorTestFunction());
 			id.SetBit(FS_BIT_COLOR_AGAINST_ZERO, IsColorTestAgainstZero());
-			// This is alos set in enableAlphaTest - color test is uncommon, but we can skip discard the same way.
+			// This is also set in enableAlphaTest - color test is uncommon, but we can skip discard the same way.
 			id.SetBit(FS_BIT_TEST_DISCARD_TO_ZERO, !NeedsTestDiscard());
 		}
 
diff --git a/GPU/Common/ShaderId.h b/GPU/Common/ShaderId.h
index e8815e1bf2..26807c3837 100644
--- a/GPU/Common/ShaderId.h
+++ b/GPU/Common/ShaderId.h
@@ -76,7 +76,7 @@ static inline VShaderBit operator +(VShaderBit bit, int i) {
 	return VShaderBit((int)bit + i);
 }
 
-// Local
+// TODO: See what we can free up. We're out of bits!
 enum FShaderBit : uint8_t {
 	FS_BIT_CLEARMODE = 0,
 	FS_BIT_DO_TEXTURE = 1,
@@ -102,7 +102,7 @@ enum FShaderBit : uint8_t {
 	FS_BIT_BLENDFUNC_A = 38,  // 4 bits
 	FS_BIT_BLENDFUNC_B = 42,  // 4 bits
 	FS_BIT_FLATSHADE = 46,
-	FS_BIT_BGRA_TEXTURE = 47,
+	// Free bit 47
 	FS_BIT_TEST_DISCARD_TO_ZERO = 48,
 	FS_BIT_NO_DEPTH_CANNOT_DISCARD_STENCIL = 49,
 	FS_BIT_COLOR_WRITEMASK = 50,
diff --git a/GPU/Common/TextureCacheCommon.cpp b/GPU/Common/TextureCacheCommon.cpp
index 579a25b5f4..b251c9327c 100644
--- a/GPU/Common/TextureCacheCommon.cpp
+++ b/GPU/Common/TextureCacheCommon.cpp
@@ -611,7 +611,6 @@ TexCacheEntry *TextureCacheCommon::SetTexture() {
 			gstate_c.SetTextureIsVideo(false);
 			gstate_c.SetTextureIs3D(entry->status & TexStatus::IS_3D);
 			gstate_c.SetTextureIsArray(false);
-			gstate_c.SetTextureIsBGRA(entry->status & TexStatus::BGRA);
 			gstate_c.SetTextureIsFramebuffer(false);
 			if (entry->status & TexStatus::CLUT8_INDEXED) {
 				gstate_c.SetShaderDepal(ShaderDepalMode::NORMAL, GE_FORMAT_CLUT8);
@@ -709,7 +708,6 @@ TexCacheEntry *TextureCacheCommon::SetTexture() {
 	entry->dim = dim;
 	entry->format = texFormat;
 	entry->maxLevel = maxLevel;
-	entry->status &= ~TexStatus::BGRA;
 
 	entry->bufw = bufw;
 
@@ -1218,7 +1216,6 @@ void TextureCacheCommon::SetTextureFramebuffer(const AttachCandidate &candidate)
 		gstate_c.curTextureWidth = texWidth;
 		gstate_c.curTextureHeight = texHeight;
 		gstate_c.SetTextureIsFramebuffer(true);
-		gstate_c.SetTextureIsBGRA(false);
 
