diff options
author | Paul Brook <paul@codesourcery.com> | 2009-11-19 16:45:20 +0000 |
---|---|---|
committer | Paul Brook <paul@codesourcery.com> | 2009-11-19 16:45:20 +0000 |
commit | 600114988cb1beede13ce29dec65398f3e62e510 (patch) | |
tree | 937055725aba7c29b402b927066333e2f8debf61 /fpu | |
parent | f165b53a896975f6c45760d72f87e44f1a2ba325 (diff) | |
download | qemu-600114988cb1beede13ce29dec65398f3e62e510.zip qemu-600114988cb1beede13ce29dec65398f3e62e510.tar.gz qemu-600114988cb1beede13ce29dec65398f3e62e510.tar.bz2 |
ARM FP16 support
Implement the ARM VFP half precision floating point extensions.
Signed-off-by: Paul Brook <paul@codesourcery.com>
Diffstat (limited to 'fpu')
-rw-r--r-- | fpu/softfloat.c | 138 | ||||
-rw-r--r-- | fpu/softfloat.h | 6 |
2 files changed, 144 insertions, 0 deletions
diff --git a/fpu/softfloat.c b/fpu/softfloat.c index 4d58744..395f9b1 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -2457,6 +2457,144 @@ float32 float64_to_float32( float64 a STATUS_PARAM ) } + +/*---------------------------------------------------------------------------- +| Packs the sign `zSign', exponent `zExp', and significand `zSig' into a +| half-precision floating-point value, returning the result. After being +| shifted into the proper positions, the three fields are simply added +| together to form the result. This means that any integer portion of `zSig' +| will be added into the exponent. Since a properly normalized significand +| will have an integer portion equal to 1, the `zExp' input should be 1 less +| than the desired result exponent whenever `zSig' is a complete, normalized +| significand. +*----------------------------------------------------------------------------*/ +static bits16 packFloat16(flag zSign, int16 zExp, bits16 zSig) +{ + return (((bits32)zSign) << 15) + (((bits32)zExp) << 10) + zSig; +} + +/* Half precision floats come in two formats: standard IEEE and "ARM" format. + The latter gains extra exponent range by omitting the NaN/Inf encodings. */ + +float32 float16_to_float32( bits16 a, flag ieee STATUS_PARAM ) +{ + flag aSign; + int16 aExp; + bits32 aSig; + + aSign = a >> 15; + aExp = (a >> 10) & 0x1f; + aSig = a & 0x3ff; + + if (aExp == 0x1f && ieee) { + if (aSig) { + /* Make sure correct exceptions are raised. */ + float32ToCommonNaN(a STATUS_VAR); + aSig |= 0x200; + } + return packFloat32(aSign, 0xff, aSig << 13); + } + if (aExp == 0) { + int8 shiftCount; + + if (aSig == 0) { + return packFloat32(aSign, 0, 0); + } + + shiftCount = countLeadingZeros32( aSig ) - 21; + aSig = aSig << shiftCount; + aExp = -shiftCount; + } + return packFloat32( aSign, aExp + 0x70, aSig << 13); +} + +bits16 float32_to_float16( float32 a, flag ieee STATUS_PARAM) +{ + flag aSign; + int16 aExp; + bits32 aSig; + bits32 mask; + bits32 increment; + int8 roundingMode; + + aSig = extractFloat32Frac( a ); + aExp = extractFloat32Exp( a ); + aSign = extractFloat32Sign( a ); + if ( aExp == 0xFF ) { + if (aSig) { + /* Make sure correct exceptions are raised. */ + float32ToCommonNaN(a STATUS_VAR); + aSig |= 0x00400000; + } + return packFloat16(aSign, 0x1f, aSig >> 13); + } + if (aExp == 0 && aSign == 0) { + return packFloat16(aSign, 0, 0); + } + /* Decimal point between bits 22 and 23. */ + aSig |= 0x00800000; + aExp -= 0x7f; + if (aExp < -14) { + mask = 0x007fffff; + if (aExp < -24) { + aExp = -25; + } else { + mask >>= 24 + aExp; + } + } else { + mask = 0x00001fff; + } + if (aSig & mask) { + float_raise( float_flag_underflow STATUS_VAR ); + roundingMode = STATUS(float_rounding_mode); + switch (roundingMode) { + case float_round_nearest_even: + increment = (mask + 1) >> 1; + if ((aSig & mask) == increment) { + increment = aSig & (increment << 1); + } + break; + case float_round_up: + increment = aSign ? 0 : mask; + break; + case float_round_down: + increment = aSign ? mask : 0; + break; + default: /* round_to_zero */ + increment = 0; + break; + } + aSig += increment; + if (aSig >= 0x01000000) { + aSig >>= 1; + aExp++; + } + } else if (aExp < -14 + && STATUS(float_detect_tininess) == float_tininess_before_rounding) { + float_raise( float_flag_underflow STATUS_VAR); + } + + if (ieee) { + if (aExp > 15) { + float_raise( float_flag_overflow | float_flag_inexact STATUS_VAR); + return packFloat16(aSign, 0x1f, 0); + } + } else { + if (aExp > 16) { + float_raise( float_flag_overflow | float_flag_inexact STATUS_VAR); + return packFloat16(aSign, 0x1f, 0x3ff); + } + } + if (aExp < -24) { + return packFloat16(aSign, 0, 0); + } + if (aExp < -14) { + aSig >>= -14 - aExp; + aExp = -14; + } + return packFloat16(aSign, aExp + 14, aSig >> 13); +} + #ifdef FLOATX80 /*---------------------------------------------------------------------------- diff --git a/fpu/softfloat.h b/fpu/softfloat.h index 789179a..636591b 100644 --- a/fpu/softfloat.h +++ b/fpu/softfloat.h @@ -243,6 +243,12 @@ float128 int64_to_float128( int64_t STATUS_PARAM ); #endif /*---------------------------------------------------------------------------- +| Software half-precision conversion routines. +*----------------------------------------------------------------------------*/ +bits16 float32_to_float16( float32, flag STATUS_PARAM ); +float32 float16_to_float32( bits16, flag STATUS_PARAM ); + +/*---------------------------------------------------------------------------- | Software IEC/IEEE single-precision conversion routines. *----------------------------------------------------------------------------*/ int float32_to_int32( float32 STATUS_PARAM ); |