Skip to content

Commit 77cc4a0

Browse files
JishinMasterJishinMaster
committed
implemented RISCV pol2cart2D & cart2pol2D
1 parent 1fac94d commit 77cc4a0

16 files changed

+479
-254
lines changed

‎README.md

+8-8
Original file line numberDiff line numberDiff line change
@@ -150,13 +150,13 @@ The following table is a work in progress, "?" means there is not yet an impleme
150150
| floorXf (a) | floorf_C | ippsFloor_32f | floorf_vec |
151151
| truncXf (a) | truncf_C | ippsTrunc_32f | truncf_vec |
152152
| modfXf (a) | modff_C | ippsModf_32f | modf_vec |
153-
| cplxvecmulXf (a) | cplxvecmul_C/precise | ippsMul_32fc_A11/24 | cplxvecmul_vec |
154-
| cplxvecmulXf_split (a) | cplxvecmul_C_split/precise | ? | cplxvecmul_vec_split |
155-
| cplxconjvecmulXf (a) | cplxconjvecmul_C | ippsMulByConj_32fc_A24 | ? |
156-
| cplxconjvecmulXf_split | cplxconjvecmul_C_split | ? | ? |
153+
| cplxvecmulXf (a) | cplxvecmul_C/precise | ippsMul_32fc_A11/24 | cplxvecmulf_vec |
154+
| cplxvecmulXf_split (a) | cplxvecmul_C_split/precise | ? | cplxvecmulf_vec_split |
155+
| cplxconjvecmulXf (a) | cplxconjvecmul_C | ippsMulByConj_32fc_A24 | cplxconjvecmulf_vec |
156+
| cplxconjvecmulXf_split | cplxconjvecmul_C_split | ? | cplxconjvecmulf_vec_split |
157157
| cplxconjXf (a) | cplxconj_C | ippsConj_32fc_A24 | cplxconjf_vec |
158-
| cplxvecdivXf (a) | cplxvecdiv_C | ? | cplxvecdiv_vec |
159-
| cplxvecdivXf_split (a) | cplxvecdiv_C_split | ? | cplxvecdiv_vec_split |
158+
| cplxvecdivXf (a) | cplxvecdiv_C | ? | cplxvecdivf_vec |
159+
| cplxvecdivXf_split (a) | cplxvecdiv_C_split | ? | cplxvecdivf_vec_split |
160160
| setXd | setd_C | ippsSet_64f | setd_vec |
161161
| zeroXd | zerod_C | ippsZero_64f | zerod_vec |
162162
| copyXd | copyd_C | ippsCopy_64f | copyd_vec |
@@ -207,8 +207,8 @@ The following table is a work in progress, "?" means there is not yet an impleme
207207
| sigmoidXf (a) | sigmoidf_C | ? | ? |
208208
| PReluXf (a) | PReluf_C | ? | ? |
209209
| softmaxXf (a) | softmaxf_C | ? | ? |
210-
| pol2cart2DXf (a) | pol2cart2Df_C | ? | ? |
211-
| cart2pol2DXf (a) | cart2pol2Df_C | ? | ? |
210+
| pol2cart2DXf (a) | pol2cart2Df_C | ? | pol2cart2Df_vec |
211+
| cart2pol2DXf (a) | cart2pol2Df_C | ? | cart2pol2Df_vec |
212212
| gatheri_256/512s | gatheri_C | ? | ? |
213213

214214

‎avx512_mathfun.h

100755100644
+6-6
Original file line numberDiff line numberDiff line change
@@ -145,9 +145,9 @@ static inline v16sf sin512_ps(v16sf x)
145145

146146

147147
v16sf swap_sign_bit = _mm512_castsi512_ps(imm0);
148-
148+
149149
#if 1
150-
//Cast integer 0000 FFFF (negative int) to mmask type. Is there a better way?
150+
// Cast integer 0000 FFFF (negative int) to mmask type. Is there a better way?
151151
__mmask16 poly_mask = _mm512_cmplt_epi32_mask(imm2, _mm512_setzero_si512());
152152
#else
153153
v16sf poly_mask = _mm512_castsi512_ps(imm2);
@@ -221,9 +221,9 @@ static inline v16sf cos512_ps(v16sf x)
221221
imm2 = (__m512i) _mm512_maskz_set1_epi32(_mm512_cmpeq_epi32_mask(imm2, *(v16si *) _pi32_512_0), -1);
222222

223223
v16sf sign_bit = _mm512_castsi512_ps(imm0);
224-
224+
225225
#if 1
226-
//Cast integer 0000 FFFF (negative int) to mmask type. Is there a better way?
226+
// Cast integer 0000 FFFF (negative int) to mmask type. Is there a better way?
227227
__mmask16 poly_mask = _mm512_cmplt_epi32_mask(imm2, _mm512_setzero_si512());
228228
#else
229229
v16sf poly_mask = _mm512_castsi512_ps(imm2);
@@ -302,9 +302,9 @@ static inline void sincos512_ps(v16sf x, v16sf *s, v16sf *c)
302302
// v16sf poly_mask = _mm512_castsi512_ps(imm2);
303303

304304
v16sf swap_sign_bit_sin = _mm512_castsi512_ps(imm0);
305-
305+
306306
#if 1
307-
//Cast integer 0000 FFFF (negative int) to mmask type. Is there a better way?
307+
// Cast integer 0000 FFFF (negative int) to mmask type. Is there a better way?
308308
__mmask16 poly_mask = _mm512_cmplt_epi32_mask(imm2, _mm512_setzero_si512());
309309
#else
310310
v16sf poly_mask = _mm512_castsi512_ps(imm2);

0 commit comments

Comments
 (0)