/external/XNNPACK/src/qs8-requantization/ |
D | q31-wasmsimd.c | 79 … const v128_t x_product_lo = wasm_i64x2_add(wasm_i64x2_mul(x_lo, vmultiplier), vtwice_q31rounding); in xnn_qs8_requantize_q31__wasmsimd() 80 … const v128_t y_product_lo = wasm_i64x2_add(wasm_i64x2_mul(y_lo, vmultiplier), vtwice_q31rounding); in xnn_qs8_requantize_q31__wasmsimd() 81 … const v128_t z_product_lo = wasm_i64x2_add(wasm_i64x2_mul(z_lo, vmultiplier), vtwice_q31rounding); in xnn_qs8_requantize_q31__wasmsimd() 82 … const v128_t w_product_lo = wasm_i64x2_add(wasm_i64x2_mul(w_lo, vmultiplier), vtwice_q31rounding); in xnn_qs8_requantize_q31__wasmsimd() 84 … const v128_t x_product_hi = wasm_i64x2_add(wasm_i64x2_mul(x_hi, vmultiplier), vtwice_q31rounding); in xnn_qs8_requantize_q31__wasmsimd() 85 … const v128_t y_product_hi = wasm_i64x2_add(wasm_i64x2_mul(y_hi, vmultiplier), vtwice_q31rounding); in xnn_qs8_requantize_q31__wasmsimd() 86 … const v128_t z_product_hi = wasm_i64x2_add(wasm_i64x2_mul(z_hi, vmultiplier), vtwice_q31rounding); in xnn_qs8_requantize_q31__wasmsimd() 87 … const v128_t w_product_hi = wasm_i64x2_add(wasm_i64x2_mul(w_hi, vmultiplier), vtwice_q31rounding); in xnn_qs8_requantize_q31__wasmsimd()
|
/external/XNNPACK/src/qu8-requantization/ |
D | q31-wasmsimd.c | 79 … const v128_t x_product_lo = wasm_i64x2_add(wasm_i64x2_mul(x_lo, vmultiplier), vtwice_q31rounding); in xnn_qu8_requantize_q31__wasmsimd() 80 … const v128_t y_product_lo = wasm_i64x2_add(wasm_i64x2_mul(y_lo, vmultiplier), vtwice_q31rounding); in xnn_qu8_requantize_q31__wasmsimd() 81 … const v128_t z_product_lo = wasm_i64x2_add(wasm_i64x2_mul(z_lo, vmultiplier), vtwice_q31rounding); in xnn_qu8_requantize_q31__wasmsimd() 82 … const v128_t w_product_lo = wasm_i64x2_add(wasm_i64x2_mul(w_lo, vmultiplier), vtwice_q31rounding); in xnn_qu8_requantize_q31__wasmsimd() 84 … const v128_t x_product_hi = wasm_i64x2_add(wasm_i64x2_mul(x_hi, vmultiplier), vtwice_q31rounding); in xnn_qu8_requantize_q31__wasmsimd() 85 … const v128_t y_product_hi = wasm_i64x2_add(wasm_i64x2_mul(y_hi, vmultiplier), vtwice_q31rounding); in xnn_qu8_requantize_q31__wasmsimd() 86 … const v128_t z_product_hi = wasm_i64x2_add(wasm_i64x2_mul(z_hi, vmultiplier), vtwice_q31rounding); in xnn_qu8_requantize_q31__wasmsimd() 87 … const v128_t w_product_hi = wasm_i64x2_add(wasm_i64x2_mul(w_hi, vmultiplier), vtwice_q31rounding); in xnn_qu8_requantize_q31__wasmsimd()
|
/external/XNNPACK/src/qs8-gavgpool/gen/ |
D | 7x-minmax-wasmsimd-c24-acc2.c | 147 const v128_t vabsprod01 = wasm_i64x2_mul(vabsacc01, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() 148 const v128_t vabsprod23 = wasm_i64x2_mul(vabsacc23, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() 149 const v128_t vabsprod45 = wasm_i64x2_mul(vabsacc45, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() 150 const v128_t vabsprod67 = wasm_i64x2_mul(vabsacc67, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() 151 const v128_t vabsprod89 = wasm_i64x2_mul(vabsacc89, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() 152 const v128_t vabsprodAB = wasm_i64x2_mul(vabsaccAB, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() 153 const v128_t vabsprodCD = wasm_i64x2_mul(vabsaccCD, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() 154 const v128_t vabsprodEF = wasm_i64x2_mul(vabsaccEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() 155 const v128_t vabsprodGH = wasm_i64x2_mul(vabsaccGH, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() 156 const v128_t vabsprodIJ = wasm_i64x2_mul(vabsaccIJ, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() [all …]
|
D | 7x-minmax-wasmsimd-c16-acc2.c | 124 const v128_t vabsprod01 = wasm_i64x2_mul(vabsacc01, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() 125 const v128_t vabsprod23 = wasm_i64x2_mul(vabsacc23, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() 126 const v128_t vabsprod45 = wasm_i64x2_mul(vabsacc45, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() 127 const v128_t vabsprod67 = wasm_i64x2_mul(vabsacc67, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() 128 const v128_t vabsprod89 = wasm_i64x2_mul(vabsacc89, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() 129 const v128_t vabsprodAB = wasm_i64x2_mul(vabsaccAB, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() 130 const v128_t vabsprodCD = wasm_i64x2_mul(vabsaccCD, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() 131 const v128_t vabsprodEF = wasm_i64x2_mul(vabsaccEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() 206 const v128_t vabsprod01 = wasm_i64x2_mul(vabsacc01, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() 207 const v128_t vabsprod23 = wasm_i64x2_mul(vabsacc23, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() [all …]
