| /third_party/mindspore/mindspore/ops/_op_impl/_custom_op/ |
| D | batchnorm_fold.py | 67 batch_mean = te.lang.cce.vmuls(x_sum, num_rec)
70 variance_div = te.lang.cce.vmuls(x_square_sum, num_rec)
78 batch_var_unbiased = te.lang.cce.vmuls(batch_var_biased, batch_var_scaler)
82 mean_mul = te.lang.cce.vmuls(batch_mean, factor)
83 mean_mul_rev = te.lang.cce.vmuls(mean, factor_reverse)
86 var_mul = te.lang.cce.vmuls(batch_var_unbiased, factor)
87 var_mul_rev = te.lang.cce.vmuls(variance, factor_reverse)
|
| D | matmul_cube_dense_right_impl.py | 118 … tik_instance.vmuls(64, resmatmul_local_ub, resmatmul_local_ub, matrix_max_scalar, 255, 1, 1, 8, 8)
119 tik_instance.vmuls(64, resmatmul_local_ub[255 * 64], resmatmul_local_ub[255 * 64],
121 tik_instance.vmuls(64, resmatmul_local_ub[510 * 64], resmatmul_local_ub[510 * 64],
171 … tik_instance.vmuls(64, resmatmul_local_ub, resmatmul_local_ub, matrix_max_scalar, 255, 1, 1, 8, 8)
172 …tik_instance.vmuls(64, resmatmul_local_ub[255 * 64], resmatmul_local_ub[255 * 64], matrix_max_scal…
174 …tik_instance.vmuls(64, resmatmul_local_ub[510 * 64], resmatmul_local_ub[510 * 64], matrix_max_scal…
195 …tik_instance.vmuls(64, resmatmul_local_ub1, resmatmul_local_ub1, matrix_max_scalar, 255, 1, 1, 8, …
196 …tik_instance.vmuls(64, resmatmul_local_ub1[255 * 64], resmatmul_local_ub1[255 * 64], matrix_max_sc…
|
| D | cholesky_trsm_impl.py | 65 … tik_instance.vmuls(64, assist_2_ub[0], assist_1_ub[0], scalar1, vector_repeat_times, 1, 1, 8, 8)
68 …tik_instance.vmuls(64, input_x_ub[i, 0], input_x_ub[i, 0], scalar2, vector_repeat_times, 1, 1, 8, …
72 … tik_instance.vmuls(64, temp_ub[j, 0], input_x_ub[i, 0], scalar3, vector_repeat_times, 1, 1, 8, 8)
96 … tik_instance.vmuls(64, assist_2_ub, temp_ub[j + index + 1, 0], chol_diag_element_loop,
105 …tik_instance.vmuls(64, temp_ub[index, 0], temp_ub[index, 0], chol_diag_element, vector_repeat_time…
|
| D | minmax_update_perchannel.py | 69 min_val = te.lang.cce.vadd(te.lang.cce.vmuls(
70 min_val, ema_decay), te.lang.cce.vmuls(x_min, (1 - ema_decay)))
71 max_val = te.lang.cce.vadd(te.lang.cce.vmuls(
72 max_val, ema_decay), te.lang.cce.vmuls(x_max, (1 - ema_decay)))
|
| D | minmax_update_perlayer.py | 66 min_val = te.lang.cce.vadd(te.lang.cce.vmuls(
67 min_val, ema_decay), te.lang.cce.vmuls(x_min, (1 - ema_decay)))
68 max_val = te.lang.cce.vadd(te.lang.cce.vmuls(
69 max_val, ema_decay), te.lang.cce.vmuls(x_max, (1 - ema_decay)))
|
| D | fake_quant_perlayer_grad.py | 67 res_max_mul = te.lang.cce.vmuls(res_max, max_value)
68 res_max_mul_max = te.lang.cce.vmuls(res_max_mul, max_value)
69 res = te.lang.cce.vmuls(res_max_mul_max, factor_value)
92 max_val = te.lang.cce.vmax(te.lang.cce.vmuls(min_val, -1.), max_val)
93 min_val = te.lang.cce.vmuls(max_val, -1.)
|
| D | fake_learned_scale_quant_perchannel_grad.py | 77 res_mul1 = te.lang.cce.vmuls(vmin, max_support_fp32)
78 res_mul2 = te.lang.cce.vmuls(res_mul1, max_support_fp32)
79 sign = te.lang.cce.vmuls(res_mul2, tvm.const(2 ** 2, dtype=dtype))
84 res_mul1 = te.lang.cce.vmuls(vmin, max_support_fp16)
85 sign = te.lang.cce.vmuls(res_mul1, max_support_fp16)
|
| D | fake_learned_scale_quant_perlayer_grad.py | 78 res_mul1 = te.lang.cce.vmuls(vmin, max_support_fp32)
79 res_mul2 = te.lang.cce.vmuls(res_mul1, max_support_fp32)
80 sign = te.lang.cce.vmuls(res_mul2, tvm.const(2 ** 2, dtype=dtype))
85 res_mul1 = te.lang.cce.vmuls(vmin, max_support_fp16)
86 sign = te.lang.cce.vmuls(res_mul1, max_support_fp16)
|
| D | fake_quant_perchannel_grad.py | 66 res_max_mul = te.lang.cce.vmuls(res_max, max_value)
67 res_max_mul_max = te.lang.cce.vmuls(res_max_mul, max_value)
68 res = te.lang.cce.vmuls(res_max_mul_max, factor_value)
|
| D | batchnorm_fold2_grad.py | 71 d_batch_mean = te.lang.cce.vmuls(d_batch_mean, -1.)
75 d_gamma = te.lang.cce.vmuls(d_gamma, -1.)
|
| D | fake_quant_perchannel.py | 61 max_val = te.lang.cce.vmax(te.lang.cce.vmuls(min_val, -1.), max_val)
62 min_val = te.lang.cce.vmuls(max_val, -1.)
|
| D | fake_quant_perlayer.py | 61 max_val = te.lang.cce.vmax(te.lang.cce.vmuls(min_val, -1.), max_val)
62 min_val = te.lang.cce.vmuls(max_val, -1.)
|
| D | matmul_cube_fracz_right_mul_impl.py | 233 tik_instance.vmuls(vectorfp32_size,
239 tik_instance.vmuls(vectorfp32_size,
|
| D | batchnorm_fold_grad.py | 71 dx = te.lang.cce.vmuls(dx, tvm.const(1. / normal_size, dtype=dx.dtype))
|
| D | matmul_dds_impl.py | 339 tik_inst.vmuls(
|
| /third_party/skia/third_party/externals/swiftshader/third_party/subzero/src/ |
| D | IceAssemblerARM32.h | 555 void vmuls(const Operand *OpSd, const Operand *OpSn, const Operand *OpSm,
|
| D | IceInstARM32.cpp | 1269 Asm->vmuls(Dest, getSrc(0), getSrc(1), CondARM32::AL); in emitIAS()
|
| D | IceAssemblerARM32.cpp | 3297 void AssemblerARM32::vmuls(const Operand *OpSd, const Operand *OpSn, in vmuls() function in Ice::ARM32::AssemblerARM32
|
| /third_party/skia/third_party/externals/swiftshader/third_party/subzero/src/DartARM32/ |
| D | assembler_arm.h | 676 void vmuls(SRegister sd, SRegister sn, SRegister sm, Condition cond = AL);
|
| D | assembler_arm.cc | 950 void Assembler::vmuls(SRegister sd, SRegister sn, SRegister sm,
|