platform/core/ml/nnfw.git
6 years ago[neurun] Introduce contains in IndexSet (#2457)
김수진/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 24 Aug 2018 03:46:10 +0000 (12:46 +0900)]
[neurun] Introduce contains in IndexSet (#2457)

This commit introduces contains helper in IndexSet for checking if index of operand is exist.

Signed-off-by: sjsujinkim <sjsujin.kim@samsung.com>
6 years agoAdd nnapi_quickcheck for split operator (multi-dimension) (#2449)
윤지영/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 24 Aug 2018 02:30:13 +0000 (11:30 +0900)]
Add nnapi_quickcheck for split operator (multi-dimension) (#2449)

Add below testcases for split operator:
  NNAPI_Quickcheck_split_2.simple_test (4 rank, 2 axis)
  NNAPI_Quickcheck_split_3.simple_test (2 rank, 1 axis)
  NNAPI_Quickcheck_split_4.simple_test (2 rank, 0 axis)

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
6 years agoEnable to pass ValidationTestModel.Finish in pureacl (#2454)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 24 Aug 2018 02:29:01 +0000 (11:29 +0900)]
Enable to pass ValidationTestModel.Finish in pureacl (#2454)

- Add checking if building model is finished or not

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years agoEnable to pass ValidationTestModel.AddOperation in pureacl (#2451)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 24 Aug 2018 02:28:46 +0000 (11:28 +0900)]
Enable to pass ValidationTestModel.AddOperation in pureacl (#2451)

In `ValidationTestModel.AddOperation`, if inputs or outputs are `nullptr`, it is always error even if `inputCount` and `outputCount` is zero.
This commit will modify those conditions, and add some checkings to pass the test.

 - Modify `nullptr` check condition
 - Add checking if building model is finished or not
 - Add checking if type is invalid or not

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years agoEnable to pass ValidationTestModel.identifyInputsAndOutputs in pureacl (#2452)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 24 Aug 2018 02:27:54 +0000 (11:27 +0900)]
Enable to pass ValidationTestModel.identifyInputsAndOutputs in pureacl (#2452)

- Add checking if building model is finished or not

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years agoIntroduce Split operation into pureACL runtime (#2447)
윤지영/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 24 Aug 2018 02:24:49 +0000 (11:24 +0900)]
Introduce Split operation into pureACL runtime (#2447)

* Introduce Split operation into pureACL runtime

This commit introduces Split operation into pureACL runtime.

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
* Remove num_split_index of Split operator

The `num_split_index` is always the same with the number of output.
So it can be omitted.

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
6 years agoRevert "Add comment for skipping fully_connected_quant8_2 of GeneratedTests (#2426...
김용섭/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 23 Aug 2018 10:54:30 +0000 (19:54 +0900)]
Revert "Add comment for skipping fully_connected_quant8_2 of GeneratedTests (#2426)" (#2443)

This reverts commit 3e32b4fda06ec739eeb078ee2b1dc7f86b616cc6.

According to V1_1's fully_connected_float_4d_simple's comment, this test
is not wrong. The shape can be guessed by output. The test is not wrong,
so this revert is needed.

Signed-off-by: Yongseop Kim <yons.kim@samsung.com>
6 years agoUpdate model.cpp of depthwise_conv2d (#2441)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 23 Aug 2018 10:43:18 +0000 (19:43 +0900)]
Update model.cpp of depthwise_conv2d (#2441)

By #2300, `depthwise_conv2d_float_large_2_weights_as_inputs` was modified
But it was not the result of `generate_test.sh`, it was my own modification
This commit will apply the result of `generate_test.sh`

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years agoPureACL handles operators that generate no kernel (e.g. CONCATENATION) (#2437)
이상규/동작제어Lab(SR)/Principal Engineer/삼성전자 [Thu, 23 Aug 2018 10:42:55 +0000 (19:42 +0900)]
PureACL handles operators that generate no kernel (e.g. CONCATENATION) (#2437)

Related Issue: #2431

Dummy stage instance is added for `CONCATENATION` because Profiler assumes
every op inserts a stage.

The output will not show `CONCATENATION` slot because it is based on
runtime execution event. There was no event for `CONCATENATION` at runtime.

