diff --git a/gpu.c b/gpu.c
index 2cc5359..e05fa67 100644
--- a/gpu.c
+++ b/gpu.c
@@ -436,9 +436,15 @@ void gpu_init(gpu_ctx_t *ctx, const gpu_config_t *config)
 	};
 
 	// Create pipeline layout
+	const VkPushConstantRange mvp_range = {
+		.stageFlags = VK_SHADER_STAGE_VERTEX_BIT,
+		.offset = 0,
+		.size = sizeof(mvp_t),
+	};
 	const VkPipelineLayoutCreateInfo pipeline_layout_config = {
 		.sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO,
-		.pushConstantRangeCount = 0,
+		.pushConstantRangeCount = 1,
+		.pPushConstantRanges = &mvp_range,
 	};
 	result = vkCreatePipelineLayout(
 	    ctx->dev, &pipeline_layout_config, nullptr, &ctx->pipeline_layout);
@@ -641,7 +647,7 @@ void gpu_cleanup(const gpu_ctx_t *ctx)
 	vkDestroyInstance(ctx->inst, nullptr);
 }
 
-void gpu_render(const gpu_ctx_t *ctx)
+void gpu_render(const gpu_ctx_t *ctx, const mvp_t *mvp)
 {
 	VkResult result;
 
@@ -679,6 +685,11 @@ void gpu_render(const gpu_ctx_t *ctx)
 	vkCmdBindPipeline(
 	    ctx->cmd_buf, VK_PIPELINE_BIND_POINT_GRAPHICS, ctx->pipeline);
 
+	// Push MVP
+	vkCmdPushConstants(
+	    ctx->cmd_buf, ctx->pipeline_layout, VK_SHADER_STAGE_VERTEX_BIT, 0,
+	    sizeof(mvp_t), mvp);
+
 	// Bind vertex and index buffers then draw
 	vkCmdBindVertexBuffers(
 	    ctx->cmd_buf, 0, 1, &ctx->vertex_buf, &(VkDeviceSize) { 0 });
diff --git a/gpu.h b/gpu.h
index 819ebd7..8bb2fc4 100644
--- a/gpu.h
+++ b/gpu.h
@@ -6,10 +6,21 @@
 
 #define SWAPCHAIN_IMG_BUFFER_SIZE 4
 
+typedef float vec2_t[3];
+typedef float vec3_t[3];
+typedef float mat4_t[16];
+
 typedef struct {
-	float position[2], colour[3];
+	vec2_t position;
+	vec3_t colour;
 } vertex_t;
 
+typedef struct {
+	mat4_t model;
+	mat4_t view;
+	mat4_t proj;
+} mvp_t;
+
 typedef struct {
 	VkInstance inst;
 	VkSurfaceKHR surface;
@@ -49,6 +60,6 @@ typedef struct {
 
 void gpu_init(gpu_ctx_t *ctx, const gpu_config_t *config);
 void gpu_cleanup(const gpu_ctx_t *ctx);
-void gpu_render(const gpu_ctx_t *ctx);
+void gpu_render(const gpu_ctx_t *ctx, const mvp_t *mvp);
 
 #endif
diff --git a/main.c b/main.c
index b824e34..5eb3813 100644
--- a/main.c
+++ b/main.c
@@ -2,16 +2,56 @@
 
 #include <assert.h>
 #include <math.h>
+#include <stdio.h>
+#include <string.h>
+
+#ifndef M_PI
+#define M_PI 3.14159265358979324
+#endif
 
 static const vertex_t vertices[] = {
-	{ { -0.5, -0.5 }, { 1.0, 0.0, 0.0 } },
-	{ { 0.5, -0.5 }, { 0.0, 1.0, 0.0 } },
-	{ { 0.5, 0.5 }, { 0.0, 0.0, 1.0 } },
-	{ { -0.5, 0.5 }, { 1.0, 1.0, 1.0 } },
+	{ { -1, -1 }, { 1, 0, 0 } },
+	{ { 1, -1 }, { 0, 1, 0 } },
+	{ { 1, 1 }, { 0, 0, 1 } },
+	{ { -1, 1 }, { 1, 1, 1 } },
 };
 
 static const uint16_t indices[] = { 0, 1, 2, 2, 3, 0 };
 
