better default scale mode

Firmware/components/backend-esp/CMakeLists.txt

@@ -21,6 +21,7 @@ idf_component_register(
         nvs_flash
 )
 
+add_subdirectory("${CMAKE_CURRENT_LIST_DIR}/../../sdk/utils" cardboy_utils_esp)
 add_subdirectory("${CMAKE_CURRENT_LIST_DIR}/../../sdk/backend_interface" backend_interface_from_backend_esp)
 
 add_library(cardboy_backend_esp INTERFACE)

Firmware/components/backend-esp/include/cardboy/backend/esp/display.hpp

@@ -6,6 +6,7 @@
 #define CB_DISPLAY_HPP
 
 #include "cardboy/backend/esp/config.hpp"
+#include "cardboy/utils/utils.hpp"
 
 #include "driver/spi_master.h"
 // (Async memcpy removed for debugging simplification)
@@ -13,7 +14,7 @@
 #include <array>
 #include <bitset>
 #include <cassert>
-
+#include <cstdint>
 
 namespace SMD {
 static constexpr size_t kLineBytes       = DISP_WIDTH / 8;
@@ -34,7 +35,7 @@ void frame_ready();
 bool frame_transfer_in_flight(); // optional diagnostic: is a frame transfer still in flight?
 
 __attribute__((always_inline)) static void set_pixel(int x, int y, bool value) {
-    assert(x >= 0 && x < DISP_WIDTH && y >= 0 && y < DISP_HEIGHT);
+    CARDBOY_CHECK(x >= 0 && x < DISP_WIDTH && y >= 0 && y < DISP_HEIGHT);
 
     unsigned lineIdx = 2 + kLineMultiSingle * y + (x / 8);
     unsigned bitIdx  = 1 << (7 - (x % 8)) % 8;
@@ -46,6 +47,14 @@ __attribute__((always_inline)) static void set_pixel(int x, int y, bool value) {
     }
 }
 
+__attribute__((always_inline)) static void set_pixel_8(int x, int y, std::uint8_t value) {
+    CARDBOY_CHECK(x >= 0 && x < DISP_WIDTH && y >= 0 && y < DISP_HEIGHT);
+    CARDBOY_CHECK((x % 8) == 0);
+    unsigned lineIdx = 2 + kLineMultiSingle * y + (x / 8);
+    dma_buf[lineIdx] = ~value;
+}
+
+
 extern "C" void s_spi_post_cb(spi_transaction_t* trans);
 
 static inline spi_device_interface_config_t _devcfg = {

Firmware/components/backend-esp/include/cardboy/backend/esp_backend.hpp

@@ -1,5 +1,7 @@
 #pragma once
 
+#include <cardboy/sdk/display_spec.hpp>
+#include "cardboy/backend/esp/display.hpp"
 #include "cardboy/sdk/platform.hpp"
 #include "cardboy/sdk/services.hpp"
 
@@ -14,9 +16,12 @@ class EspFramebuffer final : public cardboy::sdk::FramebufferFacade<EspFramebuff
 public:
     EspFramebuffer() = default;
 
-    [[nodiscard]] int  width_impl() const;
-    [[nodiscard]] int  height_impl() const;
-    void               drawPixel_impl(int x, int y, bool on);
+    [[nodiscard]] int                   width_impl() const { return cardboy::sdk::kDisplayWidth; }
+    [[nodiscard]] int                   height_impl() const { return cardboy::sdk::kDisplayHeight; }
+    void __attribute__((always_inline)) drawPixel_impl(int x, int y, bool on) { SMD::set_pixel(x, y, on); }
+    void __attribute__((always_inline)) drawBits8_impl(int x, int y, std::uint8_t bits) {
+        SMD::set_pixel_8(x, y, bits);
+    }
     void               clear_impl(bool on);
     void               frameReady_impl();
     void               sendFrame_impl(bool clearAfterSend);

Firmware/components/backend-esp/src/esp_backend.cpp

@@ -182,16 +182,6 @@ void EspRuntime::initializeHardware() {
     FsHelper::get().mount();
 }
 
-int EspFramebuffer::width_impl() const { return cardboy::sdk::kDisplayWidth; }
-
-int EspFramebuffer::height_impl() const { return cardboy::sdk::kDisplayHeight; }
-
-void EspFramebuffer::drawPixel_impl(int x, int y, bool on) {
-    if (x < 0 || y < 0 || x >= width_impl() || y >= height_impl())
-        return;
-    SMD::set_pixel(x, y, on);
-}
-
 void EspFramebuffer::clear_impl(bool on) {
     for (int y = 0; y < height_impl(); ++y)
         for (int x = 0; x < width_impl(); ++x)

Firmware/sdk/CMakeLists.txt

@@ -5,6 +5,8 @@ set(CMAKE_CXX_STANDARD 20)
 set(CMAKE_CXX_STANDARD_REQUIRED YES)
 set(CMAKE_CXX_EXTENSIONS NO)
 
+add_subdirectory(utils)
+
 add_subdirectory(backend_interface)
 
 set(CARDBOY_SDK_BACKEND_LIBRARY "" CACHE STRING "Backend implementation library for Cardboy SDK")

Firmware/sdk/apps/gameboy/include/cardboy/apps/peanut_gb.h

@@ -577,15 +577,6 @@ union cart_rtc
  */
 struct gb_s
 {
-	/**
-	 * Return byte from ROM at given address.
-	 *
-	 * \param gb_s	emulator context
-	 * \param addr	address
-	 * \return		byte at address in ROM
-	 */
-	uint8_t (*gb_rom_read)(struct gb_s*, const uint_fast32_t addr);
-
 	/**
 	 * Return byte from cart RAM at given address.
 	 *
@@ -738,9 +729,14 @@ struct gb_s
 
 		/* Implementation defined data. Set to NULL if not required. */
 		void *priv;
+		const uint8_t* rom;
 	} direct;
 };
 
+__attribute__((always_inline)) inline uint8_t gb_rom_read(struct gb_s* s, const uint_fast32_t addr) {
+	return s->direct.rom[addr];
+}
+
 #ifndef PEANUT_GB_HEADER_ONLY
 
 #define IO_JOYP	0x00
@@ -796,17 +792,17 @@ uint8_t __gb_read(struct gb_s *gb, uint16_t addr)
 	case 0x1:
 	case 0x2:
 	case 0x3:
-		return gb->gb_rom_read(gb, addr);
+		return gb_rom_read(gb, addr);
 
 	case 0x4:
 	case 0x5:
 	case 0x6:
 	case 0x7:
 		if(gb->mbc == 1 && gb->cart_mode_select)
-			return gb->gb_rom_read(gb,
+			return gb_rom_read(gb,
 					       addr + ((gb->selected_rom_bank & 0x1F) - 1) * ROM_BANK_SIZE);
 		else
-			return gb->gb_rom_read(gb, addr + (gb->selected_rom_bank - 1) * ROM_BANK_SIZE);
+			return gb_rom_read(gb, addr + (gb->selected_rom_bank - 1) * ROM_BANK_SIZE);
 
 	case 0x8:
 	case 0x9:
@@ -3542,7 +3538,7 @@ int gb_get_save_size_s(struct gb_s *gb, size_t *ram_size)
 		/* 0,  2KiB,   8KiB,  32KiB,  128KiB,   64KiB */
 		0x00, 0x800, 0x2000, 0x8000, 0x20000, 0x10000
 	};
-	uint8_t ram_size_code = gb->gb_rom_read(gb, ram_size_location);
+	uint8_t ram_size_code = gb_rom_read(gb, ram_size_location);
 