 		if ((gstate_c.curTextureXOffset == 0) != (fbInfo.xOffset == 0) || (gstate_c.curTextureYOffset == 0) != (fbInfo.yOffset == 0)) {
 			gstate_c.Dirty(DIRTY_FRAGMENTSHADER_STATE);
@@ -2240,7 +2237,6 @@ void TextureCacheCommon::ApplyTexture(bool doBind, bool flatZ) {
 		gstate_c.SetTextureSolidAlpha(false);
 		gstate_c.SetTextureIs3D(false);
 		gstate_c.SetTextureIsArray(false);
-		gstate_c.SetTextureIsBGRA(false);
 	} else {
 		if (doBind) {
 			BindTexture(entry, flatZ);
@@ -2248,7 +2244,6 @@ void TextureCacheCommon::ApplyTexture(bool doBind, bool flatZ) {
 		gstate_c.SetTextureSolidAlpha((entry->status & TexStatus::ALPHA_SOLID) != 0);
 		gstate_c.SetTextureIs3D((entry->status & TexStatus::IS_3D) != 0);
 		gstate_c.SetTextureIsArray(false);
-		gstate_c.SetTextureIsBGRA((entry->status & TexStatus::BGRA) != 0);
 		if (entry->status & TexStatus::CLUT8_INDEXED) {
 			bool smoothedDepal = false;
 			u32 depthUpperBits = 0;
@@ -3125,8 +3120,5 @@ std::string TexStatusToString(TexStatus status) {
 	if (status & TexStatus::VIDEO) {
 		result += "VIDEO ";
 	}
-	if (status & TexStatus::BGRA) {
-		result += "BGRA ";
-	}
 	return result.empty() ? "None" : result;
 }
diff --git a/GPU/Common/TextureCacheCommon.h b/GPU/Common/TextureCacheCommon.h
index a0d5d914da..ecfc39424b 100644
--- a/GPU/Common/TextureCacheCommon.h
+++ b/GPU/Common/TextureCacheCommon.h
@@ -137,7 +137,7 @@ struct TextureDefinition {
 
 enum class TexStatus : u16 {
 	VIDEO = (1 << 0),
-	BGRA = (1 << 1),
+	// Free bit 1
 	ALPHA_SOLID = (1 << 2),      // Has no alpha channel, or always solid (==1.0) alpha.
 
 	MANY_CLUT_VARIANTS = (1 << 3),   // Has multiple CLUT variants.
diff --git a/GPU/D3D11/FramebufferManagerD3D11.cpp b/GPU/D3D11/FramebufferManagerD3D11.cpp
index 3e4b5c949c..5f972b2892 100644
--- a/GPU/D3D11/FramebufferManagerD3D11.cpp
+++ b/GPU/D3D11/FramebufferManagerD3D11.cpp
@@ -25,5 +25,4 @@
 FramebufferManagerD3D11::FramebufferManagerD3D11(Draw::DrawContext *draw)
 	: FramebufferManagerCommon(draw) {
 	presentation_->SetLanguage(HLSL_D3D11);
-	preferredPixelsFormat_ = Draw::DataFormat::B8G8R8A8_UNORM;
 }
diff --git a/GPU/D3D11/GPU_D3D11.cpp b/GPU/D3D11/GPU_D3D11.cpp
index 9e9b5c7097..4e9ffa2b99 100644
--- a/GPU/D3D11/GPU_D3D11.cpp
+++ b/GPU/D3D11/GPU_D3D11.cpp
@@ -78,19 +78,6 @@ GPU_D3D11::GPU_D3D11(GraphicsContext *gfxCtx, Draw::DrawContext *draw)
 
 GPU_D3D11::~GPU_D3D11() {}
 
-u32 GPU_D3D11::CheckGPUFeatures() const {
-	u32 features = GPUCommonHW::CheckGPUFeatures();
-
-	uint32_t fmt4444 = draw_->GetDataFormatSupport(Draw::DataFormat::A4R4G4B4_UNORM_PACK16);
-	uint32_t fmt1555 = draw_->GetDataFormatSupport(Draw::DataFormat::A1R5G5B5_UNORM_PACK16);
-	uint32_t fmt565 = draw_->GetDataFormatSupport(Draw::DataFormat::R5G6B5_UNORM_PACK16);
-	if ((fmt4444 & Draw::FMT_TEXTURE) && (fmt565 & Draw::FMT_TEXTURE) && (fmt1555 & Draw::FMT_TEXTURE)) {
-		features |= GPU_USE_16BIT_FORMATS;
-	}
-
-	return CheckGPUFeaturesLate(features);
-}
-
 void GPU_D3D11::DeviceLost() {
 	draw_->Invalidate(InvalidationFlags::CACHED_RENDER_STATE);
 	// Simply drop all caches and textures.
diff --git a/GPU/D3D11/GPU_D3D11.h b/GPU/D3D11/GPU_D3D11.h
index 97448f7c9d..5486db5a13 100644
--- a/GPU/D3D11/GPU_D3D11.h
+++ b/GPU/D3D11/GPU_D3D11.h
@@ -37,8 +37,6 @@ public:
 	GPU_D3D11(GraphicsContext *gfxCtx, Draw::DrawContext *draw);
 	~GPU_D3D11();
 
