mirrors/qemu
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

arm/helper.c: re-factor recpe and add recepe_f16

Browse Source 
It looks like the ARM ARM has simplified the pseudo code for the
calculation which is done on a fixed point 9 bit integer maths. So
while adding f16 we can also clean this up to be a little less heavy
on the floating point and just return the fractional part and leave
the calle's to do the final packing of the result.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180227143852.11175-23-alex.bennee@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This commit is contained in:
Alex Bennée 2018-03-01 11:05:54 +00:00 committed by Peter Maydell
parent 15f8a233c8
commit 5eb70735af
2 changed files with 129 additions and 98 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

226
target/arm/helper.c
View File

@ -11523,80 +11523,75 @@ float32 HELPER(rsqrts_f32)(float32 a, float32 b, CPUARMState *env)
* int->float conversions at run-time. */ * int->float conversions at run-time. */
#define float64_256 make_float64(0x4070000000000000LL) #define float64_256 make_float64(0x4070000000000000LL)
#define float64_512 make_float64(0x4080000000000000LL) #define float64_512 make_float64(0x4080000000000000LL)
#define float16_maxnorm make_float16(0x7bff)
#define float32_maxnorm make_float32(0x7f7fffff) #define float32_maxnorm make_float32(0x7f7fffff)
#define float64_maxnorm make_float64(0x7fefffffffffffffLL) #define float64_maxnorm make_float64(0x7fefffffffffffffLL)
/* Reciprocal functions /* Reciprocal functions
* *
* The algorithm that must be used to calculate the estimate * The algorithm that must be used to calculate the estimate
* is specified by the ARM ARM, see FPRecipEstimate() * is specified by the ARM ARM, see FPRecipEstimate()/RecipEstimate
*/ */
static float64 recip_estimate(float64 a, float_status *real_fp_status) /* See RecipEstimate()
*
* input is a 9 bit fixed point number
* input range 256 .. 511 for a number from 0.5 <= x < 1.0.
* result range 256 .. 511 for a number from 1.0 to 511/256.
*/
static int recip_estimate(int input)
{ {
/* These calculations mustn't set any fp exception flags, int a, b, r;
* so we use a local copy of the fp_status. assert(256 <= input && input < 512);
*/ a = (input * 2) + 1;
float_status dummy_status = *real_fp_status; b = (1 << 19) / a;
float_status *s = &dummy_status; r = (b + 1) >> 1;
/* q = (int)(a * 512.0) */ assert(256 <= r && r < 512);
float64 q = float64_mul(float64_512, a, s); return r;
int64_t q_int = float64_to_int64_round_to_zero(q, s);
/* r = 1.0 / (((double)q + 0.5) / 512.0) */
q = int64_to_float64(q_int, s);
q = float64_add(q, float64_half, s);
q = float64_div(q, float64_512, s);
q = float64_div(float64_one, q, s);
/* s = (int)(256.0 * r + 0.5) */
q = float64_mul(q, float64_256, s);
q = float64_add(q, float64_half, s);
q_int = float64_to_int64_round_to_zero(q, s);
/* return (double)s / 256.0 */
return float64_div(int64_to_float64(q_int, s), float64_256, s);
} }
/* Common wrapper to call recip_estimate */ /*
static float64 call_recip_estimate(float64 num, int off, float_status *fpst) * Common wrapper to call recip_estimate
{ *
uint64_t val64 = float64_val(num); * The parameters are exponent and 64 bit fraction (without implicit
uint64_t frac = extract64(val64, 0, 52); * bit) where the binary point is nominally at bit 52. Returns a
int64_t exp = extract64(val64, 52, 11); * float64 which can then be rounded to the appropriate size by the
uint64_t sbit; * callee.
