instancing.rin
src/4.advanced_d3d11/10.1.instancing/instancing.rin · 369 lines
// 10.1 Instancing
//
// Drawing the same mesh many times, each copy placed differently, in one draw
// call. Twenty thousand rocks orbit a planet here; drawn one by one that would
// be twenty thousand draws, each with its own constant buffer update.
//
// Instead every rock's model matrix goes into one more vertex buffer, marked
// PER_INSTANCE_DATA in the input layout. The rock's own vertices are read per
// vertex as usual; the matrix advances once per copy. DrawInstanced says how
// many copies to make.
//
// A mat4 does not fit one input element (at most four floats), so it is sent as
// four: INSTANCE0..3, one row each, put back together in the shader.
import "../../../includes/win32.rin"
import "../../../includes/math.rin"
import "../../../includes/learn/shader.rin"
import "../../../includes/learn/camera.rin"
import "../../../includes/learn/model.rin"
import "std/print.rin"
import "std/memops.rin"
// settings
scr_width: i32 = 1280;
scr_height: i32 = 720;
rand: proc[external]() -> i32 = {}
srand: proc[external](seed: u32) -> void = {}
amount: u32 = 20000;
instance_layout: [7]D3D11_INPUT_ELEMENT_DESC = {
{ .SemanticName = "POSITION", .Format = DXGI_FORMAT.DXGI_FORMAT_R32G32B32_FLOAT, .AlignedByteOffset = 0 },
{ .SemanticName = "NORMAL", .Format = DXGI_FORMAT.DXGI_FORMAT_R32G32B32_FLOAT, .AlignedByteOffset = 12 },
{ .SemanticName = "TEXCOORD", .Format = DXGI_FORMAT.DXGI_FORMAT_R32G32_FLOAT, .AlignedByteOffset = 24 },
{ .SemanticName = "INSTANCE", .SemanticIndex = 0, .Format = DXGI_FORMAT.DXGI_FORMAT_R32G32B32A32_FLOAT, .InputSlot = 1, .AlignedByteOffset = 0, .InputSlotClass = D3D11_INPUT_CLASSIFICATION.D3D11_INPUT_PER_INSTANCE_DATA, .InstanceDataStepRate = 1 },
{ .SemanticName = "INSTANCE", .SemanticIndex = 1, .Format = DXGI_FORMAT.DXGI_FORMAT_R32G32B32A32_FLOAT, .InputSlot = 1, .AlignedByteOffset = 16, .InputSlotClass = D3D11_INPUT_CLASSIFICATION.D3D11_INPUT_PER_INSTANCE_DATA, .InstanceDataStepRate = 1 },
{ .SemanticName = "INSTANCE", .SemanticIndex = 2, .Format = DXGI_FORMAT.DXGI_FORMAT_R32G32B32A32_FLOAT, .InputSlot = 1, .AlignedByteOffset = 32, .InputSlotClass = D3D11_INPUT_CLASSIFICATION.D3D11_INPUT_PER_INSTANCE_DATA, .InstanceDataStepRate = 1 },
{ .SemanticName = "INSTANCE", .SemanticIndex = 3, .Format = DXGI_FORMAT.DXGI_FORMAT_R32G32B32A32_FLOAT, .InputSlot = 1, .AlignedByteOffset = 48, .InputSlotClass = D3D11_INPUT_CLASSIFICATION.D3D11_INPUT_PER_INSTANCE_DATA, .InstanceDataStepRate = 1 },
};
// Mirrors `cbuffer transforms` in the shaders, register for register.
transforms: struct = {
model: mat4;
normal_matrix: mat4;
view: mat4;
projection: mat4;
view_position: vec3;
pad: f32;
}
random01: proc() -> f32 = {
return cast(rand() % 10000, f32) / 10000.0;
}
// camera
cam: camera = {};
last_cursor: POINT = {};
scroll: f32 = 0.0;
// timing
delta_time: f32 = 0.0;
last_frame: f32 = 0.0;
perf_frequency: i64 = 0;
perf_start: i64 = 0;
// Seconds since the first call: what glfwGetTime is to LearnOpenGL.