 	/* MBC2 always has 512 half-bytes of cart RAM.
 	 * This assumes that only the lower nibble of each byte is used; the
@@ -3570,7 +3566,7 @@ uint_fast32_t gb_get_save_size(struct gb_s *gb)
 		/* 0,  2KiB,   8KiB,  32KiB,  128KiB,   64KiB */
 		0x00, 0x800, 0x2000, 0x8000, 0x20000, 0x10000
 	};
-	uint8_t ram_size_code = gb->gb_rom_read(gb, ram_size_location);
+	uint8_t ram_size_code = gb_rom_read(gb, ram_size_location);
 
 	/* MBC2 always has 512 half-bytes of cart RAM.
 	 * This assumes that only the lower nibble of each byte is used; the
@@ -3603,7 +3599,7 @@ uint8_t gb_colour_hash(struct gb_s *gb)
 	uint16_t i;
 
 	for(i = ROM_TITLE_START_ADDR; i <= ROM_TITLE_END_ADDR; i++)
-		x += gb->gb_rom_read(gb, i);
+		x += gb_rom_read(gb, i);
 
 	return x;
 }
@@ -3626,7 +3622,7 @@ void gb_reset(struct gb_s *gb)
 	if(gb->gb_bootrom_read == NULL)
 	{
 		uint8_t hdr_chk;
-		hdr_chk = gb->gb_rom_read(gb, ROM_HEADER_CHECKSUM_LOC) != 0;
+		hdr_chk = gb_rom_read(gb, ROM_HEADER_CHECKSUM_LOC) != 0;
 
 		gb->cpu_reg.a = 0x01;
 		gb->cpu_reg.f.f_bits.z = 1;
@@ -3692,11 +3688,10 @@ void gb_reset(struct gb_s *gb)
 }
 
 enum gb_init_error_e gb_init(struct gb_s *gb,
-			     uint8_t (*gb_rom_read)(struct gb_s*, const uint_fast32_t),
 			     uint8_t (*gb_cart_ram_read)(struct gb_s*, const uint_fast32_t),
 			     void (*gb_cart_ram_write)(struct gb_s*, const uint_fast32_t, const uint8_t),
 			     void (*gb_error)(struct gb_s*, const enum gb_error_e, const uint16_t),
-			     void *priv)
+			     void *priv, const uint8_t* rom_data)
 {
 	const uint16_t mbc_location = 0x0147;
 	const uint16_t bank_count_location = 0x0148;
@@ -3731,11 +3726,11 @@ enum gb_init_error_e gb_init(struct gb_s *gb,
 	 * some early homebrew ROMs supposedly may use this value. */
 	const uint8_t num_ram_banks[] = { 0, 1, 1, 4, 16, 8 };
 
-	gb->gb_rom_read = gb_rom_read;
 	gb->gb_cart_ram_read = gb_cart_ram_read;
 	gb->gb_cart_ram_write = gb_cart_ram_write;
 	gb->gb_error = gb_error;
 	gb->direct.priv = priv;
+	gb->direct.rom = rom_data;
 
 	/* Initialise serial transfer function to NULL. If the front-end does
 	 * not provide serial support, Peanut-GB will emulate no cable connected
@@ -3751,24 +3746,24 @@ enum gb_init_error_e gb_init(struct gb_s *gb,
 		uint16_t i;
 
 		for(i = 0x0134; i <= 0x014C; i++)
-			x = x - gb->gb_rom_read(gb, i) - 1;
+			x = x - gb_rom_read(gb, i) - 1;
 
-		if(x != gb->gb_rom_read(gb, ROM_HEADER_CHECKSUM_LOC))
+		if(x != gb_rom_read(gb, ROM_HEADER_CHECKSUM_LOC))
 			return GB_INIT_INVALID_CHECKSUM;
 	}
 
 	/* Check if cartridge type is supported, and set MBC type. */
 	{
-		const uint8_t mbc_value = gb->gb_rom_read(gb, mbc_location);
+		const uint8_t mbc_value = gb_rom_read(gb, mbc_location);
 
 		if(mbc_value > sizeof(cart_mbc) - 1 ||
 				(gb->mbc = cart_mbc[mbc_value]) == -1)
 			return GB_INIT_CARTRIDGE_UNSUPPORTED;
 	}
 
-	gb->num_rom_banks_mask = num_rom_banks_mask[gb->gb_rom_read(gb, bank_count_location)] - 1;
-	gb->cart_ram = cart_ram[gb->gb_rom_read(gb, mbc_location)];
-	gb->num_ram_banks = num_ram_banks[gb->gb_rom_read(gb, ram_size_location)];
+	gb->num_rom_banks_mask = num_rom_banks_mask[gb_rom_read(gb, bank_count_location)] - 1;
+	gb->cart_ram = cart_ram[gb_rom_read(gb, mbc_location)];
+	gb->num_ram_banks = num_ram_banks[gb_rom_read(gb, ram_size_location)];
 
 	/* If the ROM says that it support RAM, but has 0 RAM banks, then
 	 * disable RAM reads from the cartridge. */
@@ -3804,7 +3799,7 @@ const char* gb_get_rom_name(struct gb_s* gb, char *title_str)
 
 	for(; title_loc <= title_end; title_loc++)
 	{
-		const char title_char = gb->gb_rom_read(gb, title_loc);
+		const char title_char = gb_rom_read(gb, title_loc);
 
 		if(title_char >= ' ' && title_char <= '_')
 		{
@@ -3886,11 +3881,10 @@ void gb_set_rtc(struct gb_s *gb, const struct tm * const time)
  * \returns	0 on success or an enum that describes the error.
  */
 enum gb_init_error_e gb_init(struct gb_s *gb,
-			     uint8_t (*gb_rom_read)(struct gb_s*, const uint_fast32_t),
 			     uint8_t (*gb_cart_ram_read)(struct gb_s*, const uint_fast32_t),
 			     void (*gb_cart_ram_write)(struct gb_s*, const uint_fast32_t, const uint8_t),
 			     void (*gb_error)(struct gb_s*, const enum gb_error_e, const uint16_t),
-			     void *priv);
+			     void *priv, const uint8_t* rom_data);
 
 /**
  * Executes the emulator and runs for the duration of time equal to one frame.

