Skip to content

Commit c77741a

Browse files
author
[email protected]
committed
resolved linter errors
1 parent a8083ae commit c77741a

File tree

8 files changed

+215
-203
lines changed

8 files changed

+215
-203
lines changed

‎backends/cadence/fusion_g3/operators/op_add.cpp

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -331,7 +331,7 @@ Tensor& add_scalar_out(
331331
alpha_val,
332332
out.numel());
333333

334-
} else {
334+
} else {
335335
ET_SWITCH_REALB_TYPES(compute_type, ctx, op_name, CTYPE_COMPUTE, [&]() {
336336
torch::executor::native::utils::
337337
apply_unitensor_elementwise_fn<CTYPE_COMPUTE, op_name>(

‎backends/cadence/fusion_g3/operators/op_cat.cpp

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -43,23 +43,24 @@ Tensor& cat_out(
4343
exec_aten::ArrayRef<Tensor> tensors,
4444
int64_t dim,
4545
Tensor& out) {
46-
4746
if (dim < 0) {
4847
dim += out.dim();
4948
}
5049

51-
#ifdef OPT_ARG_CHECK
50+
int kTensorDimensionLimit = executorch::runtime::kTensorDimensionLimit;
51+
52+
#ifdef OP_ARG_CHECK
5253
ET_KERNEL_CHECK(
5354
ctx,
5455
torch::executor::check_cat_args(tensors, dim, out),
5556
InvalidArgument,
5657
out);
57-
58+
5859
Tensor::SizesType expected_out_size[kTensorDimensionLimit];
5960
size_t expected_out_dim = 0;
6061
torch::executor::get_cat_out_target_size(
6162
tensors, dim, expected_out_size, &expected_out_dim);
62-
63+
6364
ET_KERNEL_CHECK(
6465
ctx,
6566
executorch::runtime::resize_tensor(
@@ -81,7 +82,6 @@ Tensor& cat_out(
8182
return out;
8283
}
8384

84-
int kTensorDimensionLimit = executorch::runtime::kTensorDimensionLimit;
8585

8686
const signed char* inp_tensors[tensors.size()];
8787
const int* inp_tensors_shapes[tensors.size()];

‎backends/cadence/fusion_g3/operators/op_dequantize.cpp

Lines changed: 94 additions & 86 deletions
Original file line numberDiff line numberDiff line change
@@ -92,7 +92,8 @@ void check_dequantize_per_tensor_args(
9292
} // namespace
9393

