From: Alex Bennée <alex.ben...@linaro.org> For float16 ARM supports an alternative half-precision format which sacrifices the ability to represent NaN/Inf in return for a higher dynamic range. To support this I've added an additional FloatFmt (float16_params_ahp).
The new FloatFmt flag (arm_althp) is then used to modify the behaviour of canonicalize and round_canonical with respect to representation and exception raising. Finally the float16_to_floatN and floatN_to_float16 conversion routines select the new alternative FloatFmt when !ieee. Signed-off-by: Alex Bennée <alex.ben...@linaro.org> Signed-off-by: Richard Henderson <richard.hender...@linaro.org> --- v3 - squash NaN to 0 if destination is AHP F16 v4 - handle inf -> ahp max in float_to_float not round_canonical - assert no nan and inf for ahp in round_canonical - check ahp before snan in float_to_float --- fpu/softfloat.c | 95 +++++++++++++++++++++++++++++++++++++++++-------- 1 file changed, 81 insertions(+), 14 deletions(-) diff --git a/fpu/softfloat.c b/fpu/softfloat.c index aa219223ff..15a272759d 100644 --- a/fpu/softfloat.c +++ b/fpu/softfloat.c @@ -211,8 +211,10 @@ typedef struct { * frac_shift: shift to normalise the fraction with DECOMPOSED_BINARY_POINT * The following are computed based the size of fraction * frac_lsb: least significant bit of fraction - * fram_lsbm1: the bit bellow the least significant bit (for rounding) + * frac_lsbm1: the bit bellow the least significant bit (for rounding) * round_mask/roundeven_mask: masks used for rounding + * The following optional modifiers are available: + * arm_althp: handle ARM Alternative Half Precision */ typedef struct { int exp_size; @@ -224,6 +226,7 @@ typedef struct { uint64_t frac_lsbm1; uint64_t round_mask; uint64_t roundeven_mask; + bool arm_althp; } FloatFmt; /*---------------------------------------------------------------------------- @@ -252,6 +255,11 @@ static const FloatFmt float16_params = { FLOAT_PARAMS(5, 10) }; +static const FloatFmt float16_params_ahp = { + FLOAT_PARAMS(5, 10), + .arm_althp = true +}; + static const FloatFmt float32_params = { FLOAT_PARAMS(8, 23) }; @@ -315,7 +323,7 @@ static inline float64 float64_pack_raw(FloatParts p) static FloatParts canonicalize(FloatParts part, const FloatFmt *parm, float_status *status) { - if (part.exp == parm->exp_max) { + if (part.exp == parm->exp_max && !parm->arm_althp) { if (part.frac == 0) { part.cls = float_class_inf; } else { @@ -404,7 +412,15 @@ static FloatParts round_canonical(FloatParts p, float_status *s, } frac >>= frac_shift; - if (unlikely(exp >= exp_max)) { + if (parm->arm_althp) { + /* ARM Alt HP eschews Inf and NaN for a wider exponent. */ + if (unlikely(exp > exp_max)) { + /* Overflow. Return the maximum normal. */ + flags = float_flag_invalid; + exp = exp_max; + frac = -1; + } + } else if (unlikely(exp >= exp_max)) { flags |= float_flag_overflow | float_flag_inexact; if (overflow_norm) { exp = exp_max - 1; @@ -455,12 +471,14 @@ static FloatParts round_canonical(FloatParts p, float_status *s, case float_class_inf: do_inf: + assert(!parm->arm_althp); exp = exp_max; frac = 0; break; case float_class_qnan: case float_class_snan: + assert(!parm->arm_althp); exp = exp_max; frac >>= parm->frac_shift; break; @@ -475,14 +493,27 @@ static FloatParts round_canonical(FloatParts p, float_status *s, return p; } +/* Explicit FloatFmt version */ +static FloatParts float16a_unpack_canonical(float16 f, float_status *s, + const FloatFmt *params) +{ + return canonicalize(float16_unpack_raw(f), params, s); +} + static FloatParts float16_unpack_canonical(float16 f, float_status *s) { - return canonicalize(float16_unpack_raw(f), &float16_params, s); + return