Firmware/sdk/apps/gameboy/src/gameboy_app.cpp

@@ -4,7 +4,9 @@
 #include "cardboy/gfx/font16x8.hpp"
 #include "cardboy/sdk/app_framework.hpp"
 #include "cardboy/sdk/app_system.hpp"
+#include "cardboy/sdk/display_spec.hpp"
 #include "cardboy/sdk/services.hpp"
+#include "cardboy/utils/utils.hpp"
 
 #include <inttypes.h>
 
@@ -34,6 +36,70 @@
 #define GB_PERF_ONLY(...)
 #endif
 
+static constexpr uint8_t kPattern2x2[4] = {0, 2, 3, 1}; // [ (y&1)<<1 | (x&1) ]
+
+// exact predicate per your shouldPixelBeOn
+static constexpr uint8_t on_bit(uint8_t v, uint8_t threshold) {
+    if (v >= 3)
+        return 1;
+    if (v == 0)
+        return 0;
+    return static_cast<uint8_t>(v > threshold);
+}
+
+// Build 4 output bits (MSB = first pixel) from 4 packed 2-bit pixels.
+// packed4 layout: p0 | p1<<2 | p2<<4 | p3<<6
+static constexpr uint8_t makeNibble(uint8_t packed4, int xParity, int yParity) {
+    uint8_t out = 0;
+    int     xp  = xParity;
+    for (int i = 0; i < 4; ++i) {
+        const uint8_t v  = (packed4 >> (i * 2)) & 0x3;
+        const uint8_t th = kPattern2x2[(yParity << 1) | (xp & 1)];
+        out              = static_cast<uint8_t>((out << 1) | on_bit(v, th));
+        xp ^= 1; // toggle x parity each pixel
+    }
+    return out;
+}
+
+using LUT256   = std::array<uint8_t, 256>;
+using LUT2x256 = std::array<LUT256, 2>;
+using LUTFull  = std::array<LUT2x256, 2>;
+
+static constexpr LUTFull buildNibbleLUT() {
+    LUTFull L{};
+    for (int yp = 0; yp < 2; ++yp)
+        for (int xp = 0; xp < 2; ++xp)
+            for (int p = 0; p < 256; ++p)
+                L[yp][xp][p] = makeNibble(static_cast<uint8_t>(p), xp, yp);
+    return L;
+}
+
+inline constexpr LUTFull kNibbleLUT = buildNibbleLUT();
+
+static constexpr LUTFull buildFullHeightWideByteLUT() {
+    LUTFull L{};
+    for (int yp = 0; yp < 2; ++yp)
+        for (int xp = 0; xp < 2; ++xp)
+            for (int p = 0; p < 256; ++p) {
+                const uint8_t p0 = static_cast<uint8_t>(p & 0x03);
+                const uint8_t p1 = static_cast<uint8_t>((p >> 2) & 0x03);
+                const uint8_t p2 = static_cast<uint8_t>((p >> 4) & 0x03);
+                const uint8_t p3 = static_cast<uint8_t>((p >> 6) & 0x03);
+
+                const uint8_t p4a = static_cast<uint8_t>(p0 | (p0 << 2) | (p1 << 4) | (p1 << 6));
+                const uint8_t p4b = static_cast<uint8_t>(p2 | (p2 << 2) | (p3 << 4) | (p3 << 6));
+
+                const uint8_t n0 = makeNibble(p4a, xp, yp);
+                const uint8_t n1 = makeNibble(p4b, xp, yp);
+
+                L[yp][xp][p] = static_cast<uint8_t>((n0 << 4) | (n1 & 0x0F));
+            }
+    return L;
+}
+
+inline constexpr LUTFull kFullHeightWideByteLUT = buildFullHeightWideByteLUT();
+
+
 namespace apps {
 namespace {
 
@@ -102,49 +168,6 @@ void appendEmbeddedRoms(std::vector<RomEntry>& out) {
     }
 }
 
-int measureVerticalText(std::string_view text, int scale = 1, int letterSpacing = 1) {
-    if (text.empty())
-        return 0;
-    const int advance = (font16x8::kGlyphWidth + letterSpacing) * scale;
-    return static_cast<int>(text.size()) * advance - letterSpacing * scale;
-}
-
-void drawGlyphRotated(Framebuffer& fb, int x, int y, char ch, bool clockwise, int scale = 1, bool on = true) {
-    const auto& rows = font16x8::glyphBitmap(ch);
-    for (int row = 0; row < font16x8::kGlyphHeight; ++row) {
-        const uint8_t rowBits = rows[row];
-        for (int col = 0; col < font16x8::kGlyphWidth; ++col) {
-            const uint8_t mask = static_cast<uint8_t>(1u << (font16x8::kGlyphWidth - 1 - col));
-            if ((rowBits & mask) == 0)
-                continue;
-            for (int sx = 0; sx < scale; ++sx) {
-                for (int sy = 0; sy < scale; ++sy) {
-                    int dstX;
-                    int dstY;
-                    if (clockwise) {
-                        dstX = x + row * scale + sx;
-                        dstY = y + (font16x8::kGlyphWidth - 1 - col) * scale + sy;
-                    } else {
-                        dstX = x + (font16x8::kGlyphHeight - 1 - row) * scale + sx;
-                        dstY = y + col * scale + sy;
-                    }
-                    fb.drawPixel(dstX, dstY, on);
-                }
-            }
-        }
-    }
-}
-
-void drawTextRotated(Framebuffer& fb, int x, int y, std::string_view text, bool clockwise, int scale = 1,
-                     bool on = true, int letterSpacing = 1) {
-    int       cursor  = y;
-    const int advance = (font16x8::kGlyphWidth + letterSpacing) * scale;
-    for (char ch: text) {
-        drawGlyphRotated(fb, x, cursor, ch, clockwise, scale, on);
-        cursor += advance;
-    }
-}
-
 class GameboyApp final : public cardboy::sdk::IApp {
 public:
     explicit GameboyApp(AppContext& ctx) :
@@ -157,8 +180,7 @@ public:
         prevInput = {};
         statusMessage.clear();
         resetFpsStats();
-        scaleMode     = ScaleMode::Original;
-        geometryDirty = true;
+        scaleMode = ScaleMode::FullHeightWide;
         ensureFilesystemReady();
         refreshRomList();
         mode         = Mode::Browse;
@@ -222,9 +244,7 @@ public:
                 }
 
 
-                GB_PERF_ONLY(const uint64_t geometryStartUs = nowMicros();)
-                ensureRenderGeometry();
-                GB_PERF_ONLY(perf.geometryUs = nowMicros() - geometryStartUs;)
+                GB_PERF_ONLY(perf.geometryUs = 0;)
 
                 GB_PERF_ONLY(const uint64_t runStartUs = nowMicros();)
                 gb_run_frame(&gb);
@@ -248,7 +268,7 @@ public:
 
 private:
     enum class Mode { Browse, Running };
-    enum class ScaleMode { Original, FullHeight };
+    enum class ScaleMode { Original, FullHeight, FullHeightWide };
 
     struct PerfTracker {
         enum class CallbackKind { RomRead, CartRamRead, CartRamWrite, LcdDraw, Error };
@@ -489,20 +509,40 @@ private:
 #endif
     };
 