9494
/* Local function which calls the kernels based on the input datatype */
95-
Tensor & dequantize_impl(KernelRuntimeContext& ctx,
95+
Tensor& dequantize_impl(
96+
KernelRuntimeContext& ctx,
9697
Tensor& out,
9798
const Tensor& input,
9899
float* scale_data,
@@ -132,82 +133,82 @@ Tensor & dequantize_impl(KernelRuntimeContext& ctx,
132133
if (is_asym_dequant) {
133134
if (input.scalar_type() == ScalarType::Byte) {
134135
const uint8_t* input_data = input.const_data_ptr<uint8_t>();
135-
XT_KERNEL_CHECK(
136+
XT_KERNEL_CHECK(
136137
ctx,
137138
out,
138-
xa_nn_elm_dequantize_asym8u_f32,
139-
out_data,
140-
input_data,
141-
inp_shape,
142-
input.dim(),
143-
axis,
144-
zero_point_data,
145-
scale_data);
139+
xa_nn_elm_dequantize_asym8u_f32,
140+
out_data,
141+
input_data,
142+
inp_shape,
143+
input.dim(),
144+
axis,
145+
zero_point_data,
146+
scale_data);
146147
} else if (input.scalar_type() == ScalarType::Char) {
147148
const int8_t* input_data = input.const_data_ptr<int8_t>();
148-
XT_KERNEL_CHECK(
149+
XT_KERNEL_CHECK(
149150
ctx,
150151
out,
151-
xa_nn_elm_dequantize_asym8_f32,
152-
out_data,
153-
input_data,
154-
inp_shape,
155-
input.dim(),
156-
axis,
157-
zero_point_data,
158-
scale_data);
152+
xa_nn_elm_dequantize_asym8_f32,
153+
out_data,
154+
input_data,
155+
inp_shape,
156+
input.dim(),
157+
axis,
158+
zero_point_data,
159+
scale_data);
159160
} else if (input.scalar_type() == (ScalarType)Ushort) {
160161
const uint16_t* input_data = input.const_data_ptr<uint16_t>();
161-
XT_KERNEL_CHECK(
162+
XT_KERNEL_CHECK(
162163
ctx,
163164
out,
164-
xa_nn_elm_dequantize_asym16u_f32,
165-
out_data,
166-
input_data,
167-
inp_shape,
168-
input.dim(),
169-
axis,
170-
zero_point_data,
171-
scale_data);
165+
xa_nn_elm_dequantize_asym16u_f32,
166+
out_data,
167+
input_data,
168+
inp_shape,
169+
input.dim(),
170+
axis,
171+
zero_point_data,
172+
scale_data);
172173
} else if (input.scalar_type() == ScalarType::Short) {
173174
const int16_t* input_data = input.const_data_ptr<int16_t>();
174-
XT_KERNEL_CHECK(
175+
XT_KERNEL_CHECK(
175176
ctx,
176177
out,
177-
xa_nn_elm_dequantize_asym16_f32,
178-
out_data,
179-
input_data,
180-
inp_shape,
181-
input.dim(),
182-
axis,
183-
zero_point_data,
184-
scale_data);
178+
xa_nn_elm_dequantize_asym16_f32,
179+
out_data,
180+
input_data,
181+
inp_shape,
182+
input.dim(),
183+
axis,
184+
zero_point_data,
185+
scale_data);
185186
} else if (input.scalar_type() == (ScalarType)Bits4u) {
186187
const uint8_t* input_data = input.const_data_ptr<uint8_t>();
187-
XT_KERNEL_CHECK(
188+
XT_KERNEL_CHECK(
188189
ctx,
189190
out,
190-
xa_nn_elm_dequantize_asym4u_f32,
191-
out_data,
192-
input_data,
193-
inp_shape,
194-
input.dim(),
195-
axis,
196-
zero_point_data,
197-
scale_data);
191+
xa_nn_elm_dequantize_asym4u_f32,
192+
out_data,
193+
input_data,
194+
inp_shape,
195+
input.dim(),
196+
axis,
197+
zero_point_data,
198+
scale_data);
198199
} else if (input.scalar_type() == (ScalarType)Bits4) {
199200
const int8_t* input_data = input.const_data_ptr<int8_t>();
200-
XT_KERNEL_CHECK(
201+
XT_KERNEL_CHECK(
201202
ctx,
202203
out,
203-
xa_nn_elm_dequantize_asym4_f32,
204-
out_data,
205-
input_data,
206-
inp_shape,
207-
input.dim(),
208-
axis,
209-
zero_point_data,
210-
scale_data);
204+
xa_nn_elm_dequantize_asym4_f32,
205+
out_data,
206+
input_data,
207+
inp_shape,