|
D | 7x-minmax-wasmsimd-c8-acc2.c | 101 const v128_t vabsprod01 = wasm_i64x2_mul(vabsacc01, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c8_acc2() 102 const v128_t vabsprod23 = wasm_i64x2_mul(vabsacc23, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c8_acc2() 103 const v128_t vabsprod45 = wasm_i64x2_mul(vabsacc45, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c8_acc2() 104 const v128_t vabsprod67 = wasm_i64x2_mul(vabsacc67, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c8_acc2() 170 const v128_t vabsprod01 = wasm_i64x2_mul(vabsacc01, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c8_acc2() 171 const v128_t vabsprod23 = wasm_i64x2_mul(vabsacc23, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c8_acc2() 172 const v128_t vabsprod45 = wasm_i64x2_mul(vabsacc45, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c8_acc2() 173 const v128_t vabsprod67 = wasm_i64x2_mul(vabsacc67, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c8_acc2()
|
D | 7p7x-minmax-wasmsimd-c16-acc2.c | 251 const v128_t vabsprod01 = wasm_i64x2_mul(vabsacc01, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() 252 const v128_t vabsprod23 = wasm_i64x2_mul(vabsacc23, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() 253 const v128_t vabsprod45 = wasm_i64x2_mul(vabsacc45, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() 254 const v128_t vabsprod67 = wasm_i64x2_mul(vabsacc67, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() 255 const v128_t vabsprod89 = wasm_i64x2_mul(vabsacc89, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() 256 const v128_t vabsprodAB = wasm_i64x2_mul(vabsaccAB, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() 257 const v128_t vabsprodCD = wasm_i64x2_mul(vabsaccCD, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() 258 const v128_t vabsprodEF = wasm_i64x2_mul(vabsaccEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() 334 const v128_t vabsprod01 = wasm_i64x2_mul(vabsacc01, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() 335 const v128_t vabsprod23 = wasm_i64x2_mul(vabsacc23, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() [all …]
|
D | 7p7x-minmax-wasmsimd-c24-acc2.c | 382 const v128_t vabsprod01 = wasm_i64x2_mul(vabsacc01, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() 383 const v128_t vabsprod23 = wasm_i64x2_mul(vabsacc23, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() 384 const v128_t vabsprod45 = wasm_i64x2_mul(vabsacc45, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() 385 const v128_t vabsprod67 = wasm_i64x2_mul(vabsacc67, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() 386 const v128_t vabsprod89 = wasm_i64x2_mul(vabsacc89, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() 387 const v128_t vabsprodAB = wasm_i64x2_mul(vabsaccAB, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() 388 const v128_t vabsprodCD = wasm_i64x2_mul(vabsaccCD, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() 389 const v128_t vabsprodEF = wasm_i64x2_mul(vabsaccEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() 390 const v128_t vabsprodGH = wasm_i64x2_mul(vabsaccGH, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() 391 const v128_t vabsprodIJ = wasm_i64x2_mul(vabsaccIJ, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() [all …]
|
D | 7p7x-minmax-wasmsimd-c8-acc2.c | 194 const v128_t vabsprod01 = wasm_i64x2_mul(vabsacc01, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c8_acc2() 195 const v128_t vabsprod23 = wasm_i64x2_mul(vabsacc23, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c8_acc2() 196 const v128_t vabsprod45 = wasm_i64x2_mul(vabsacc45, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c8_acc2() 197 const v128_t vabsprod67 = wasm_i64x2_mul(vabsacc67, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c8_acc2() 264 const v128_t vabsprod01 = wasm_i64x2_mul(vabsacc01, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c8_acc2() 265 const v128_t vabsprod23 = wasm_i64x2_mul(vabsacc23, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c8_acc2() 266 const v128_t vabsprod45 = wasm_i64x2_mul(vabsacc45, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c8_acc2() 267 const v128_t vabsprod67 = wasm_i64x2_mul(vabsacc67, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c8_acc2()