float16a_unpack_canonical(f, s, &float16_params); +} + +static float16 float16a_round_pack_canonical(FloatParts p, float_status *s, + const FloatFmt *params) +{ + return float16_pack_raw(round_canonical(p, s, params)); } static float16 float16_round_pack_canonical(FloatParts p, float_status *s) { - return float16_pack_raw(round_canonical(p, s, &float16_params)); + return float16a_round_pack_canonical(p, s, &float16_params); } static FloatParts float32_unpack_canonical(float32 f, float_status *s) @@ -1174,7 +1205,33 @@ static FloatParts float_to_float(FloatParts a, const FloatFmt *srcf, const FloatFmt *dstf, float_status *s) { - if (is_nan(a.cls)) { + if (dstf->arm_althp) { + switch (a.cls) { + case float_class_qnan: + case float_class_snan: + /* There is no NaN in the destination format. Raise Invalid + * and return a zero with the sign of the input NaN. + */ + s->float_exception_flags |= float_flag_invalid; + a.cls = float_class_zero; + a.frac = 0; + a.exp = 0; + break; + + case float_class_inf: + /* There is no Inf in the destination format. Raise Invalid + * and return the maximum normal with the correct sign. + */ + s->float_exception_flags |= float_flag_invalid; + a.cls = float_class_normal; + a.exp = dstf->exp_max; + a.frac = ((1ull << dstf->frac_size) - 1) << dstf->frac_shift; + break; + + default: + break; + } + } else if (is_nan(a.cls)) { if (is_snan(a.cls)) { s->float_exception_flags |= float_flag_invalid; a = parts_silence_nan(a, s); @@ -1186,25 +1243,34 @@ static FloatParts float_to_float(FloatParts a, return a; } +/* + * Currently non-ieee implies ARM Alternative Half Precision handling + * for float16 values. If more are needed we'll need to expand the API + * into softfloat. + */ + float32 float16_to_float32(float16 a, bool ieee, float_status *s) { - FloatParts p = float16_unpack_canonical(a, s); - FloatParts pr = float_to_float(p, &float16_params, &float32_params, s); + const FloatFmt *fmt16 = ieee ? &float16_params : &float16_params_ahp; + FloatParts p = float16a_unpack_canonical(a, s, fmt16); + FloatParts pr = float_to_float(p, fmt16, &float32_params, s); return float32_round_pack_canonical(pr, s); } float64 float16_to_float64(float16 a, bool ieee, float_status *s) { - FloatParts p = float16_unpack_canonical(a, s); - FloatParts pr = float_to_float(p, &float16_params, &float64_params, s); + const FloatFmt *fmt16 = ieee ? &float16_params : &float16_params_ahp; + FloatParts p = float16a_unpack_canonical(a, s, fmt16); + FloatParts pr = float_to_float(p, fmt16, &float64_params, s); return float64_round_pack_canonical(pr, s); } float16 float32_to_float16(float32 a, bool ieee, float_status *s) { + const FloatFmt *fmt16 = ieee ? &float16_params : &float16_params_ahp; FloatParts p = float32_unpack_canonical(a, s); - FloatParts pr = float_to_float(p, &float32_params, &float16_params, s); - return float16_round_pack_canonical(pr, s); + FloatParts pr = float_to_float(p, &float32_params, fmt16, s); + return float16a_round_pack_canonical(pr, s, fmt16); } float64 float32_to_float64(float32 a, float_status *s) @@ -1216,9 +1282,10 @@ float64 float32_to_float64(float32 a, float_status *s) float16 float64_to_float16(float64 a, bool ieee, float_status *s) { + const FloatFmt *fmt16 = ieee ? &float16_params : &float16_params_ahp; FloatParts p = float64_unpack_canonical(a, s); - FloatParts pr = float_to_float(p, &float64_params, &float16_params, s); - return float16_round_pack_canonical(pr, s); + FloatParts pr = float_to_float(p, &float64_params, fmt16, s); + return float16a_round_pack_canonical(pr, s, fmt16); } float32 float64_to_float32(float64 a, float_status *s) -- 2.17.0