208+
input.dim(),
209+
axis,
210+
zero_point_data,
211+
scale_data);
211212
} else {
212213
if (axis == NULL) {
213214
// calculate the dequantized output, cast scale to float to match fbgemm
@@ -343,10 +344,10 @@ Tensor & dequantize_impl(KernelRuntimeContext& ctx,
343344
} else {
344345
if (input.scalar_type() == ScalarType::Byte) {
345346
const uint8_t* input_data = input.const_data_ptr<uint8_t>();
346-
XT_KERNEL_CHECK(
347+
XT_KERNEL_CHECK(
347348
ctx,
348349
out,
349-
xa_nn_elm_dequantize_sym8u_f32,
350+
xa_nn_elm_dequantize_sym8u_f32,
350351
out_data,
351352
input_data,
352353
inp_shape,
@@ -358,19 +359,19 @@ Tensor & dequantize_impl(KernelRuntimeContext& ctx,
358359
XT_KERNEL_CHECK(
359360
ctx,
360361
out,
361-
xa_nn_elm_dequantize_sym8_f32,
362-
out_data,
363-
input_data,
364-
inp_shape,
365-
input.dim(),
366-
axis,
367-
scale_data);
362+
xa_nn_elm_dequantize_sym8_f32,
363+
out_data,
364+
input_data,
365+
inp_shape,
366+
input.dim(),
367+
axis,
368+
scale_data);
368369
} else if (input.scalar_type() == (ScalarType)Ushort) {
369370
const uint16_t* input_data = input.const_data_ptr<uint16_t>();
370-
XT_KERNEL_CHECK(
371+
XT_KERNEL_CHECK(
371372
ctx,
372373
out,
373-
xa_nn_elm_dequantize_sym16u_f32,
374+
xa_nn_elm_dequantize_sym16u_f32,
374375
out_data,
375376
input_data,
376377
inp_shape,
@@ -379,10 +380,10 @@ Tensor & dequantize_impl(KernelRuntimeContext& ctx,
379380
scale_data);
380381
} else if (input.scalar_type() == ScalarType::Short) {
381382
const int16_t* input_data = input.const_data_ptr<int16_t>();
382-
XT_KERNEL_CHECK(
383+
XT_KERNEL_CHECK(
383384
ctx,
384385
out,
385-
xa_nn_elm_dequantize_sym16_f32,
386+
xa_nn_elm_dequantize_sym16_f32,
386387
out_data,
387388
input_data,
388389
inp_shape,
@@ -391,10 +392,10 @@ Tensor & dequantize_impl(KernelRuntimeContext& ctx,
391392
scale_data);
392393
} else if (input.scalar_type() == (ScalarType)Bits4u) {
393394
const uint8_t* input_data = input.const_data_ptr<uint8_t>();
394-
XT_KERNEL_CHECK(
395+
XT_KERNEL_CHECK(
395396
ctx,
396397
out,
397-
xa_nn_elm_dequantize_sym4u_f32,
398+
xa_nn_elm_dequantize_sym4u_f32,
398399
out_data,
399400
input_data,
400401
inp_shape,
@@ -403,10 +404,10 @@ Tensor & dequantize_impl(KernelRuntimeContext& ctx,
403404
scale_data);
404405
} else if (input.scalar_type() == (ScalarType)Bits4) {
405406
const int8_t* input_data = input.const_data_ptr<int8_t>();
406-
XT_KERNEL_CHECK(
407+
XT_KERNEL_CHECK(
407408
ctx,
408409
out,
409-
xa_nn_elm_dequantize_sym4_f32,
410+
xa_nn_elm_dequantize_sym4_f32,
410411
out_data,
411412
input_data,
412413
inp_shape,
@@ -558,7 +559,8 @@ Tensor & dequantize_impl(KernelRuntimeContext& ctx,
558559
* https://github.com/pytorch/pytorch/pull/87093#discussion_r1000841181 for more
559560
* info.
560561
*/
561-
Tensor& dequantize_per_tensor_out(KernelRuntimeContext& context,
562+
Tensor& dequantize_per_tensor_out(
563+
KernelRuntimeContext& context,
562564
const Tensor& input,
563565
double scale,
564566
int64_t zero_point,
@@ -572,20 +574,22 @@ Tensor& dequantize_per_tensor_out(KernelRuntimeContext& context,
572574
ET_CHECK_MSG(
573575
err == torch::executor::Error::Ok,
574576
"Failed to resize out Tensor in dequantize_per_tensor_out");
575-
577+
576578
check_dequantize_per_tensor_args(
577579
input, quant_min, quant_max, dtype, out_dtype, out);
578580
#endif
579581

