Skip to content

Commit

Permalink
target/arm: Implement SVE2 fp multiply-add long
Browse files Browse the repository at this point in the history 
Implements both vectored and indexed FMLALB, FMLALT, FMLSLB, FMLSLT

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Stephen Long <steplong@quicinc.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20210525010358.152808-83-richard.henderson@linaro.org
Message-Id: <20200504171240.11220-1-steplong@quicinc.com>
[rth: Rearrange to use float16_to_float32_by_bits.]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
  • Loading branch information
@steplong @pm215
steplong authored and pm215 committed May 25, 2021
1 parent 93966af commit 50d102b
Show file tree
Hide file tree
Showing 4 changed files with 141 additions and 0 deletions.
5 changes: 5 additions & 0 deletions target/arm/helper.h
View file
Expand Up @@ -986,6 +986,11 @@ DEF_HELPER_FLAGS_4(sve2_sqrdmulh_idx_s, TCG_CALL_NO_RWG,
DEF_HELPER_FLAGS_4(sve2_sqrdmulh_idx_d, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, i32)

DEF_HELPER_FLAGS_6(sve2_fmlal_zzzw_s, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_6(sve2_fmlal_zzxw_s, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, ptr, i32)

DEF_HELPER_FLAGS_4(gvec_xar_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)

#ifdef TARGET_AARCH64
Expand Down
14 changes: 14 additions & 0 deletions target/arm/sve.decode
View file
Expand Up @@ -132,6 +132,8 @@
&rrrr_esz ra=%reg_movprfx

# Four operand with unused vector element size
@rda_rn_rm_e0 ........ ... rm:5 ... ... rn:5 rd:5 \
&rrrr_esz esz=0 ra=%reg_movprfx
@rdn_ra_rm_e0 ........ ... rm:5 ... ... ra:5 rd:5 \
&rrrr_esz esz=0 rn=%reg_movprfx

Expand Down Expand Up @@ -1608,3 +1610,15 @@ FCVTLT_sd 01100100 11 0010 11 101 ... ..... ..... @rd_pg_rn_e0

### SVE2 floating-point convert to integer
FLOGB 01100101 00 011 esz:2 0101 pg:3 rn:5 rd:5 &rpr_esz

### SVE2 floating-point multiply-add long (vectors)
FMLALB_zzzw 01100100 10 1 ..... 10 0 00 0 ..... ..... @rda_rn_rm_e0
FMLALT_zzzw 01100100 10 1 ..... 10 0 00 1 ..... ..... @rda_rn_rm_e0
FMLSLB_zzzw 01100100 10 1 ..... 10 1 00 0 ..... ..... @rda_rn_rm_e0
FMLSLT_zzzw 01100100 10 1 ..... 10 1 00 1 ..... ..... @rda_rn_rm_e0

### SVE2 floating-point multiply-add long (indexed)
FMLALB_zzxw 01100100 10 1 ..... 0100.0 ..... ..... @rrxr_3a esz=2
FMLALT_zzxw 01100100 10 1 ..... 0100.1 ..... ..... @rrxr_3a esz=2
FMLSLB_zzxw 01100100 10 1 ..... 0110.0 ..... ..... @rrxr_3a esz=2
FMLSLT_zzxw 01100100 10 1 ..... 0110.1 ..... ..... @rrxr_3a esz=2
75 changes: 75 additions & 0 deletions target/arm/translate-sve.c
View file
Expand Up @@ -8535,3 +8535,78 @@ static bool trans_FLOGB(DisasContext *s, arg_rpr_esz *a)
}
return true;
}

static bool do_FMLAL_zzzw(DisasContext *s, arg_rrrr_esz *a, bool sub, bool sel)
{
if (!dc_isar_feature(aa64_sve2, s)) {
return false;
}
if (sve_access_check(s)) {
unsigned vsz = vec_full_reg_size(s);
tcg_gen_gvec_4_ptr(vec_full_reg_offset(s, a->rd),
vec_full_reg_offset(s, a->rn),
vec_full_reg_offset(s, a->rm),
vec_full_reg_offset(s, a->ra),
cpu_env, vsz, vsz, (sel << 1) | sub,
gen_helper_sve2_fmlal_zzzw_s);
}
return true;
}

static bool trans_FMLALB_zzzw(DisasContext *s, arg_rrrr_esz *a)
{
return do_FMLAL_zzzw(s, a, false, false);
}

static bool trans_FMLALT_zzzw(DisasContext *s, arg_rrrr_esz *a)
{
return do_FMLAL_zzzw(s, a, false, true);
}

static bool trans_FMLSLB_zzzw(DisasContext *s, arg_rrrr_esz *a)
{
return do_FMLAL_zzzw(s, a, true, false);
}

static bool trans_FMLSLT_zzzw(DisasContext *s, arg_rrrr_esz *a)
{
return do_FMLAL_zzzw(s, a, true, true);
}