float64 scaled, estimate; */
/* Generate the scaled number for the estimate function */ static uint64_t call_recip_estimate(int *exp, int exp_off, uint64_t frac)
if (exp == 0) { {
uint32_t scaled, estimate;
uint64_t result_frac;
int result_exp;
/* Handle sub-normals */
if (*exp == 0) {
if (extract64(frac, 51, 1) == 0) { if (extract64(frac, 51, 1) == 0) {
exp = -1; *exp = -1;
frac = extract64(frac, 0, 50) << 2; frac <<= 2;
} else { } else {
frac = extract64(frac, 0, 51) << 1; frac <<= 1;
} }
} }
/* scaled = '0' : '01111111110' : fraction<51:44> : Zeros(44); */ /* scaled = UInt('1':fraction<51:44>) */
scaled = make_float64((0x3feULL << 52) scaled = deposit32(1 << 8, 0, 8, extract64(frac, 44, 8));
| extract64(frac, 44, 8) << 44); estimate = recip_estimate(scaled);
estimate = recip_estimate(scaled, fpst); result_exp = exp_off - *exp;
result_frac = deposit64(0, 44, 8, estimate);
/* Build new result */ if (result_exp == 0) {
val64 = float64_val(estimate); result_frac = deposit64(result_frac >> 1, 51, 1, 1);
sbit = 0x8000000000000000ULL & val64; } else if (result_exp == -1) {
exp = off - exp; result_frac = deposit64(result_frac >> 2, 50, 2, 1);
frac = extract64(val64, 0, 52); result_exp = 0;
if (exp == 0) {
frac = 1ULL << 51 | extract64(frac, 1, 51);
} else if (exp == -1) {
frac = 1ULL << 50 | extract64(frac, 2, 50);
exp = 0;
} }
return make_float64(sbit | (exp << 52) | frac); *exp = result_exp;
return result_frac;
} }
static bool round_to_inf(float_status *fpst, bool sign_bit) static bool round_to_inf(float_status *fpst, bool sign_bit)
@ -11615,18 +11610,63 @@ static bool round_to_inf(float_status *fpst, bool sign_bit)
g_assert_not_reached(); g_assert_not_reached();
} }
float16 HELPER(recpe_f16)(float16 input, void *fpstp)
{
float_status *fpst = fpstp;
float16 f16 = float16_squash_input_denormal(input, fpst);
uint32_t f16_val = float16_val(f16);
uint32_t f16_sign = float16_is_neg(f16);
int f16_exp = extract32(f16_val, 10, 5);
uint32_t f16_frac = extract32(f16_val, 0, 10);
uint64_t f64_frac;
if (float16_is_any_nan(f16)) {
float16 nan = f16;
if (float16_is_signaling_nan(f16, fpst)) {
float_raise(float_flag_invalid, fpst);
nan = float16_maybe_silence_nan(f16, fpst);
}
if (fpst->default_nan_mode) {
nan = float16_default_nan(fpst);
}
return nan;
} else if (float16_is_infinity(f16)) {
return float16_set_sign(float16_zero, float16_is_neg(f16));
} else if (float16_is_zero(f16)) {
float_raise(float_flag_divbyzero, fpst);
return float16_set_sign(float16_infinity, float16_is_neg(f16));
} else if (float16_abs(f16) < (1 << 8)) {
/* Abs(value) < 2.0^-16 */
float_raise(float_flag_overflow | float_flag_inexact, fpst);
if (round_to_inf(fpst, f16_sign)) {
return float16_set_sign(float16_infinity, f16_sign);
} else {
return float16_set_sign(float16_maxnorm, f16_sign);
}
} else if (f16_exp >= 29 && fpst->flush_to_zero) {
float_raise(float_flag_underflow, fpst);
return float16_set_sign(float16_zero, float16_is_neg(f16));
}
f64_frac = call_recip_estimate(&f16_exp, 29,
((uint64_t) f16_frac) << (52 - 10));
/* result = sign : result_exp<4:0> : fraction<51:42> */
f16_val = deposit32(0, 15, 1, f16_sign);
f16_val = deposit32(f16_val, 10, 5, f16_exp);
f16_val = deposit32(f16_val, 0, 10, extract64(f64_frac, 52 - 10, 10));
return make_float16(f16_val);
}
float32 HELPER(recpe_f32)(float32 input, void *fpstp) float32 HELPER(recpe_f32)(float32 input, void *fpstp)