580582
float scale_data = (float)scale;
581583
int zero_point_data = (int)zero_point;
582584

583-
dequantize_impl(context, out, input, &scale_data, &zero_point_data, NULL, out_dtype);
585+
dequantize_impl(
586+
context, out, input, &scale_data, &zero_point_data, NULL, out_dtype);
584587

585588
return out;
586589
}
587590

588-
Tensor& dequantize_per_tensor_tensor_args_out(KernelRuntimeContext& context,
591+
Tensor& dequantize_per_tensor_tensor_args_out(
592+
KernelRuntimeContext& context,
589593
const Tensor& input,
590594
const Tensor& scale,
591595
const Tensor& zero_point,
@@ -613,7 +617,8 @@ Tensor& dequantize_per_tensor_tensor_args_out(KernelRuntimeContext& context,
613617
ssize_t(zero_point.numel()));
614618
#endif
615619

616-
dequantize_per_tensor_out(context,
620+
dequantize_per_tensor_out(
621+
context,
617622
input,
618623
scale.const_data_ptr<double>()[0],
619624
zero_point.const_data_ptr<int64_t>()[0],
@@ -626,7 +631,8 @@ Tensor& dequantize_per_tensor_tensor_args_out(KernelRuntimeContext& context,
626631
return out;
627632
}
628633

629-
Tensor& dequantize_per_channel_out(KernelRuntimeContext& context,
634+
Tensor& dequantize_per_channel_out(
635+
KernelRuntimeContext& context,
630636
const Tensor& input,
631637
const Tensor& scale,
632638
const exec_aten::optional<Tensor>& opt_zero_points,
@@ -636,14 +642,13 @@ Tensor& dequantize_per_channel_out(KernelRuntimeContext& context,
636642
ScalarType dtype,
637643
exec_aten::optional<ScalarType> out_dtype,
638644
Tensor& out) {
639-
640645
if (axis < 0) {
641646
axis += executorch::runtime::nonzero_dim(input);
642647
}
643-
/* if the arguments are passed properly to the operator disable the Macro - "OP_ARG_CHECK"
644-
* if not the case, enable the Macro - "OP_ARG_CHECK", to have the checks only in
645-
* operator level(As there are no checks in kernel).
646-
*/
648+
/* if the arguments are passed properly to the operator disable the Macro -
649+
* "OP_ARG_CHECK" if not the case, enable the Macro - "OP_ARG_CHECK", to have
650+
* the checks only in operator level(As there are no checks in kernel).
651+
*/
647652
#ifdef OP_ARG_CHECK
648653
torch::executor::Error err = resize_tensor(out, input.sizes());
649654

@@ -705,12 +710,14 @@ Tensor& dequantize_per_channel_out(KernelRuntimeContext& context,
705710
for (int i = 0; i < scale.numel(); i++) {
706711
scale_data[i] = (float)scale_dt[i];
707712
}
708-
dequantize_impl(context, out, input, scale_data, zero_point_ptr, axis_ptr, out_dtype);
713+
dequantize_impl(
714+
context, out, input, scale_data, zero_point_ptr, axis_ptr, out_dtype);
709715

710716
return out;
711717
}
712718

713-
Tensor& dequantize_per_token_out(KernelRuntimeContext& context,
719+
Tensor& dequantize_per_token_out(
720+
KernelRuntimeContext& context,
714721
const Tensor& input,
715722
const Tensor& scale,
716723
const Tensor& zero_points,
@@ -757,7 +764,8 @@ Tensor& dequantize_per_token_out(KernelRuntimeContext& context,
757764
"Failed to resize out Tensor in dequantize_per_channel_out");
758765
#endif
759766

760-
return dequantize_per_channel_out(context,
767+
return dequantize_per_channel_out(
768+
context,
761769
reshaped_input,
762770
scale,
763771
zero_points,

0 commit comments

Comments
 (0)