static bool do_FMLAL_zzxw(DisasContext *s, arg_rrxr_esz *a, bool sub, bool sel)
{
if (!dc_isar_feature(aa64_sve2, s)) {
return false;
}
if (sve_access_check(s)) {
unsigned vsz = vec_full_reg_size(s);
tcg_gen_gvec_4_ptr(vec_full_reg_offset(s, a->rd),
vec_full_reg_offset(s, a->rn),
vec_full_reg_offset(s, a->rm),
vec_full_reg_offset(s, a->ra),
cpu_env, vsz, vsz,
(a->index << 2) | (sel << 1) | sub,
gen_helper_sve2_fmlal_zzxw_s);
}
return true;
}

static bool trans_FMLALB_zzxw(DisasContext *s, arg_rrxr_esz *a)
{
return do_FMLAL_zzxw(s, a, false, false);
}

static bool trans_FMLALT_zzxw(DisasContext *s, arg_rrxr_esz *a)
{
return do_FMLAL_zzxw(s, a, false, true);
}

static bool trans_FMLSLB_zzxw(DisasContext *s, arg_rrxr_esz *a)
{
return do_FMLAL_zzxw(s, a, true, false);
}

static bool trans_FMLSLT_zzxw(DisasContext *s, arg_rrxr_esz *a)
{
return do_FMLAL_zzxw(s, a, true, true);
}
47 changes: 47 additions & 0 deletions target/arm/vec_helper.c
View file
Expand Up @@ -1668,6 +1668,27 @@ void HELPER(gvec_fmlal_a64)(void *vd, void *vn, void *vm,
get_flush_inputs_to_zero(&env->vfp.fp_status_f16));
}

void HELPER(sve2_fmlal_zzzw_s)(void *vd, void *vn, void *vm, void *va,
void *venv, uint32_t desc)
{
intptr_t i, oprsz = simd_oprsz(desc);
uint16_t negn = extract32(desc, SIMD_DATA_SHIFT, 1) << 15;
intptr_t sel = extract32(desc, SIMD_DATA_SHIFT + 1, 1) * sizeof(float16);
CPUARMState *env = venv;
float_status *status = &env->vfp.fp_status;
bool fz16 = get_flush_inputs_to_zero(&env->vfp.fp_status_f16);

for (i = 0; i < oprsz; i += sizeof(float32)) {
float16 nn_16 = *(float16 *)(vn + H1_2(i + sel)) ^ negn;
float16 mm_16 = *(float16 *)(vm + H1_2(i + sel));
float32 nn = float16_to_float32_by_bits(nn_16, fz16);
float32 mm = float16_to_float32_by_bits(mm_16, fz16);
float32 aa = *(float32 *)(va + H1_4(i));

*(float32 *)(vd + H1_4(i)) = float32_muladd(nn, mm, aa, 0, status);
}
}

static void do_fmlal_idx(float32 *d, void *vn, void *vm, float_status *fpst,
uint32_t desc, bool fz16)
{
Expand Down Expand Up @@ -1712,6 +1733,32 @@ void HELPER(gvec_fmlal_idx_a64)(void *vd, void *vn, void *vm,
get_flush_inputs_to_zero(&env->vfp.fp_status_f16));
}

void HELPER(sve2_fmlal_zzxw_s)(void *vd, void *vn, void *vm, void *va,
void *venv, uint32_t desc)
{
intptr_t i, j, oprsz = simd_oprsz(desc);
uint16_t negn = extract32(desc, SIMD_DATA_SHIFT, 1) << 15;
intptr_t sel = extract32(desc, SIMD_DATA_SHIFT + 1, 1) * sizeof(float16);
intptr_t idx = extract32(desc, SIMD_DATA_SHIFT + 2, 3) * sizeof(float16);
CPUARMState *env = venv;
float_status *status = &env->vfp.fp_status;
bool fz16 = get_flush_inputs_to_zero(&env->vfp.fp_status_f16);

for (i = 0; i < oprsz; i += 16) {
float16 mm_16 = *(float16 *)(vm + i + idx);
float32 mm = float16_to_float32_by_bits(mm_16, fz16);

for (j = 0; j < 16; j += sizeof(float32)) {
float16 nn_16 = *(float16 *)(vn + H1_2(i + j + sel)) ^ negn;
float32 nn = float16_to_float32_by_bits(nn_16, fz16);
float32 aa = *(float32 *)(va + H1_4(i + j));

*(float32 *)(vd + H1_4(i + j)) =
float32_muladd(nn, mm, aa, 0, status);
}
}
}

void HELPER(gvec_sshl_b)(void *vd, void *vn, void *vm, uint32_t desc)
{
intptr_t i, opr_sz = simd_oprsz(desc);
Expand Down

0 comments on commit 50d102b

Please sign in to comment.