{ {
float_status *fpst = fpstp; float_status *fpst = fpstp;
float32 f32 = float32_squash_input_denormal(input, fpst); float32 f32 = float32_squash_input_denormal(input, fpst);
uint32_t f32_val = float32_val(f32); uint32_t f32_val = float32_val(f32);
uint32_t f32_sbit = 0x80000000ULL & f32_val; bool f32_sign = float32_is_neg(f32);
int32_t f32_exp = extract32(f32_val, 23, 8); int f32_exp = extract32(f32_val, 23, 8);
uint32_t f32_frac = extract32(f32_val, 0, 23); uint32_t f32_frac = extract32(f32_val, 0, 23);
float64 f64, r64; uint64_t f64_frac;
uint64_t r64_val;
int64_t r64_exp;
uint64_t r64_frac;
if (float32_is_any_nan(f32)) { if (float32_is_any_nan(f32)) {
float32 nan = f32; float32 nan = f32;
@ -11643,30 +11683,27 @@ float32 HELPER(recpe_f32)(float32 input, void *fpstp)
} else if (float32_is_zero(f32)) { } else if (float32_is_zero(f32)) {
float_raise(float_flag_divbyzero, fpst); float_raise(float_flag_divbyzero, fpst);
return float32_set_sign(float32_infinity, float32_is_neg(f32)); return float32_set_sign(float32_infinity, float32_is_neg(f32));
} else if ((f32_val & ~(1ULL << 31)) < (1ULL << 21)) { } else if (float32_abs(f32) < (1ULL << 21)) {
/* Abs(value) < 2.0^-128 */ /* Abs(value) < 2.0^-128 */
float_raise(float_flag_overflow | float_flag_inexact, fpst); float_raise(float_flag_overflow | float_flag_inexact, fpst);
if (round_to_inf(fpst, f32_sbit)) { if (round_to_inf(fpst, f32_sign)) {
return float32_set_sign(float32_infinity, float32_is_neg(f32)); return float32_set_sign(float32_infinity, f32_sign);
} else { } else {
return float32_set_sign(float32_maxnorm, float32_is_neg(f32)); return float32_set_sign(float32_maxnorm, f32_sign);
} }
} else if (f32_exp >= 253 && fpst->flush_to_zero) { } else if (f32_exp >= 253 && fpst->flush_to_zero) {
float_raise(float_flag_underflow, fpst); float_raise(float_flag_underflow, fpst);
return float32_set_sign(float32_zero, float32_is_neg(f32)); return float32_set_sign(float32_zero, float32_is_neg(f32));
} }
f64_frac = call_recip_estimate(&f32_exp, 253,
((uint64_t) f32_frac) << (52 - 23));
f64 = make_float64(((int64_t)(f32_exp) << 52) | (int64_t)(f32_frac) << 29); /* result = sign : result_exp<7:0> : fraction<51:29> */
r64 = call_recip_estimate(f64, 253, fpst); f32_val = deposit32(0, 31, 1, f32_sign);
r64_val = float64_val(r64); f32_val = deposit32(f32_val, 23, 8, f32_exp);
r64_exp = extract64(r64_val, 52, 11); f32_val = deposit32(f32_val, 0, 23, extract64(f64_frac, 52 - 23, 23));
r64_frac = extract64(r64_val, 0, 52); return make_float32(f32_val);
/* result = sign : result_exp<7:0> : fraction<51:29>; */
return make_float32(f32_sbit |
(r64_exp & 0xff) << 23 |
extract64(r64_frac, 29, 24));
} }
float64 HELPER(recpe_f64)(float64 input, void *fpstp) float64 HELPER(recpe_f64)(float64 input, void *fpstp)
@ -11674,12 +11711,9 @@ float64 HELPER(recpe_f64)(float64 input, void *fpstp)
float_status *fpst = fpstp; float_status *fpst = fpstp;
float64 f64 = float64_squash_input_denormal(input, fpst); float64 f64 = float64_squash_input_denormal(input, fpst);
uint64_t f64_val = float64_val(f64); uint64_t f64_val = float64_val(f64);
uint64_t f64_sbit = 0x8000000000000000ULL & f64_val; bool f64_sign = float64_is_neg(f64);
int64_t f64_exp = extract64(f64_val, 52, 11); int f64_exp = extract64(f64_val, 52, 11);
float64 r64; uint64_t f64_frac = extract64(f64_val, 0, 52);