get_time: proc() -> f32 = {
now: LARGE_INTEGER = {};
if (perf_frequency == 0) {
QueryPerformanceFrequency(now&);
perf_frequency = now.QuadPart;
QueryPerformanceCounter(now&);
perf_start = now.QuadPart;
}
QueryPerformanceCounter(now&);
return cast(now.QuadPart - perf_start, f32) / cast(perf_frequency, f32);
}
// win32: every event for the window (keys, mouse, closing) arrives here
// ---------------------------------------------------------------------
window_proc: proc[callconv(CALLBACK)](hwnd: HWND, msg: UINT, wparam: WPARAM, lparam: LPARAM) -> LRESULT = {
if (msg == WM_DESTROY) {
PostQuitMessage(0);
return 0;
}
if (msg == WM_MOUSEWHEEL) {
// The high word of wparam is the distance turned, 120 per notch.
scroll += cast(cast((wparam shr 16) & 0xffff, i16), f32) / 120.0;
return 0;
}
return DefWindowProcA(hwnd, msg, wparam, lparam);
}
// process all input: ask Windows whether relevant keys are down this frame and react accordingly
// ----------------------------------------------------------------------------------------------
// The top bit of GetAsyncKeyState is set while a key is held, which makes the SHORT negative.
key_down: proc(key: i32) -> b32 = {
return GetAsyncKeyState(key) < 0;
}
process_input: proc(window: HWND) -> void = {
cursor: POINT = {};
GetCursorPos(cursor&);
// GetAsyncKeyState reads the whole keyboard, so ignore it while another
// window has the focus; glfwGetKey only ever sees its own window's keys.
if (GetForegroundWindow() != window) {
last_cursor = cursor;
scroll = 0.0;
return;
}
if (key_down(VK_ESCAPE)) { DestroyWindow(window); }
if (key_down('W')) { camera_process_keyboard(cam&, camera_movement.forward, delta_time); }
if (key_down('S')) { camera_process_keyboard(cam&, camera_movement.backward, delta_time); }
if (key_down('A')) { camera_process_keyboard(cam&, camera_movement.left, delta_time); }
if (key_down('D')) { camera_process_keyboard(cam&, camera_movement.right, delta_time); }
if (key_down(VK_SPACE)) { camera_process_keyboard(cam&, camera_movement.up, delta_time); }
if (key_down(VK_SHIFT)) { camera_process_keyboard(cam&, camera_movement.down, delta_time); }
// Hold the right mouse button to look around.
if (key_down(VK_RBUTTON)) {
// Screen y grows downwards, so moving the mouse up is a negative dy: pitch up.
camera_process_mouse(cam&, cast(cursor.x - last_cursor.x, f32), cast(last_cursor.y - cursor.y, f32));
}
last_cursor = cursor;
camera_process_scroll(cam&, scroll);
scroll = 0.0;
}
// An immutable buffer: filled once at creation, never written by the CPU again.
create_buffer: proc(device: *ID3D11Device, bind: UINT, data: *const void, size: u64) -> *ID3D11Buffer = {
desc: D3D11_BUFFER_DESC = {
.ByteWidth = cast(size, UINT),
.Usage = D3D11_USAGE.D3D11_USAGE_IMMUTABLE,
.BindFlags = bind,
};
init: D3D11_SUBRESOURCE_DATA = { .pSysMem = data };
buffer: *ID3D11Buffer = null;
ID3D11Device_CreateBuffer(device, desc&, init&, buffer&);
return buffer;
}
// A constant buffer the CPU rewrites, usually every frame. Its size must be a
// multiple of 16 bytes, the size of one HLSL register.
create_constant_buffer: proc(device: *ID3D11Device, size: u64) -> *ID3D11Buffer = {
desc: D3D11_BUFFER_DESC = {
.ByteWidth = cast(size, UINT),
.Usage = D3D11_USAGE.D3D11_USAGE_DYNAMIC,
.BindFlags = D3D11_BIND_CONSTANT_BUFFER,
.CPUAccessFlags = D3D11_CPU_ACCESS_WRITE,
};
buffer: *ID3D11Buffer = null;
ID3D11Device_CreateBuffer(device, desc&, null, buffer&);
return buffer;
}
// WRITE_DISCARD hands back fresh memory, so the GPU never waits on the CPU.