-    struct RenderGeometry {
-        float                       scaleX       = 1.0f;
-        float                       scaleY       = 1.0f;
-        int                         scaledWidth  = LCD_WIDTH;
-        int                         scaledHeight = LCD_HEIGHT;
-        int                         offsetX      = 0;
-        int                         offsetY      = 0;
-        int                         leftMargin   = 0;
-        int                         rightMargin  = 0;
-        std::array<int, LCD_HEIGHT> lineYStart{};
-        std::array<int, LCD_HEIGHT> lineYEnd{};
-        std::array<int, LCD_WIDTH>  colXStart{};
-        std::array<int, LCD_WIDTH>  colXEnd{};
-    };
+    static constexpr int kOriginalOffsetX = (cardboy::sdk::kDisplayWidth - LCD_WIDTH) / 2;
+    static constexpr int kOriginalOffsetY = (cardboy::sdk::kDisplayHeight - LCD_HEIGHT) / 2;
+    static constexpr int kFullHeightScaledWidth =
+            (((LCD_WIDTH * cardboy::sdk::kDisplayHeight + LCD_HEIGHT / 2) / LCD_HEIGHT) + 7) & ~7;
+    static constexpr int kFullHeightOffsetX = (((cardboy::sdk::kDisplayWidth - kFullHeightScaledWidth) / 2) / 8) * 8;
+
+    static constexpr int kFullHeightWideWidth   = LCD_WIDTH * 2;
+    static constexpr int kFullHeightWideOffsetX = (((cardboy::sdk::kDisplayWidth - kFullHeightWideWidth) / 2) / 8) * 8;
+
+    static_assert(kFullHeightScaledWidth % 8 == 0);
+    static_assert(kFullHeightOffsetX % 8 == 0);
+    static_assert(kOriginalOffsetX % 8 == 0);
+    static_assert(kFullHeightOffsetX + kFullHeightScaledWidth <= cardboy::sdk::kDisplayWidth);
+    static_assert(kFullHeightWideWidth % 8 == 0);
+    static_assert(kFullHeightWideOffsetX % 8 == 0);
+    static_assert(kFullHeightWideOffsetX + kFullHeightWideWidth <= cardboy::sdk::kDisplayWidth);
+
+    inline static constexpr std::array<int, LCD_WIDTH + 1> kFullHeightColumnBounds = []() constexpr {
+        std::array<int, LCD_WIDTH + 1> bounds{};
+        for (int x = 0; x <= LCD_WIDTH; ++x)
+            bounds[static_cast<std::size_t>(x)] = kFullHeightOffsetX + (kFullHeightScaledWidth * x) / LCD_WIDTH;
+        return bounds;
+    }();
+
+    inline static constexpr std::array<int, LCD_HEIGHT + 1> kFullHeightRowBounds = []() constexpr {
+        std::array<int, LCD_HEIGHT + 1> bounds{};
+        for (int y = 0; y <= LCD_HEIGHT; ++y)
+            bounds[static_cast<std::size_t>(y)] = (cardboy::sdk::kDisplayHeight * y) / LCD_HEIGHT;
+        return bounds;
+    }();
+
+    static_assert(kFullHeightColumnBounds[0] % 8 == 0);
+    static_assert(kFullHeightColumnBounds[LCD_WIDTH] - kFullHeightColumnBounds[0] == kFullHeightScaledWidth);
+    static_assert(kFullHeightRowBounds[LCD_HEIGHT] == cardboy::sdk::kDisplayHeight);
 
     AppContext&                  context;
     Framebuffer&                 framebuffer;
@@ -514,9 +554,7 @@ private:
     static constexpr uint32_t    kTargetFrameUs    = 1000000 / 60; // ~16.6 ms
 
     Mode                  mode      = Mode::Browse;
-    ScaleMode             scaleMode     = ScaleMode::Original;
-    bool                  geometryDirty = true;
-    RenderGeometry        geometry{};
+    ScaleMode             scaleMode = ScaleMode::FullHeightWide;
     std::vector<RomEntry> roms;
     std::size_t           selectedIndex = 0;
     bool                  browserDirty  = true;
@@ -695,88 +733,21 @@ private:
     }
 
     void toggleScaleMode() {
-        if (scaleMode == ScaleMode::Original)
+        switch (scaleMode) {
+            case ScaleMode::Original:
                 scaleMode = ScaleMode::FullHeight;
-        else
+                setStatus("Scale: Full height");
+                break;
+            case ScaleMode::FullHeight:
+                scaleMode = ScaleMode::FullHeightWide;
+                setStatus("Scale: Full height 2x");
+                break;
+            case ScaleMode::FullHeightWide:
                 scaleMode = ScaleMode::Original;
-        geometryDirty = true;
+                setStatus("Scale: Original");
+                break;
+        }
         frameDirty = true;
-        setStatus(scaleMode == ScaleMode::FullHeight ? "Scale: Full height" : "Scale: Original");
-    }
-
-    void ensureRenderGeometry() {
-        if (!geometryDirty)
-            return;
-        geometryDirty = false;
-
-        auto&      geom      = geometry;
-        const int  fbWidth   = framebuffer.width();
-        const int  fbHeight  = framebuffer.height();
-        const auto resetGeom = [&]() {
-            geom.scaleX       = 1.0f;
-            geom.scaleY       = 1.0f;
-            geom.scaledWidth  = 0;
-            geom.scaledHeight = 0;
-            geom.offsetX      = 0;
-            geom.offsetY      = 0;
-            geom.leftMargin   = 0;
-            geom.rightMargin  = 0;
-            std::fill(geom.lineYStart.begin(), geom.lineYStart.end(), 0);
-            std::fill(geom.lineYEnd.begin(), geom.lineYEnd.end(), 0);
-            std::fill(geom.colXStart.begin(), geom.colXStart.end(), 0);
-            std::fill(geom.colXEnd.begin(), geom.colXEnd.end(), 0);
-        };
-
-        if (fbWidth <= 0 || fbHeight <= 0) {
-            resetGeom();
-            return;
-        }
-
-        int scaledWidth;
-        int scaledHeight;
-        if (scaleMode == ScaleMode::FullHeight) {
-            int targetHeight = fbHeight;
-            int targetWidth  = static_cast<int>((static_cast<int64_t>(LCD_WIDTH) * targetHeight + LCD_HEIGHT / 2) /
-                                                std::max(1, LCD_HEIGHT));
-            if (targetWidth > fbWidth) {
-                targetWidth  = fbWidth;
-                targetHeight = static_cast<int>((static_cast<int64_t>(LCD_HEIGHT) * targetWidth + LCD_WIDTH / 2) /
-                                                std::max(1, LCD_WIDTH));
-            }
-            scaledWidth  = std::clamp(targetWidth, 1, fbWidth);
-            scaledHeight = std::clamp(targetHeight, 1, fbHeight);
-        } else {
-            scaledWidth  = std::clamp(fbWidth, 1, LCD_WIDTH);
-            scaledHeight = std::clamp(fbHeight, 1, LCD_HEIGHT);
-        }
-
-        geom.scaledWidth  = scaledWidth;
-        geom.scaledHeight = scaledHeight;
-        geom.offsetX      = std::max(0, (fbWidth - scaledWidth) / 2);
-        geom.offsetY      = std::max(0, (fbHeight - scaledHeight) / 2);
-        geom.leftMargin   = geom.offsetX;
-        geom.rightMargin  = std::max(0, fbWidth - (geom.offsetX + scaledWidth));
-        geom.scaleX       = static_cast<float>(scaledWidth) / static_cast<float>(LCD_WIDTH);
-        geom.scaleY       = static_cast<float>(scaledHeight) / static_cast<float>(LCD_HEIGHT);
-
-        for (int srcLine = 0; srcLine < LCD_HEIGHT; ++srcLine) {
-            int start = geom.offsetY + static_cast<int>((static_cast<int64_t>(scaledHeight) * srcLine) / LCD_HEIGHT);
-            int end =
-                    geom.offsetY + static_cast<int>((static_cast<int64_t>(scaledHeight) * (srcLine + 1)) / LCD_HEIGHT);
-            start                    = std::clamp(start, 0, fbHeight);
-            end                      = std::clamp(end, start, fbHeight);
-            geom.lineYStart[srcLine] = start;
-            geom.lineYEnd[srcLine]   = end;
-        }
-
-        for (int srcCol = 0; srcCol < LCD_WIDTH; ++srcCol) {
-            int start = geom.offsetX + static_cast<int>((static_cast<int64_t>(scaledWidth) * srcCol) / LCD_WIDTH);
-            int end   = geom.offsetX + static_cast<int>((static_cast<int64_t>(scaledWidth) * (srcCol + 1)) / LCD_WIDTH);
-            start     = std::clamp(start, 0, fbWidth);
-            end       = std::clamp(end, start, fbWidth);
-            geom.colXStart[srcCol] = start;
-            geom.colXEnd[srcCol]   = end;
-        }
     }
 