Signed-off-by: Sanggyu Lee <sg5.lee@samsung.com>
6 years ago[neurun] Change internal datatype enum name (#2435)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 23 Aug 2018 08:33:25 +0000 (17:33 +0900)]
[neurun] Change internal datatype enum name (#2435)

Change enum name used for neurun internal data type

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years ago[neurun] Use correct scale for Softmax (#2433)
김수진/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 23 Aug 2018 06:31:19 +0000 (15:31 +0900)]
[neurun] Use correct scale for Softmax (#2433)

This commit uses correct scale for Softmax likes #1728.

Signed-off-by: sjsujinkim <sjsujin.kim@samsung.com>
6 years ago[neurun] Use internal data type in CPU backend (#2432)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 23 Aug 2018 06:09:57 +0000 (15:09 +0900)]
[neurun] Use internal data type in CPU backend (#2432)

Define convert function to make CPU backend shape from graph operand
Use internal data type to make backend shape
Change CPU backend interface to use backend shape instead of graph shape
Change CPU backend operation to save input data type

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoPass initing weight and bias on compilation time if data is null (#2427)
김용섭/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 23 Aug 2018 05:17:49 +0000 (14:17 +0900)]
Pass initing weight and bias on compilation time if data is null (#2427)

Fc's weight and bias can be passed on execution time not only model time.

these generated tests can be fixed as workaround method.
- fully_connected_quant8_weights_as_inputs
- fully_connected_quant8_large_weights_as_inputs

This should be fixed structurally on neuron runtime.
Ref: https://github.sec.samsung.net/STAR/nnfw/issues/2319

Signed-off-by: Yongseop Kim <yons.kim@samsung.com>
6 years agoAdd comment for skipping fully_connected_quant8_2 of GeneratedTests (#2426)
김용섭/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 23 Aug 2018 05:17:31 +0000 (14:17 +0900)]
Add comment for skipping fully_connected_quant8_2 of GeneratedTests (#2426)

Add comment for the reason why skip fully_connected_quant8_2 of
GeneratedTests. It's because the GeneratedTests.fully_connected_quant8_2
code is wrong. Skip this test as is if the test code is not fixed.

Signed-off-by: Yongseop Kim <yons.kim@samsung.com>
6 years ago[neurun] Use internal data type in cpu InitializeGenerator (#2428)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 23 Aug 2018 04:56:25 +0000 (13:56 +0900)]
[neurun] Use internal data type in cpu InitializeGenerator (#2428)

Define getter for internal data type
Use internal data type in cpu InitializeGenerator instead of shape

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoAvoid compile warning of narrowing uint32_t (#2429)
김용섭/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 23 Aug 2018 04:44:59 +0000 (13:44 +0900)]
Avoid compile warning of narrowing uint32_t (#2429)

When compilation.cc is compiled, warning message prints about narrowing
uint32_t -> int32_t. The code doesn't matter whether the vars are uint32_t
or int32_t.

Signed-off-by: Yongseop Kim <yons.kim@samsung.com>
6 years agoModify gather operation and testdata (#2407)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 23 Aug 2018 04:43:47 +0000 (13:43 +0900)]
Modify gather operation and testdata (#2407)

`GeneratedTests.gather_1D_uint8` and `GeneratedTests.gather_2D_uint8` are tests for `gather` operation using quant8 type
However, despite not correctly given test data format, two tests are passed because of weak dataset
And more seriously, `gather` operation do not support quant8 yet
Therefore we should disable the two tests

 - Modify `_uint8` to `_quant8` in test name
 - Modify test data for `GeneratedTests.gather_1D_uint8` and `GeneratedTests.gather_2D_uint8`
 - Disable `GeneratedTests.gather_1D_uint8` and `GeneratedTests.gather_2D_uint8`

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years agoEnable GeneratedTests.relu* tests (#2420)
윤지영/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 23 Aug 2018 04:05:49 +0000 (13:05 +0900)]
Enable GeneratedTests.relu* tests (#2420)

Remove below testcases in skip list:
  GeneratedTests.relu6_float_2
  GeneratedTests.relu_float_2

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
6 years agoUpdate with supporting LOGISTIC quantization (#2422)
윤지영/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 23 Aug 2018 04:05:32 +0000 (13:05 +0900)]
Update with supporting LOGISTIC quantization (#2422)