-	u32 CheckGPUFeatures() const override;
-
 	void GetStats(StringWriter &w) override;
 	void DeviceLost() override;  // Destroy all device objects.
 	void DeviceRestore(Draw::DrawContext *draw) override;
diff --git a/GPU/D3D11/TextureCacheD3D11.cpp b/GPU/D3D11/TextureCacheD3D11.cpp
index fbf50a8c95..b44932e034 100644
--- a/GPU/D3D11/TextureCacheD3D11.cpp
+++ b/GPU/D3D11/TextureCacheD3D11.cpp
@@ -57,7 +57,7 @@ static Draw::DataFormat FromD3D11Format(u32 fmt) {
 	case DXGI_FORMAT_B5G5R5A1_UNORM: return Draw::DataFormat::A1R5G5B5_UNORM_PACK16;
 	case DXGI_FORMAT_B5G6R5_UNORM: return Draw::DataFormat::R5G6B5_UNORM_PACK16;
 	case DXGI_FORMAT_R8_UNORM: return Draw::DataFormat::R8_UNORM;
-	case DXGI_FORMAT_B8G8R8A8_UNORM: default: return Draw::DataFormat::R8G8B8A8_UNORM;
+	case DXGI_FORMAT_R8G8B8A8_UNORM: default: return Draw::DataFormat::R8G8B8A8_UNORM;
 	}
 }
 
@@ -69,7 +69,7 @@ static DXGI_FORMAT ToDXGIFormat(Draw::DataFormat fmt) {
 	case Draw::DataFormat::BC4_UNORM_BLOCK: return DXGI_FORMAT_BC4_UNORM;
 	case Draw::DataFormat::BC5_UNORM_BLOCK: return DXGI_FORMAT_BC5_UNORM;
 	case Draw::DataFormat::BC7_UNORM_BLOCK: return DXGI_FORMAT_BC7_UNORM;
-	case Draw::DataFormat::R8G8B8A8_UNORM: return DXGI_FORMAT_B8G8R8A8_UNORM;
+	case Draw::DataFormat::R8G8B8A8_UNORM: return DXGI_FORMAT_R8G8B8A8_UNORM;
 	default: _dbg_assert_(false); return DXGI_FORMAT_UNKNOWN;
 	}
 }
@@ -254,7 +254,7 @@ void TextureCacheD3D11::BuildTexture(TexCacheEntry *const entry) {
 	if (plan.doReplace) {
 		dstFmt = ToDXGIFormat(plan.replaced->Format());
 	} else if (plan.scaleFactor > 1 || plan.saveTexture) {
-		dstFmt = DXGI_FORMAT_B8G8R8A8_UNORM;
+		dstFmt = DXGI_FORMAT_R8G8B8A8_UNORM;
 	} else if (plan.decodeToClut8) {
 		dstFmt = DXGI_FORMAT_R8_UNORM;
 	}
@@ -303,7 +303,7 @@ void TextureCacheD3D11::BuildTexture(TexCacheEntry *const entry) {
 			if (plan.scaleFactor > 1) {
 				bpp = 4;
 			} else {
-				bpp = dstFmt == DXGI_FORMAT_B8G8R8A8_UNORM ? 4 : 2;
+				bpp = dstFmt == DXGI_FORMAT_R8G8B8A8_UNORM ? 4 : 2;
 			}
 			stride = std::max(mipWidth * bpp, 16);
 			dataSize = stride * mipHeight;
@@ -404,12 +404,6 @@ void TextureCacheD3D11::BuildTexture(TexCacheEntry *const entry) {
 
 	if (plan.doReplace) {
 		entry->SetAlphaStatus(plan.replaced->AlphaStatus());
-
-		if (!Draw::DataFormatIsBlockCompressed(plan.replaced->Format(), nullptr)) {
-			entry->status |= TexStatus::BGRA;
-		}
-	} else {
-		entry->status |= TexStatus::BGRA;
 	}
 }
 