     void handleBrowserInput(const InputState& input) {
@@ -921,8 +892,8 @@ private:
         }
 
         std::memset(&gb, 0, sizeof(gb));
-        const auto initResult = gb_init(&gb, &GameboyApp::romRead, &GameboyApp::cartRamRead, &GameboyApp::cartRamWrite,
-                                        &GameboyApp::errorCallback, this);
+        const auto initResult = gb_init(&gb, &GameboyApp::cartRamRead, &GameboyApp::cartRamWrite,
+                                        &GameboyApp::errorCallback, this, romDataView);
         if (initResult != GB_INIT_NO_ERROR) {
             setStatus(initErrorToString(initResult));
             romData.clear();
@@ -954,7 +925,6 @@ private:
         gbReady       = true;
         mode          = Mode::Running;
         frameDirty    = true;
-        geometryDirty = true;
         activeRomName = rom.name.empty() ? "Game" : rom.name;
 
         std::string statusText = "Running " + activeRomName;
@@ -973,7 +943,6 @@ private:
             cartRam.clear();
             activeRomName.clear();
             activeRomSavePath.clear();
-            geometryDirty = true;
             return;
         }
 
@@ -987,7 +956,6 @@ private:
         cartRam.clear();
         activeRomName.clear();
         activeRomSavePath.clear();
-        geometryDirty = true;
         std::memset(&gb, 0, sizeof(gb));
         mode         = Mode::Browse;
         browserDirty = true;
@@ -1034,8 +1002,6 @@ private:
             return;
         frameDirty = false;
 
-        ensureRenderGeometry();
-
         ++fpsFrameCounter;
         ++totalFrameCounter;
         const uint32_t nowMs = context.clock.millis();
@@ -1054,19 +1020,24 @@ private:
         const std::string fpsValue(fpsValueBuf);
         const std::string fpsLabel = "FPS";
         const std::string fpsText  = fpsValue + " FPS";
-        const std::string scaleHint = (scaleMode == ScaleMode::FullHeight) ? "START+B NORMAL" : "START+B SCALE";
 
-        if (scaleMode == ScaleMode::FullHeight) {
-            const auto& geom          = geometry;
+        std::string scaleHint;
+        switch (scaleMode) {
+            case ScaleMode::Original:
+                scaleHint = "START+B FULL";
+                break;
+            case ScaleMode::FullHeight:
+                scaleHint = "START+B WIDE";
+                break;
+            case ScaleMode::FullHeightWide:
+                scaleHint = "START+B NORMAL";
+                break;
+        }
+
+        if (scaleMode == ScaleMode::FullHeight || scaleMode == ScaleMode::FullHeightWide) {
             const int textScale    = 1;
-            const int   rotatedWidth  = font16x8::kGlyphHeight * textScale;
             const int screenHeight = framebuffer.height();
             const int screenWidth  = framebuffer.width();
-            const int   leftMargin    = std::max(0, geom.leftMargin);
-            const int   rightMargin   = std::max(0, geom.rightMargin);
-            const int   maxLeftX      = std::max(0, screenWidth - rotatedWidth);
-            const int   maxRightXBase = std::max(0, screenWidth - rotatedWidth);
-
 
             const int horizontalPadding = 8;
             const int fpsLineGap        = 4;
@@ -1081,18 +1052,24 @@ private:
 
             if (!activeRomName.empty()) {
                 const std::string rotatedRomName(activeRomName.rbegin(), activeRomName.rend());
-                const int         textHeight = measureVerticalText(rotatedRomName, textScale);
+                const auto        textBounds =
+                        font16x8::measureTextBounds(rotatedRomName, textScale, 1, font16x8::Rotation::Clockwise90);
+                const int textHeight = textBounds.height;
                 const int maxOrigin  = std::max(0, screenHeight - textHeight);
                 int       leftX      = 8;
                 int       leftY      = std::clamp((screenHeight - textHeight) / 2, 0, maxOrigin);
-                drawTextRotated(framebuffer, leftX, leftY, rotatedRomName, true, textScale, true, 1);
+                font16x8::drawText(framebuffer, leftX, leftY, rotatedRomName, textScale, true, 1,
+                                   font16x8::Rotation::Clockwise90);
                 if (!statusMessage.empty()) {
                     const std::string rotatedStatusMessage(statusMessage.rbegin(), statusMessage.rend());
-                    const int         textHeight = measureVerticalText(rotatedStatusMessage, textScale);
+                    const auto        statusBounds = font16x8::measureTextBounds(rotatedStatusMessage, textScale, 1,
+                                                                                 font16x8::Rotation::Clockwise90);
+                    const int         textHeight   = statusBounds.height;
                     const int         maxOrigin    = std::max(0, screenHeight - textHeight);
                     leftX                          = leftX + 20;
                     leftY                          = std::clamp((screenHeight - textHeight) / 2, 0, maxOrigin);
-                    drawTextRotated(framebuffer, leftX, leftY, rotatedStatusMessage, true, textScale, true, 1);
+                    font16x8::drawText(framebuffer, leftX, leftY, rotatedStatusMessage, textScale, true, 1,
+                                       font16x8::Rotation::Clockwise90);
                 }
             }
 