* Enable QASYMM8 in LOGISTIC of CLActivationLayerKernel. (#42)

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
6 years agoEnable GeneratedTests.rnn* tests (#2423)
장지섭/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 23 Aug 2018 03:46:20 +0000 (12:46 +0900)]
Enable GeneratedTests.rnn* tests (#2423)

This commit enables GeneratedTests.rnn* tests.
Remove below testcases in skip list:
  GeneratedTests.rnn
  GeneratedTests.rnn_state

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
6 years agoHandle multi-step operator's profile data (#2414)
이상규/동작제어Lab(SR)/Principal Engineer/삼성전자 [Thu, 23 Aug 2018 02:11:09 +0000 (11:11 +0900)]
Handle multi-step operator's profile data (#2414)

- Add profiler_summarizer in tools/tfite_benchmark_model

profile_summarizer must understand multi-step expansion for
pureACL runtime. So I need to introduce our own profiler_summarizer
which is based on original tflite_benchmark_model's profiler_summarizer.

- Handle multi-step operator's profile data

Now, it can handle multi step expanded operators.

The only modifications are in profile_summarizer.cc's ProcessProfiles().
You can do diff with tensorflow-v1.9.0-rc2.

Related Issue: #2376, #1696

Signed-off-by: Sanggyu Lee <sg5.lee@samsung.com>
6 years agoEnable GeneratedTests.floor_ test (#2418)
윤지영/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 23 Aug 2018 01:31:29 +0000 (10:31 +0900)]
Enable GeneratedTests.floor_ test (#2418)

Remove the GeneratedTests.floor_ testcase in skip list.

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
6 years ago[neurun] Set type info in internal operand (#2416)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 23 Aug 2018 01:25:30 +0000 (10:25 +0900)]
[neurun] Set type info in internal operand (#2416)

* [neurun] Set type info in internal operand

Assign TypeInfo into internal operand.
It is duplicated data with shape yet.

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
* Fix format and remove global namespace prefix

6 years agopureACl profiler keeps track of operator's index (#2412)
이상규/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 22 Aug 2018 10:03:51 +0000 (19:03 +0900)]
pureACl profiler keeps track of operator's index (#2412)

Some operators in model are expanded into multiple Steps.
To analyze the profile data, it need to keep track of operator index
to interpreter op's.

Now, we know where the ACL kernel is generated by seeing Step's _op_idx
member.

Related Issue: #2376

Signed-off-by: Sanggyu Lee <sg5.lee@samsung.com>
6 years ago[PureACL] NEON implicit CONV2D operation (#2356)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 22 Aug 2018 10:03:30 +0000 (19:03 +0900)]
[PureACL] NEON implicit CONV2D operation (#2356)

Added NEON implicit CONV2D operation.
From 20 `conv` generated tests, 6 tests passes.

```
$ NEON=1 LD_LIBRARY_PATH=Product/out/lib Product/out/unittest/runtime_run_android_nn_test --gtest_filter=GeneratedTests.conv*

[==========] Running 20 tests from 1 test case.
[----------] Global test environment set-up.
[----------] 20 tests from GeneratedTests
[       OK ] GeneratedTests.conv_1_h3_w2_SAME (51 ms)
[       OK ] GeneratedTests.conv_1_h3_w2_VALID (3 ms)
[       OK ] GeneratedTests.conv_3_h3_w2_SAME (10 ms)
[       OK ] GeneratedTests.conv_3_h3_w2_VALID (7 ms)
[       OK ] GeneratedTests.conv_float_2 (2 ms)
[       OK ] GeneratedTests.conv_quant8_2 (3 ms)
[==========] 20 tests from 1 test case ran. (86 ms total)
[  PASSED  ] 6 tests.

[  FAILED  ] GeneratedTests.conv_float_channels
[  FAILED  ] GeneratedTests.conv_float_channels_weights_as_inputs
[  FAILED  ] GeneratedTests.conv_float_large
[  FAILED  ] GeneratedTests.conv_float_large_weights_as_inputs
[  FAILED  ] GeneratedTests.conv_float
[  FAILED  ] GeneratedTests.conv_float_weights_as_inputs
[  FAILED  ] GeneratedTests.conv_quant8_channels
[  FAILED  ] GeneratedTests.conv_quant8_channels_weights_as_inputs
[  FAILED  ] GeneratedTests.conv_quant8_large
[  FAILED  ] GeneratedTests.conv_quant8_large_weights_as_inputs
[  FAILED  ] GeneratedTests.conv_quant8
[  FAILED  ] GeneratedTests.conv_quant8_overflow
[  FAILED  ] GeneratedTests.conv_quant8_overflow_weights_as_inputs
[  FAILED  ] GeneratedTests.conv_quant8_weights_as_inputs

14 FAILED TESTS
```