@@ -422,15 +416,15 @@ DXGI_FORMAT GetClutDestFormatD3D11(GEPaletteFormat format) {
 	case GE_CMODE_16BIT_BGR5650:
 		return DXGI_FORMAT_B5G6R5_UNORM;
 	case GE_CMODE_32BIT_ABGR8888:
-		return DXGI_FORMAT_B8G8R8A8_UNORM;
+		return DXGI_FORMAT_R8G8B8A8_UNORM;
 	}
 	// Should never be here !
-	return DXGI_FORMAT_B8G8R8A8_UNORM;
+	return DXGI_FORMAT_R8G8B8A8_UNORM;
 }
 
 DXGI_FORMAT TextureCacheD3D11::GetDestFormat(GETextureFormat format, GEPaletteFormat clutFormat) const {
 	if (!gstate_c.Use(GPU_USE_16BIT_FORMATS)) {
-		return DXGI_FORMAT_B8G8R8A8_UNORM;
+		return DXGI_FORMAT_R8G8B8A8_UNORM;
 	}
 
 	switch (format) {
@@ -450,7 +444,7 @@ DXGI_FORMAT TextureCacheD3D11::GetDestFormat(GETextureFormat format, GEPaletteFo
 	case GE_TFMT_DXT3:
 	case GE_TFMT_DXT5:
 	default:
-		return DXGI_FORMAT_B8G8R8A8_UNORM;
+		return DXGI_FORMAT_R8G8B8A8_UNORM;
 	}
 }
 
@@ -475,7 +469,7 @@ bool TextureCacheD3D11::GetCurrentTextureDebug(GPUDebugBuffer &buffer, int level
 	int height = desc.Height >> level;
 
 	switch (desc.Format) {
-	case DXGI_FORMAT_B8G8R8A8_UNORM:
+	case DXGI_FORMAT_R8G8B8A8_UNORM:
 		buffer.Allocate(width, height, GPU_DBG_FORMAT_8888);
 		break;
 
diff --git a/GPU/GPUState.h b/GPU/GPUState.h
index c30b2acfcf..a6bf015949 100644
--- a/GPU/GPUState.h
+++ b/GPU/GPUState.h
@@ -483,7 +483,7 @@ enum : u32 {
 	// Free bit: 18
 	GPU_USE_FRAMEBUFFER_ARRAYS = FLAG_BIT(19),
 	GPU_USE_FRAMEBUFFER_FETCH = FLAG_BIT(20),
-	// Free bit: 21,
+	// Free bit: 21
 	GPU_ROUND_FRAGMENT_DEPTH_TO_16BIT = FLAG_BIT(22),
 	GPU_ROUND_DEPTH_TO_16BIT = FLAG_BIT(23),  // Can be disabled either per game or if we use a real 16-bit depth buffer
 	GPU_USE_CLIP_DISTANCE = FLAG_BIT(24),
@@ -563,12 +563,6 @@ struct GPUStateCache {
 	void SetTextureIsVideo(bool isVideo) {
 		textureIsVideo = isVideo;
 	}
-	void SetTextureIsBGRA(bool isBGRA) {
-		if (bgraTexture != isBGRA) {
-			bgraTexture = isBGRA;
-			Dirty(DIRTY_FRAGMENTSHADER_STATE);
-		}
-	}
 	void SetTextureIsFramebuffer(bool isFramebuffer) {
 		if (textureIsFramebuffer != isFramebuffer) {
 			textureIsFramebuffer = isFramebuffer;
@@ -609,7 +603,6 @@ public:
 
 	int skipDrawReason;
 
-	bool bgraTexture;
 	bool needShaderTexClamp;
 	bool textureIsArray;
 	bool textureIsFramebuffer;