@@ -1110,7 +1087,8 @@ private:
                 std::string rotated(text.rbegin(), text.rend());
                 if (totalRightHeight > 0)
                     totalRightHeight += gap;
-                totalRightHeight += measureVerticalText(rotated, textScale);
+                const auto bounds = font16x8::measureTextBounds(rotated, textScale, 1, font16x8::Rotation::Clockwise90);
+                totalRightHeight += bounds.height;
             }
 
             const int maxRightOrigin = std::max(0, screenHeight - totalRightHeight);
@@ -1125,8 +1103,10 @@ private:
                 if (text.empty())
                     continue;
                 std::string rotated(text.rbegin(), text.rend());
-                rightY = screenHeight - measureVerticalText(rotated, textScale) - 8;
-                drawTextRotated(framebuffer, rightX, rightY, rotated, true, textScale, true, 1);
+                const auto bounds = font16x8::measureTextBounds(rotated, textScale, 1, font16x8::Rotation::Clockwise90);
+                rightY            = screenHeight - bounds.height - 8;
+                font16x8::drawText(framebuffer, rightX, rightY, rotated, textScale, true, 1,
+                                   font16x8::Rotation::Clockwise90);
                 rightX -= 20;
             }
 
@@ -1221,8 +1201,7 @@ private:
     }
 
     static GameboyApp* fromGb(struct gb_s* gb) {
-        if (!gb)
-            return nullptr;
+        CARDBOY_CHECK_CODE(if (!gb) return nullptr;);
         return static_cast<GameboyApp*>(gb->direct.priv);
     }
 
@@ -1236,16 +1215,122 @@ private:
         return value > threshold;
     }
 
+    static void drawLineOriginal(GameboyApp& self, const uint8_t pixels[160], int srcLine) {
+        Framebuffer& fb      = self.framebuffer;
+        const int    dstY    = kOriginalOffsetY + srcLine;
+        const int    baseX   = kOriginalOffsetX;
+        const int    yParity = dstY & 1;
+
+        CARDBOY_CHECK((baseX % 8) == 0);
+        CARDBOY_CHECK((LCD_WIDTH % 8) == 0); // 160 is fine
+
+        int x = 0;
+        while (x < LCD_WIDTH) {
+            // Build two 4-pixel packs (2 bits per pixel → 8-bit pack)
+            uint8_t p4a = pixels[x + 0] | (pixels[x + 1] << 2) | (pixels[x + 2] << 4) | (pixels[x + 3] << 6);
+
+            uint8_t p4b = pixels[x + 4] | (pixels[x + 5] << 2) | (pixels[x + 6] << 4) | (pixels[x + 7] << 6);
+
+            const int     xParity = (baseX + x) & 1;                   // start parity for this byte
+            const uint8_t n0      = kNibbleLUT[yParity][xParity][p4a];
+            const uint8_t n1      = kNibbleLUT[yParity][xParity][p4b]; // same parity after 4 pixels
+
+            const uint8_t pack = static_cast<uint8_t>((n0 << 4) | (n1 & 0x0F));
+            fb.drawBits8(baseX + x, dstY, pack);
+
+            x += 8;
+        }
+    }
+
+
+    static void drawLineFullHeight(GameboyApp& self, const uint8_t pixels[160], int srcLine) {
+        Framebuffer& fb     = self.framebuffer;
+        const int    yStart = kFullHeightRowBounds[static_cast<std::size_t>(srcLine)];
+        const int    yEnd   = kFullHeightRowBounds[static_cast<std::size_t>(srcLine) + 1];
+
+        CARDBOY_CHECK(yEnd > yStart);
+        CARDBOY_CHECK((kFullHeightColumnBounds[0] % 8) == 0);
+
+        for (int dstY = yStart; dstY < yEnd; ++dstY) {
+            std::array<uint8_t, 8> vals{}; // collected 2-bit pixel values for this output byte
+            int                    collected = 0;
+            const int              yParity   = dstY & 1;
+
+            for (int x = 0; x < LCD_WIDTH; ++x) {
+                const int colStart = kFullHeightColumnBounds[static_cast<std::size_t>(x)];
+                const int colEnd   = kFullHeightColumnBounds[static_cast<std::size_t>(x) + 1];
+                CARDBOY_CHECK(colEnd > colStart);
+
+                // expand this source pixel across its horizontal span
+                const uint8_t v = static_cast<uint8_t>(pixels[x]);
+                for (int dstX = colStart; dstX < colEnd; ++dstX) {
+                    vals[static_cast<std::size_t>(collected)] = v;
+                    ++collected;
+
+                    if (collected == 8) {
+                        const int byteStart = dstX - 7;
+                        CARDBOY_CHECK((byteStart % 8) == 0);
+                        const int xParity = byteStart & 1;
+
+                        // build two 4-pixel packs from the 8 collected values (mask to 2 bits)
+                        const uint8_t p4a = (vals[0]) | ((vals[1]) << 2) | ((vals[2]) << 4) | ((vals[3]) << 6);
+
+                        const uint8_t p4b = (vals[4]) | ((vals[5]) << 2) | ((vals[6]) << 4) | ((vals[7]) << 6);
+
+                        // two LUT hits → two nibbles → one byte
+                        const uint8_t n0   = kNibbleLUT[yParity][xParity][p4a];
+                        const uint8_t n1   = kNibbleLUT[yParity][xParity][p4b];
+                        const uint8_t pack = static_cast<uint8_t>((n0 << 4) | (n1 & 0x0F));
+
+                        fb.drawBits8(byteStart, dstY, pack);
+                        collected = 0; // reset for next byte
+                    }
+                }
+            }
+            CARDBOY_CHECK(collected == 0); // must end on byte boundary
+        }
+    }
+
+    static void drawLineFullHeightWide(GameboyApp& self, const uint8_t pixels[160], int srcLine) {
+        Framebuffer& fb     = self.framebuffer;
+        const int    yStart = kFullHeightRowBounds[static_cast<std::size_t>(srcLine)];
+        const int    yEnd   = kFullHeightRowBounds[static_cast<std::size_t>(srcLine) + 1];
+
+        CARDBOY_CHECK(yEnd > yStart);
+        CARDBOY_CHECK((kFullHeightWideOffsetX % 8) == 0);
+
+        const int xParityBase = kFullHeightWideOffsetX & 1;
+
+        for (int dstY = yStart; dstY < yEnd; ++dstY) {
+            const int yParity = dstY & 1;
+            int       dstX    = kFullHeightWideOffsetX;
+
+            for (int srcX = 0; srcX < LCD_WIDTH; srcX += 4) {
+                const uint8_t packIndex = static_cast<uint8_t>((pixels[srcX + 0]) | ((pixels[srcX + 1]) << 2) |
+                                                               ((pixels[srcX + 2]) << 4) | ((pixels[srcX + 3]) << 6));
+
+                const uint8_t pack = kFullHeightWideByteLUT[yParity][xParityBase][packIndex];
+
+                fb.drawBits8(dstX, dstY, pack);
+                dstX += 8;
+            }
+
+            CARDBOY_CHECK(dstX == kFullHeightWideOffsetX + kFullHeightWideWidth);
+        }
+    }
+
+
+public:
     static uint8_t romRead(struct gb_s* gb, const uint_fast32_t addr) {
         auto* self = fromGb(gb);
-        if (!self)
-            return 0xFF;
-        // ScopedCallbackTimer timer(self, PerfTracker::CallbackKind::RomRead);
-        if (!self->romDataView || addr >= self->romDataViewSize)
-            return 0xFF;
+        CARDBOY_CHECK_CODE(if (!self) return 0xFF;
+                           if (!self->romDataView || addr >= self->romDataViewSize) return 0xFF;);
+        // ScopedCallbackTimer timer(self,
+        // PerfTracker::CallbackKind::RomRead);
         return self->romDataView[static_cast<std::size_t>(addr)];
     }
 