update_buffer: proc(context: *ID3D11DeviceContext, buffer: *ID3D11Buffer, data: *const void, size: u64) -> void = {
mapped: D3D11_MAPPED_SUBRESOURCE = {};
ID3D11DeviceContext_Map(context, cast(buffer, *ID3D11Resource), 0, D3D11_MAP.D3D11_MAP_WRITE_DISCARD, 0, mapped&);
memcpy(mapped.pData, data, size);
ID3D11DeviceContext_Unmap(context, cast(buffer, *ID3D11Resource), 0);
}
// The same buffer in the same register for the vertex, geometry and pixel stages.
bind_constants: proc(context: *ID3D11DeviceContext, slot: UINT, buffer: *ID3D11Buffer) -> void = {
buffers: [1]*ID3D11Buffer = { buffer };
ID3D11DeviceContext_VSSetConstantBuffers(context, slot, 1, buffers[0]&);
ID3D11DeviceContext_GSSetConstantBuffers(context, slot, 1, buffers[0]&);
ID3D11DeviceContext_PSSetConstantBuffers(context, slot, 1, buffers[0]&);
}
main: proc() -> i32 = {
// win32: register a window class and create the window
// ----------------------------------------------------
SetProcessDPIAware();
instance: HMODULE = GetModuleHandleA(null);
wc: WNDCLASSA = {
.lpfnWndProc = window_proc,
.hInstance = instance,
.hCursor = LoadCursorA(null, IDC_ARROW),
.lpszClassName = "LearnD3D11",
};
RegisterClassA(wc&);
// A fixed size: the swap chain and every render target keep the size they
// were created with, so nothing needs rebuilding on resize.
style: DWORD = WS_OVERLAPPEDWINDOW & ~(WS_THICKFRAME | WS_MAXIMIZEBOX);
rect: RECT = { .right = scr_width, .bottom = scr_height };
AdjustWindowRect(rect&, style, 0);
window: HWND = CreateWindowExA(0, wc.lpszClassName, "LearnD3D11", style | WS_VISIBLE, CW_USEDEFAULT, CW_USEDEFAULT,
rect.right - rect.left, rect.bottom - rect.top, null, null, instance, null);
if (window == null) {
printf("Failed to create the window\n");
return -1;
}
// d3d11: create the device, its immediate context and a swap chain
// ----------------------------------------------------------------
swap_desc: DXGI_SWAP_CHAIN_DESC = {
.BufferDesc = {
.Width = cast(scr_width, UINT),
.Height = cast(scr_height, UINT),
.Format = DXGI_FORMAT.DXGI_FORMAT_R8G8B8A8_UNORM,
},
.SampleDesc = { .Count = 1 },
.BufferUsage = DXGI_USAGE_RENDER_TARGET_OUTPUT,
.BufferCount = 2,
.OutputWindow = window,
.Windowed = 1,
.SwapEffect = DXGI_SWAP_EFFECT.DXGI_SWAP_EFFECT_FLIP_DISCARD,
};
feature_level: D3D_FEATURE_LEVEL = D3D_FEATURE_LEVEL.D3D_FEATURE_LEVEL_11_0;
swap_chain: *IDXGISwapChain = null;
device: *ID3D11Device = null;
context: *ID3D11DeviceContext = null;
if (D3D11CreateDeviceAndSwapChain(null, D3D_DRIVER_TYPE.D3D_DRIVER_TYPE_HARDWARE, null, 0, feature_level&, 1, D3D11_SDK_VERSION,
swap_desc&, swap_chain&, device&, null, context&) < 0) {
printf("Failed to create the D3D11 device\n");
return -1;
}
backbuffer: *ID3D11Texture2D = null;
IDXGISwapChain_GetBuffer(swap_chain, 0, IID_ID3D11Texture2D&, cast(backbuffer&, **void));
backbuffer_rtv: *ID3D11RenderTargetView = null;
ID3D11Device_CreateRenderTargetView(device, cast(backbuffer, *ID3D11Resource), null, backbuffer_rtv&);
// A depth buffer the size of the window: 24 bits of depth, 8 of stencil.