|
/external/XNNPACK/src/qs8-dwconv/gen/ |
D | up24x9-minmax-wasmsimd-mul16.c | 286 const v128_t vprod01 = wasm_i64x2_add(wasm_i64x2_mul(vacc01, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16() 287 const v128_t vprod23 = wasm_i64x2_add(wasm_i64x2_mul(vacc23, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16() 288 const v128_t vprod45 = wasm_i64x2_add(wasm_i64x2_mul(vacc45, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16() 289 const v128_t vprod67 = wasm_i64x2_add(wasm_i64x2_mul(vacc67, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16() 290 const v128_t vprod89 = wasm_i64x2_add(wasm_i64x2_mul(vacc89, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16() 291 const v128_t vprodAB = wasm_i64x2_add(wasm_i64x2_mul(vaccAB, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16() 292 const v128_t vprodCD = wasm_i64x2_add(wasm_i64x2_mul(vaccCD, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16() 293 const v128_t vprodEF = wasm_i64x2_add(wasm_i64x2_mul(vaccEF, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16() 294 const v128_t vprodGH = wasm_i64x2_add(wasm_i64x2_mul(vaccGH, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16() 295 const v128_t vprodIJ = wasm_i64x2_add(wasm_i64x2_mul(vaccIJ, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16() [all …]
|
D | up16x9-minmax-wasmsimd-mul16.c | 233 const v128_t vprod01 = wasm_i64x2_add(wasm_i64x2_mul(vacc01, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16() 234 const v128_t vprod23 = wasm_i64x2_add(wasm_i64x2_mul(vacc23, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16() 235 const v128_t vprod45 = wasm_i64x2_add(wasm_i64x2_mul(vacc45, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16() 236 const v128_t vprod67 = wasm_i64x2_add(wasm_i64x2_mul(vacc67, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16() 237 const v128_t vprod89 = wasm_i64x2_add(wasm_i64x2_mul(vacc89, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16() 238 const v128_t vprodAB = wasm_i64x2_add(wasm_i64x2_mul(vaccAB, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16() 239 const v128_t vprodCD = wasm_i64x2_add(wasm_i64x2_mul(vaccCD, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16() 240 const v128_t vprodEF = wasm_i64x2_add(wasm_i64x2_mul(vaccEF, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16() 372 const v128_t vprod01 = wasm_i64x2_add(wasm_i64x2_mul(vacc01, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16() 373 const v128_t vprod23 = wasm_i64x2_add(wasm_i64x2_mul(vacc23, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16() [all …]
|
D | up8x9-minmax-wasmsimd-mul16.c | 180 const v128_t vprod01 = wasm_i64x2_add(wasm_i64x2_mul(vacc01, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16() 181 const v128_t vprod23 = wasm_i64x2_add(wasm_i64x2_mul(vacc23, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16() 182 const v128_t vprod45 = wasm_i64x2_add(wasm_i64x2_mul(vacc45, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16() 183 const v128_t vprod67 = wasm_i64x2_add(wasm_i64x2_mul(vacc67, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16() 296 const v128_t vprod01 = wasm_i64x2_add(wasm_i64x2_mul(vacc01, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16() 297 const v128_t vprod23 = wasm_i64x2_add(wasm_i64x2_mul(vacc23, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16() 298 const v128_t vprod45 = wasm_i64x2_add(wasm_i64x2_mul(vacc45, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16() 299 const v128_t vprod67 = wasm_i64x2_add(wasm_i64x2_mul(vacc67, vmultiplier), vrounding); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16()
|
/external/XNNPACK/src/qs8-gemm/gen/ |
D | 3x4c8-xw-minmax-wasmsimd.c | 150 const v128_t vprod0x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x01, vmultiplier), vrounding); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd() 152 const v128_t vprod1x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x01, vmultiplier), vrounding); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd() 154 const v128_t vprod2x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc2x01, vmultiplier), vrounding); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd() 157 const v128_t vprod0x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x23, vmultiplier), vrounding); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd() 158 const v128_t vprod1x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x23, vmultiplier), vrounding); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd() 159 const v128_t vprod2x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc2x23, vmultiplier), vrounding); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
