SmallDouble 提供了一种压缩方案，将小浮点数压缩到8bit.
uint8_t SmallDouble::doubleToByte(double f) { if (f < 0.0) { // round negatives up to zero f = 0.0; } if (f == 0.0) { // zero is a special case return 0; } int32_t bits = MiscUtils::doubleToIntBits(f); int32_t mantissa = (bits & 0xffffff) >> 21; int32_t exponent = (((bits >> 24) & 0x7f) - 63) + 15; if (exponent > 31) { // overflow: use max value exponent = 31; mantissa = 7; } if (exponent < 0) { // underflow: use min value exponent = 0; mantissa = 1; } return (uint8_t)((exponent << 3) | mantissa); // pack into a uint8_t } double SmallDouble::byteToDouble(uint8_t b) { if (b == 0) { // zero is a special case return 0.0; } int32_t mantissa = b & 7; int32_t exponent = (b >> 3) & 31; int32_t bits = ((exponent + (63 - 15)) << 24) | (mantissa << 21); return MiscUtils::intBitsToDouble(bits); }