depth_desc: D3D11_TEXTURE2D_DESC = {
.Width = cast(scr_width, UINT),
.Height = cast(scr_height, UINT),
.MipLevels = 1,
.ArraySize = 1,
.Format = DXGI_FORMAT.DXGI_FORMAT_D24_UNORM_S8_UINT,
.SampleDesc = { .Count = 1 },
.BindFlags = D3D11_BIND_DEPTH_STENCIL,
};
depth_texture: *ID3D11Texture2D = null;
depth_dsv: *ID3D11DepthStencilView = null;
ID3D11Device_CreateTexture2D(device, depth_desc&, null, depth_texture&);
ID3D11Device_CreateDepthStencilView(device, cast(depth_texture, *ID3D11Resource), null, depth_dsv&);
// configure global d3d11 state
// ----------------------------
// GL's conventions rather than D3D's: counter-clockwise triangles face the
// viewer, and nothing is culled.
raster_desc: D3D11_RASTERIZER_DESC = {
.FillMode = D3D11_FILL_MODE.D3D11_FILL_SOLID,
.CullMode = D3D11_CULL_MODE.D3D11_CULL_NONE,
.FrontCounterClockwise = 1,
.DepthClipEnable = 1,
};
raster_state: *ID3D11RasterizerState = null;
ID3D11Device_CreateRasterizerState(device, raster_desc&, raster_state&);
ID3D11DeviceContext_RSSetState(context, raster_state);
viewport: D3D11_VIEWPORT = {
.Width = cast(scr_width, f32),
.Height = cast(scr_height, f32),
.MaxDepth = 1.0,
};
// A sampler: how textures are filtered and what happens past their edges.
sampler_desc: D3D11_SAMPLER_DESC = {
.Filter = D3D11_FILTER.D3D11_FILTER_ANISOTROPIC,
.AddressU = D3D11_TEXTURE_ADDRESS_MODE.D3D11_TEXTURE_ADDRESS_WRAP,
.AddressV = D3D11_TEXTURE_ADDRESS_MODE.D3D11_TEXTURE_ADDRESS_WRAP,
.AddressW = D3D11_TEXTURE_ADDRESS_MODE.D3D11_TEXTURE_ADDRESS_WRAP,
.MaxAnisotropy = 8,
.MaxLOD = D3D11_FLOAT32_MAX,
};
sampler: *ID3D11SamplerState = null;
ID3D11Device_CreateSamplerState(device, sampler_desc&, sampler&);
ID3D11DeviceContext_PSSetSamplers(context, 0, 1, sampler&);
// build and compile shaders
// -------------------------
planet_shader: shader = shader_create(device, filesystem_path("src/4.advanced_d3d11/10.1.instancing/planet.hlsl"), model_vertex_layout);
rock_shader: shader = shader_create(device, filesystem_path("src/4.advanced_d3d11/10.1.instancing/asteroids.hlsl"), instance_layout);
// load models
// -----------
planet: model = model_load(device, context, "resources/objects/planet/", "planet.obj", 1);
rock: model = model_load(device, context, "resources/objects/rock/", "rock.obj", 1);
transforms_buffer: *ID3D11Buffer = create_constant_buffer(device, sizeof(transforms));
cam = camera_create(v3(0.0, 12.0, 70.0), -90.0, -10.0);
cam.speed = 15.0;
cam.zoom = 50.0;
// generate a large list of semi-random model transformation matrices: a
// ring, each rock at a random angle, nudged off the circle, then given a
// random size and spin. Seeded, so every run is the same ring.