uint64_t r64_val;
int64_t r64_exp;
uint64_t r64_frac;
/* Deal with any special cases */ /* Deal with any special cases */
if (float64_is_any_nan(f64)) { if (float64_is_any_nan(f64)) {
@ -11700,25 +11734,23 @@ float64 HELPER(recpe_f64)(float64 input, void *fpstp)
} else if ((f64_val & ~(1ULL << 63)) < (1ULL << 50)) { } else if ((f64_val & ~(1ULL << 63)) < (1ULL << 50)) {
/* Abs(value) < 2.0^-1024 */ /* Abs(value) < 2.0^-1024 */
float_raise(float_flag_overflow | float_flag_inexact, fpst); float_raise(float_flag_overflow | float_flag_inexact, fpst);
if (round_to_inf(fpst, f64_sbit)) { if (round_to_inf(fpst, f64_sign)) {
return float64_set_sign(float64_infinity, float64_is_neg(f64)); return float64_set_sign(float64_infinity, f64_sign);
} else { } else {
return float64_set_sign(float64_maxnorm, float64_is_neg(f64)); return float64_set_sign(float64_maxnorm, f64_sign);
} }
} else if (f64_exp >= 2045 && fpst->flush_to_zero) { } else if (f64_exp >= 2045 && fpst->flush_to_zero) {
float_raise(float_flag_underflow, fpst); float_raise(float_flag_underflow, fpst);
return float64_set_sign(float64_zero, float64_is_neg(f64)); return float64_set_sign(float64_zero, float64_is_neg(f64));
} }
r64 = call_recip_estimate(f64, 2045, fpst); f64_frac = call_recip_estimate(&f64_exp, 2045, f64_frac);
r64_val = float64_val(r64);
r64_exp = extract64(r64_val, 52, 11);
r64_frac = extract64(r64_val, 0, 52);
/* result = sign : result_exp<10:0> : fraction<51:0> */ /* result = sign : result_exp<10:0> : fraction<51:0>; */
return make_float64(f64_sbit | f64_val = deposit64(0, 63, 1, f64_sign);
((r64_exp & 0x7ff) << 52) | f64_val = deposit64(f64_val, 52, 11, f64_exp);
r64_frac); f64_val = deposit64(f64_val, 0, 52, f64_frac);
return make_float64(f64_val);
} }
/* The algorithm that must be used to calculate the estimate /* The algorithm that must be used to calculate the estimate
@ -11907,19 +11939,17 @@ float64 HELPER(rsqrte_f64)(float64 input, void *fpstp)
uint32_t HELPER(recpe_u32)(uint32_t a, void *fpstp) uint32_t HELPER(recpe_u32)(uint32_t a, void *fpstp)
{ {
float_status *s = fpstp; /* float_status *s = fpstp; */
float64 f64; int input, estimate;
if ((a & 0x80000000) == 0) { if ((a & 0x80000000) == 0) {
return 0xffffffff; return 0xffffffff;
} }
f64 = make_float64((0x3feULL << 52) input = extract32(a, 23, 9);
| ((int64_t)(a & 0x7fffffff) << 21)); estimate = recip_estimate(input);
f64 = recip_estimate(f64, s); return deposit32(0, (32 - 9), 9, estimate);
return 0x80000000 | ((float64_val(f64) >> 21) & 0x7fffffff);
} }
uint32_t HELPER(rsqrte_u32)(uint32_t a, void *fpstp) uint32_t HELPER(rsqrte_u32)(uint32_t a, void *fpstp)

1
target/arm/helper.h
View File

@ -192,6 +192,7 @@ DEF_HELPER_4(vfp_muladds, f32, f32, f32, f32, ptr)
DEF_HELPER_3(recps_f32, f32, f32, f32, env) DEF_HELPER_3(recps_f32, f32, f32, f32, env)
DEF_HELPER_3(rsqrts_f32, f32, f32, f32, env) DEF_HELPER_3(rsqrts_f32, f32, f32, f32, env)
DEF_HELPER_FLAGS_2(recpe_f16, TCG_CALL_NO_RWG, f16, f16, ptr)
DEF_HELPER_FLAGS_2(recpe_f32, TCG_CALL_NO_RWG, f32, f32, ptr) DEF_HELPER_FLAGS_2(recpe_f32, TCG_CALL_NO_RWG, f32, f32, ptr)
DEF_HELPER_FLAGS_2(recpe_f64, TCG_CALL_NO_RWG, f64, f64, ptr) DEF_HELPER_FLAGS_2(recpe_f64, TCG_CALL_NO_RWG, f64, f64, ptr)
DEF_HELPER_FLAGS_2(rsqrte_f32, TCG_CALL_NO_RWG, f32, f32, ptr) DEF_HELPER_FLAGS_2(rsqrte_f32, TCG_CALL_NO_RWG, f32, f32, ptr)