+// clang-format off
+#define ID \
+	{ \
+		1, 0, 0, 0, \
+		0, 1, 0, 0, \
+		0, 0, 1, 0, \
+		0, 0, 0, 1, \
+	}
+
+#define SCALE(sf) \
+	{ \
+		sf, 0,  0,  0, \
+		0,  sf, 0,  0, \
+		0,  0,  sf, 0, \
+		0,  0,  0,  1, \
+	}
+
+#define TRANSLATE(x, y, z) \
+	{ \
+		1, 0, 0, 0, \
+		0, 1, 0, 0, \
+		0, 0, 1, 0, \
+		x, y, z, 1, \
+	}
+
+#define ROT_Y(a) \
+	{ \
+		cosf(a), 0, -sinf(a), 0, \
+		0,       1, 0,        0, \
+		sinf(a), 0, cosf(a),  0, \
+		0,       0, 0,        1, \
+	}
+// clang-format on
+
 static void
 key_callback(GLFWwindow *window, int key, int scancode, int action, int mods)
 {
@@ -27,13 +67,24 @@ key_callback(GLFWwindow *window, int key, int scancode, int action, int mods)
 	}
 }
 
+static void mul(mat4_t out, const mat4_t left, const mat4_t right)
+{
+	for (unsigned y = 0; y < 4; ++y) {
+		for (unsigned x = 0; x < 4; ++x) {
+			out[y + 4 * x] = 0;
+			for (unsigned i = 0; i < 4; ++i)
+				out[y + 4 * x] += left[y + 4 * i] * right[i + 4 * x];
+		}
+	}
+}
+
 int main()
 {
 	// Set up window
 	glfwInit();
 	glfwWindowHint(GLFW_CLIENT_API, GLFW_NO_API);
 	glfwWindowHint(GLFW_RESIZABLE, GLFW_FALSE);
-	GLFWwindow *window = glfwCreateWindow(800, 600, "Foo", nullptr, nullptr);
+	GLFWwindow *window = glfwCreateWindow(800, 800, "Foo", nullptr, nullptr);
 	assert(window);
 
 	// Register input callbacks
@@ -51,9 +102,21 @@ int main()
 	gpu_ctx_t gpu;
 	gpu_init(&gpu, &gpu_config);
 
+	mvp_t mvp = {
+		.model = SCALE(0.2),
+		.view = TRANSLATE(0, 0, 0.5),
+		.proj = ID,
+	};
+
+	const mat4_t rot = ROT_Y(M_PI / 128);
 	while (!glfwWindowShouldClose(window)) {
 		glfwPollEvents();
-		gpu_render(&gpu);
+
+		mat4_t tmp;
+		memcpy(tmp, mvp.model, sizeof(mat4_t));
+		mul(mvp.model, rot, tmp);
+
+		gpu_render(&gpu, &mvp);
 	}
 
 	// Clean up
diff --git a/vert.glsl b/vert.glsl
index 3e2a810..1b02ecb 100644
--- a/vert.glsl
+++ b/vert.glsl
@@ -1,11 +1,17 @@
 #version 450
 
-layout(location = 0) in vec2 position;
+layout(location = 0) in vec2 pos;
 layout(location = 1) in vec3 colour;
 
 layout(location = 0) out vec3 frag_colour;
 
+layout(push_constant) uniform MVP {
+	mat4 model;
+	mat4 view;
+	mat4 proj;
+} mvp;
+
 void main() {
-	gl_Position = vec4(position, 0.0, 1.0);
+	gl_Position = mvp.proj * mvp.view * mvp.model * vec4(pos, 0.0, 1.0);
 	frag_colour = colour;
 }