+private:
     static uint8_t cartRamRead(struct gb_s* gb, const uint_fast32_t addr) {
         auto* self = fromGb(gb);
         if (!self)
@@ -1281,48 +1366,26 @@ private:
     __attribute__((optimize("Ofast"))) static void lcdDrawLine(struct gb_s* gb, const uint8_t pixels[160],
                                                                const uint_fast8_t line) {
         auto* self = fromGb(gb);
-        if (!self || line >= LCD_HEIGHT)
-            return;
+        CARDBOY_CHECK(self && line < LCD_HEIGHT);
 
         ScopedCallbackTimer timer(self, PerfTracker::CallbackKind::LcdDraw);
 
-        self->ensureRenderGeometry();
-        const auto& geom = self->geometry;
-        if (geom.scaledWidth <= 0 || geom.scaledHeight <= 0)
-            return;
-
-        const int yStart = geom.lineYStart[line];
-        const int yEnd   = geom.lineYEnd[line];
-        if (yStart >= yEnd)
-            return;
-
         self->frameDirty = true;
 
         Framebuffer& fb = self->framebuffer;
         fb.frameReady();
 
-        if (geom.scaledWidth == LCD_WIDTH && geom.scaledHeight == LCD_HEIGHT) {
-            const int dstY     = yStart;
-            const int dstXBase = geom.offsetX;
-            for (int x = 0; x < LCD_WIDTH; ++x) {
-                const bool on = shouldPixelBeOn(pixels[x], dstXBase + x, dstY);
-                fb.drawPixel(dstXBase + x, dstY, on);
-            }
-            return;
-        }
-
-        const auto& colStart = geom.colXStart;
-        const auto& colEnd   = geom.colXEnd;
-        for (int x = 0; x < LCD_WIDTH; ++x) {
-            const int drawStart = colStart[x];
-            const int drawEnd   = colEnd[x];
-            if (drawStart >= drawEnd)
-                continue;
-            for (int dstY = yStart; dstY < yEnd; ++dstY)
-                for (int dstX = drawStart; dstX < drawEnd; ++dstX) {
-                    const bool on = shouldPixelBeOn(pixels[x], dstX, dstY);
-                    fb.drawPixel(dstX, dstY, on);
-                }
+        switch (self->scaleMode) {
+            case ScaleMode::FullHeight:
+                drawLineFullHeight(*self, pixels, static_cast<int>(line));
+                break;
+            case ScaleMode::FullHeightWide:
+                drawLineFullHeightWide(*self, pixels, static_cast<int>(line));
+                break;
+            case ScaleMode::Original:
+            default:
+                drawLineOriginal(*self, pixels, static_cast<int>(line));
+                break;
         }
     }
 

Firmware/sdk/backend_interface/CMakeLists.txt

@@ -1,5 +1,7 @@
 add_library(cardboy_backend_interface INTERFACE)
 
+target_link_libraries(cardboy_backend_interface INTERFACE cardboy_utils)
+
 set_target_properties(cardboy_backend_interface PROPERTIES
         EXPORT_NAME backend_interface
 )

Firmware/sdk/backend_interface/include/cardboy/sdk/platform.hpp

@@ -23,6 +23,9 @@ concept HasSendFrameImpl = requires(Impl& impl, bool flag) {
     { impl.sendFrame_impl(flag) };
 };
 
+template<typename Impl>
+concept HasDrawBits8Impl = requires(Impl& impl, int x, int y, std::uint8_t bits) { impl.drawBits8_impl(x, y, bits); };
+
 template<typename Impl>
 concept HasFrameInFlightImpl = requires(const Impl& impl) {
     { impl.frameInFlight_impl() } -> std::convertible_to<bool>;
@@ -37,11 +40,19 @@ concept HasSleepMsImpl = requires(Impl& impl, std::uint32_t value) {
 template<typename Impl>
 class FramebufferFacade {
 public:
-    [[nodiscard]] int width() const { return impl().width_impl(); }
-    [[nodiscard]] int height() const { return impl().height_impl(); }
+    [[nodiscard]] __attribute__((always_inline)) int width() const { return impl().width_impl(); }
+    [[nodiscard]] __attribute__((always_inline)) int height() const { return impl().height_impl(); }
 
     __attribute__((always_inline)) void drawPixel(int x, int y, bool on) { impl().drawPixel_impl(x, y, on); }
 
+    __attribute__((always_inline)) void drawBits8(int x, int y, std::uint8_t bits) {
+        if constexpr (detail::HasDrawBits8Impl<Impl>) {
+            impl().drawBits8_impl(x, y, bits);
+        } else {
+            defaultDrawBits8(x, y, bits);
+        }
+    }
+
     void clear(bool on) {
         if constexpr (detail::HasClearImpl<Impl>) {
             impl().clear_impl(on);
@@ -79,6 +90,14 @@ private:
             for (int x = 0; x < width(); ++x)
                 drawPixel(x, y, on);
     }
+
+    void defaultDrawBits8(int x, int y, std::uint8_t bits) {
+        for (int col = 0; col < 8; ++col) {
+            const std::uint8_t mask    = static_cast<std::uint8_t>(1u << (7 - col));
+            const bool         pixelOn = (bits & mask) != 0;
+            drawPixel(x + col, y, pixelOn);
+        }
+    }
 };
 
 template<typename Impl>

Firmware/sdk/core/include/cardboy/gfx/font16x8.hpp

@@ -26,21 +26,75 @@ inline const std::array<uint8_t, kGlyphHeight>& glyphBitmap(char ch) {
     return fonts_Terminess_Powerline[uc];
 }
 