|
D | 3x4c8-minmax-wasmsimd-ld128.c | 154 const v128_t vprod0x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x01, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128() 156 const v128_t vprod1x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x01, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128() 158 const v128_t vprod2x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc2x01, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128() 161 const v128_t vprod0x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x23, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128() 162 const v128_t vprod1x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x23, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128() 163 const v128_t vprod2x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc2x23, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
|
D | 3x4c8-minmax-wasmsimd-ld64.c | 150 const v128_t vprod0x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x01, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64() 152 const v128_t vprod1x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x01, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64() 154 const v128_t vprod2x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc2x01, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64() 157 const v128_t vprod0x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x23, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64() 158 const v128_t vprod1x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x23, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64() 159 const v128_t vprod2x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc2x23, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
|
D | 2x4c8-xw-minmax-wasmsimd.c | 121 const v128_t vprod0x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x01, vmultiplier), vrounding); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd() 123 const v128_t vprod1x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x01, vmultiplier), vrounding); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd() 126 const v128_t vprod0x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x23, vmultiplier), vrounding); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd() 127 const v128_t vprod1x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x23, vmultiplier), vrounding); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
|
D | 2x4c8-minmax-wasmsimd-ld128.c | 125 const v128_t vprod0x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x01, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128() 127 const v128_t vprod1x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x01, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128() 130 const v128_t vprod0x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x23, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128() 131 const v128_t vprod1x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x23, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
|
D | 2x4c8-minmax-wasmsimd-ld64.c | 121 const v128_t vprod0x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x01, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64() 123 const v128_t vprod1x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x01, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64() 126 const v128_t vprod0x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x23, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64() 127 const v128_t vprod1x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x23, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
|
D | 1x4c8-minmax-wasmsimd-ld64.c | 92 const v128_t vprod0x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x01, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld64() 95 const v128_t vprod0x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x23, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
|