// -----------------------------------------------------------------------
arena: memops_arena = {};
memops_arena_initialize(arena&);
srand(7);
matrices: [*]mat4 = cast(memops_arena_push(arena&, cast(amount, u64) * sizeof(mat4), 16), [*]mat4);
for (i: u32 = 0; i < amount; i += 1) {
angle: f32 = cast(i, f32) / cast(amount, f32) * 2.0 * pi;
radius: f32 = 45.0;
offset: f32 = 6.0;
x: f32 = sinf(angle) * radius + (random01() * 2.0 - 1.0) * offset;
y: f32 = (random01() * 2.0 - 1.0) * offset * 0.3;
z: f32 = cosf(angle) * radius + (random01() * 2.0 - 1.0) * offset;
scale: f32 = 0.05 + random01() * 0.2;
spin: mat4 = m4_rotate(random01() * 2.0 * pi, v3(0.4, 0.6, 0.8));
matrices[i] = m4_mul(m4_translate(v3(x, y, z)), m4_mul(spin, m4_scale(v3(scale, scale, scale))));
}
instance_buffer: *ID3D11Buffer = create_buffer(device, D3D11_BIND_VERTEX_BUFFER, matrices, cast(amount, u64) * sizeof(mat4));
// render loop
// -----------
running: b32 = 1;
while (running) {
msg: MSG = {};
while (PeekMessageA(msg&, null, 0, 0, PM_REMOVE) != 0) {
if (msg.message == WM_QUIT) { running = 0; }
TranslateMessage(msg&);
DispatchMessageA(msg&);
}
// per-frame time logic
// --------------------
current_frame: f32 = get_time();
delta_time = current_frame - last_frame;
last_frame = current_frame;
// input
// -----
process_input(window);
// render
// ------
clear_color: [4]f32 = { 0.02, 0.02, 0.03, 1.0 };
ID3D11DeviceContext_OMSetRenderTargets(context, 1, backbuffer_rtv&, depth_dsv);
ID3D11DeviceContext_RSSetViewports(context, 1, viewport&);
ID3D11DeviceContext_ClearRenderTargetView(context, backbuffer_rtv, clear_color[0]&);
ID3D11DeviceContext_ClearDepthStencilView(context, depth_dsv, D3D11_CLEAR_DEPTH | D3D11_CLEAR_STENCIL, 1.0, 0);
t: transforms = {
.view = camera_view(cam&),
.projection = m4_perspective(radians(cam.zoom), cast(scr_width, f32) / cast(scr_height, f32), 0.1, 500.0),
.view_position = cam.position,
};
bind_constants(context, 0, transforms_buffer);
// draw planet: one ordinary draw
t.model = m4_mul(m4_rotate(current_frame * 0.1, v3(0.0, 1.0, 0.0)), m4_scale(v3(4.0, 4.0, 4.0)));
t.normal_matrix = m4_normal_matrix(t.model);
update_buffer(context, transforms_buffer, t&, sizeof(transforms));
shader_use(context, planet_shader&);
model_draw(context, planet&);
// draw meteorites: slot 0 is the rock, slot 1 the matrices, one draw
t.model = m4_rotate(current_frame * 0.02, v3(0.0, 1.0, 0.0));
update_buffer(context, transforms_buffer, t&, sizeof(transforms));
shader_use(context, rock_shader&);
// rock.mtl lists its colour map as map_Bump, so it is in the normal slot
views: [1]*ID3D11ShaderResourceView = { rock.normal };
ID3D11DeviceContext_PSSetShaderResources(context, 0, 1, views[0]&);
buffers: [2]*ID3D11Buffer = { rock.vertices, instance_buffer };
strides: [2]UINT = { sizeof(model_vertex), sizeof(mat4) };
offsets: [2]UINT = { 0, 0 };
ID3D11DeviceContext_IASetVertexBuffers(context, 0, 2, buffers[0]&, strides[0]&, offsets[0]&);
ID3D11DeviceContext_DrawInstanced(context, rock.vertex_count, amount, 0, 0);
// dxgi: show the image just drawn, waiting for the next vertical blank
// --------------------------------------------------------------------
IDXGISwapChain_Present(swap_chain, 1, 0);
}
return 0;
}