+enum class Rotation {
+    None,
+    Clockwise90,
+    CounterClockwise90,
+};
+
+struct TextBounds {
+    int width  = 0;
+    int height = 0;
+};
+
 template<typename Framebuffer>
-inline void drawGlyph(Framebuffer& fb, int x, int y, char ch, int scale = 1, bool on = true) {
+inline void drawGlyph(Framebuffer& fb, int x, int y, char ch, int scale = 1, bool on = true,
+                      Rotation rotation = Rotation::None) {
     const auto& rows = glyphBitmap(ch);
+    if (rotation == Rotation::None && scale == 1 && on && ((x % 8) == 0)) {
+        for (int row = 0; row < kGlyphHeight; ++row) {
+            const uint8_t rowBits = rows[row];
+            fb.drawBits8(x, y + row, rowBits);
+        }
+        return;
+    }
     for (int row = 0; row < kGlyphHeight; ++row) {
         const uint8_t rowBits = rows[row];
         for (int col = 0; col < kGlyphWidth; ++col) {
             const uint8_t mask = static_cast<uint8_t>(1u << (kGlyphWidth - 1 - col));
-            if (rowBits & mask) {
-                for (int sx = 0; sx < scale; ++sx)
-                    for (int sy = 0; sy < scale; ++sy)
-                        fb.drawPixel(x + col * scale + sx, y + row * scale + sy, on);
+            if ((rowBits & mask) == 0)
+                continue;
+            for (int sx = 0; sx < scale; ++sx) {
+                for (int sy = 0; sy < scale; ++sy) {
+                    int dstX = x;
+                    int dstY = y;
+                    switch (rotation) {
+                        case Rotation::None:
+                            dstX += col * scale + sx;
+                            dstY += row * scale + sy;
+                            break;
+                        case Rotation::Clockwise90:
+                            dstX += row * scale + sx;
+                            dstY += (kGlyphWidth - 1 - col) * scale + sy;
+                            break;
+                        case Rotation::CounterClockwise90:
+                            dstX += (kGlyphHeight - 1 - row) * scale + sx;
+                            dstY += col * scale + sy;
+                            break;
+                    }
+                    fb.drawPixel(dstX, dstY, on);
                 }
             }
         }
     }
+}
+
+inline TextBounds measureTextBounds(std::string_view text, int scale = 1, int letterSpacing = 1,
+                                    Rotation rotation = Rotation::None) {
+    if (text.empty())
+        return {};
+    const int advance = (kGlyphWidth + letterSpacing) * scale;
+    const int extent  = static_cast<int>(text.size()) * advance - letterSpacing * scale;
+    const int height  = kGlyphHeight * scale;
+    switch (rotation) {
+        case Rotation::None:
+            return {extent, height};
+        case Rotation::Clockwise90:
+        case Rotation::CounterClockwise90:
+            return {height, extent};
+    }
+    return {extent, height};
+}
 
 inline int measureText(std::string_view text, int scale = 1, int letterSpacing = 1) {
     if (text.empty())
@@ -51,11 +105,22 @@ inline int measureText(std::string_view text, int scale = 1, int letterSpacing =
 
 template<typename Framebuffer>
 inline void drawText(Framebuffer& fb, int x, int y, std::string_view text, int scale = 1, bool on = true,
-                     int letterSpacing = 1) {
+                     int letterSpacing = 1, Rotation rotation = Rotation::None) {
+    if (text.empty())
+        return;
+    const int advance = (kGlyphWidth + letterSpacing) * scale;
+    if (rotation == Rotation::None) {
         int cursor = x;
         for (char ch: text) {
-        drawGlyph(fb, cursor, y, ch, scale, on);
-        cursor += (kGlyphWidth + letterSpacing) * scale;
+            drawGlyph(fb, cursor, y, ch, scale, on, rotation);
+            cursor += advance;
+        }
+    } else {
+        int cursor = y;
+        for (char ch: text) {
+            drawGlyph(fb, x, cursor, ch, scale, on, rotation);
+            cursor += advance;
+        }
     }
 }
 

Firmware/sdk/utils/CMakeLists.txt

@@ -0,0 +1,21 @@
+add_library(cardboy_utils INTERFACE)
+
+option(CARDBOY_MORE_CHECKS "More checks" OFF)
+
+set_target_properties(cardboy_utils PROPERTIES
+        EXPORT_NAME utils
+)
+
+target_include_directories(cardboy_utils
+        INTERFACE
+        ${CMAKE_CURRENT_SOURCE_DIR}/include
+)
+
+if(CARDBOY_MORE_CHECKS)
+    target_compile_definitions(cardboy_utils INTERFACE CARDBOY_MORE_CHECKS=1)
+endif()
+
+target_sources(cardboy_utils
+    INTERFACE
+    ${CMAKE_CURRENT_SOURCE_DIR}/include/cardboy/utils/utils.hpp
+)

Firmware/sdk/utils/include/cardboy/utils/utils.hpp

@@ -0,0 +1,59 @@
+//
+// Created by Stepan Usatiuk on 11.10.2025.
+//
+
+#ifndef CARDBOY_SDK_UTILS_HPP
+#define CARDBOY_SDK_UTILS_HPP
+
+#ifndef CARDBOY_MORE_CHECKS
+#define CARDBOY_MORE_CHECKS 0
+#endif
+
+#if CARDBOY_MORE_CHECKS
+#include <cstdio>
+#include <cstdlib>
+
+// Fails the program with a message. Internal use.
+#define CARDBOY__CHECK_FAIL_IMPL(expr_str, file, line, func, msg_opt)            \
+    do {                                                                         \
+        std::fprintf(stderr,                                                     \
+                      "CARDBOY_CHECK failed: %s\n  at %s:%d in %s%s%s\n",        \
+                      (expr_str), (file), (line), (func),                         \
+                      ((msg_opt) ? "\n  message: " : ""),                        \
+                      ((msg_opt) ? (msg_opt) : ""));                             \
+        std::fflush(stderr);                                                     \
+        std::abort();                                                            \
+    } while (0)
+
+// Runtime check that is active only when CARDBOY_MORE_CHECKS != 0.
+// Evaluates the expression exactly once.
+#define CARDBOY_CHECK(expr)                                                      \
+    do {                                                                         \
+        if (!(expr)) {                                                           \
+            CARDBOY__CHECK_FAIL_IMPL(#expr, __FILE__, __LINE__, __func__, NULL); \
+        }                                                                        \
+    } while (0)
+
+// Same as CARDBOY_CHECK but allows providing a custom C-string message.
+#define CARDBOY_CHECK_MSG(expr, msg)                                             \
+    do {                                                                         \
+        if (!(expr)) {                                                           \
+            CARDBOY__CHECK_FAIL_IMPL(#expr, __FILE__, __LINE__, __func__, (msg));\
+        }                                                                        \
+    } while (0)
+
+// Execute arbitrary code only when checks are enabled.
+#define CARDBOY_CHECK_CODE(code)                                                 \
+    do {                                                                         \
+        code;                                                                    \
+    } while (0)
+
+#else
+// Checks compiled out when CARDBOY_MORE_CHECKS == 0.
+#define CARDBOY_CHECK(expr)        do { (void)sizeof(expr); } while (0)
+#define CARDBOY_CHECK_MSG(expr, _) do { (void)sizeof(expr); } while (0)
+#define CARDBOY_CHECK_CODE(code)   do { } while (0)
+#endif
+
+
+#endif // CARDBOY_SDK_UTILS_HPP

Firmware/sdkconfig

@@ -644,7 +644,7 @@ CONFIG_COMPILER_RT_LIB_GCCLIB=y
 CONFIG_COMPILER_RT_LIB_NAME="gcc"
 CONFIG_COMPILER_ORPHAN_SECTIONS_WARNING=y
 # CONFIG_COMPILER_ORPHAN_SECTIONS_PLACE is not set
-CONFIG_COMPILER_STATIC_ANALYZER=y
+# CONFIG_COMPILER_STATIC_ANALYZER is not set
 # end of Compiler options
 
 #