D | 1x4c8-minmax-wasmsimd-ld128.c | 96 const v128_t vprod0x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x01, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld128() 99 const v128_t vprod0x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x23, vmultiplier), vrounding); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
|
/external/XNNPACK/src/qs8-igemm/gen/ |
D | 3x4c8-minmax-wasmsimd-ld128.c | 171 const v128_t vprod0x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x01, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128() 173 const v128_t vprod1x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x01, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128() 175 const v128_t vprod2x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc2x01, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128() 178 const v128_t vprod0x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x23, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128() 179 const v128_t vprod1x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x23, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128() 180 const v128_t vprod2x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc2x23, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
|
D | 3x4c8-minmax-wasmsimd-ld64.c | 167 const v128_t vprod0x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x01, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64() 169 const v128_t vprod1x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x01, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64() 171 const v128_t vprod2x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc2x01, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64() 174 const v128_t vprod0x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x23, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64() 175 const v128_t vprod1x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x23, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64() 176 const v128_t vprod2x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc2x23, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
|
D | 2x4c8-minmax-wasmsimd-ld128.c | 140 const v128_t vprod0x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x01, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128() 142 const v128_t vprod1x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x01, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128() 145 const v128_t vprod0x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x23, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128() 146 const v128_t vprod1x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x23, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
|
D | 2x4c8-minmax-wasmsimd-ld64.c | 136 const v128_t vprod0x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x01, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64() 138 const v128_t vprod1x01 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x01, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64() 141 const v128_t vprod0x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc0x23, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64() 142 const v128_t vprod1x23 = wasm_i64x2_add(wasm_i64x2_mul(vacc1x23, vmultiplier), vrounding); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
|
/external/XNNPACK/src/qs8-gavgpool/ |
D | unipass-wasmsimd.c.in | 89 const v128_t vabsprod${ABC[C:C+2]} = wasm_i64x2_mul(vabsacc${ABC[C:C+2]}, vmultiplier); 90 const v128_t vabsprod${ABC[C+2:C+4]} = wasm_i64x2_mul(vabsacc${ABC[C+2:C+4]}, vmultiplier); 161 const v128_t vabsprod${ABC[0:2]} = wasm_i64x2_mul(vabsacc${ABC[0:2]}, vmultiplier); 162 const v128_t vabsprod${ABC[2:4]} = wasm_i64x2_mul(vabsacc${ABC[2:4]}, vmultiplier); 163 const v128_t vabsprod${ABC[4:6]} = wasm_i64x2_mul(vabsacc${ABC[4:6]}, vmultiplier); 164 const v128_t vabsprod${ABC[6:8]} = wasm_i64x2_mul(vabsacc${ABC[6:8]}, vmultiplier);
|
/external/XNNPACK/src/qs8-dwconv/ |
D | unipass-wasmsimd-mul16.c.in | 77 …const v128_t vprod${ABC[C:C+2]} = wasm_i64x2_add(wasm_i64x2_mul(vacc${ABC[C:C+2]}, vmultiplier), v… 153 …const v128_t vprod${ABC[0:2]} = wasm_i64x2_add(wasm_i64x2_mul(vacc${ABC[0:2]}, vmultiplier), vroun… 154 …const v128_t vprod${ABC[2:4]} = wasm_i64x2_add(wasm_i64x2_mul(vacc${ABC[2:4]}, vmultiplier), vroun… 155 …const v128_t vprod${ABC[4:6]} = wasm_i64x2_add(wasm_i64x2_mul(vacc${ABC[4:6]}, vmultiplier), vroun… 156 …const v128_t vprod${ABC[6:8]} = wasm_i64x2_add(wasm_i64x2_mul(vacc${ABC[6:8]}, vmultiplier), vroun…
|