A fast 2D and 3D game engine — TypeScript SDK, C++/WebAssembly core, visual editor. Ship one project to web, desktop, WeChat MiniGames, playable ads, and native Android / iOS.
TypeScript 2d-game-engine3d3d-game-engineandroidcppecsgame-editorgame-enginegamedeviostypescriptwebassemblywebglwebgpuwechat-minigame
// SPDX-License-Identifier: Apache-2.0
// SPDX-FileCopyrightText: Copyright (c) 2024-present ESEngine Team
/**
* @file Half.hpp
* @brief IEEE binary32 → binary16, for the data the GPU takes at half width.
*
* @details One author: a mesh's morph deltas and a frame's probe coefficients
* both reach the GPU as RGBA16F, and two conversions would be two
* roundings to keep agreeing.
*/
#pragma once
#include "./Types.hpp"
#include <cstring>
namespace esengine {
/**
* @brief IEEE binary32 to binary16.
* @details The end that matters is the subnormal one — hence the explicit path
* rather than a cast. A value too large for half saturates to infinity,
* which is what a number that big would look like on screen either way.
*/
inline u16 packHalf(f32 value) {
u32 bits = 0;
std::memcpy(&bits, &value, sizeof(bits));
const u32 sign = (bits >> 16) & 0x8000u;
const i32 exponent = static_cast<i32>((bits >> 23) & 0xFFu) - 127 + 15;
const u32 mantissa = bits & 0x7FFFFFu;
if (exponent >= 31) return static_cast<u16>(sign | 0x7C00u);
if (exponent <= 0) {
if (exponent < -10) return static_cast<u16>(sign);
const u32 shift = static_cast<u32>(14 - exponent);
const u32 subnormal = (mantissa | 0x800000u) >> shift;
return static_cast<u16>(sign | (subnormal + ((mantissa >> (shift - 1)) & 1u)));
}
const u32 half = sign | (static_cast<u32>(exponent) << 10) | (mantissa >> 13);
return static_cast<u16>(half + ((mantissa & 0x1FFFu) > 0x1000u ? 1u : 0u));
}
} // namespace esengine