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
6 years ago[neurun] Define TypeInfo class (#2415)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 22 Aug 2018 08:53:13 +0000 (17:53 +0900)]
[neurun] Define TypeInfo class (#2415)

Define TypeInfo class to remove data type information from Shape

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoEnable to check state of model (#2411)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Wed, 22 Aug 2018 08:28:13 +0000 (17:28 +0900)]
Enable to check state of model (#2411)

This commit will enable to check model state if it is finished or not
As a result, checking `ANUERALNETWORKS_BAD_STATE` is available

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years ago[neurun] Introduce NodeVisitor for Graph operation (#2405)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Wed, 22 Aug 2018 08:26:24 +0000 (17:26 +0900)]
[neurun] Introduce NodeVisitor for Graph operation (#2405)

Introduce a NodeVisitor `neurun::graph::operation::Node` which is a
visitor interface for our graph nodes(operations). Once this class
gets used, NodeVisitor for `internal::tflite::op::Node` will no longer
be used.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[neurun] Add setInput/setOutput into graph node (#2406)
김수진/동작제어Lab(SR)/Engineer/삼성전자 [Wed, 22 Aug 2018 07:28:13 +0000 (16:28 +0900)]
[neurun] Add setInput/setOutput into graph node (#2406)

* [neurun] Add setInput/setOutput into graph node

This commit adds setInput/setOutput into graph node.

Signed-off-by: sjsujinkim sjsujin.kim@samsung.com
* change function names from set{In/Out}put to {in/out}put

* fix format in Set test

6 years agoRemove TopK tests from generated test skip file (#2401)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 22 Aug 2018 07:19:42 +0000 (16:19 +0900)]
Remove TopK tests from generated test skip file (#2401)

ACL to support TopK was enhanced so TopK tests in skip file was removed.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
6 years agoIntroduce LSTM operation into pureACL (#2408)
장지섭/동작제어Lab(SR)/Engineer/삼성전자 [Wed, 22 Aug 2018 06:45:43 +0000 (15:45 +0900)]
Introduce LSTM operation into pureACL (#2408)

This commit introduces LSTM operation into pureACL.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
6 years agoImplement RNN operation in pureacl (#2403)
장지섭/동작제어Lab(SR)/Engineer/삼성전자 [Wed, 22 Aug 2018 05:45:31 +0000 (14:45 +0900)]
Implement RNN operation in pureacl (#2403)

This commit implements RNN operation in pureacl.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
6 years ago[PureACL] NEON Fully_Connected op (#2378)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 22 Aug 2018 05:45:16 +0000 (14:45 +0900)]
[PureACL] NEON Fully_Connected op (#2378)

Added NEON FullyConnected operation. This commit was tested with #2370.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
6 years ago[neurun] Define internal datatype (#2399)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Wed, 22 Aug 2018 03:50:12 +0000 (12:50 +0900)]
[neurun] Define internal datatype (#2399)

Define internal datatype (enum class DataType)
Now enum value of DataType is exactly matched with OperandCode in NeuralNetworks.h

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoAdd nnapi_quickcheck for split operator (#2398)
윤지영/동작제어Lab(SR)/Engineer/삼성전자 [Wed, 22 Aug 2018 02:19:02 +0000 (11:19 +0900)]
Add nnapi_quickcheck for split operator (#2398)

Add nnapi_quickcheck_split_1 file to verify split operator

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
6 years agoIntroduce operation::LowerInfo (#2389)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Wed, 22 Aug 2018 01:38:09 +0000 (10:38 +0900)]
Introduce operation::LowerInfo (#2389)

Introduce `operation::LowerInfo` which contains backend id as string.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[neurun] Remove legacy tensor conversion (#2390)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Wed, 22 Aug 2018 01:34:44 +0000 (10:34 +0900)]
[neurun] Remove legacy tensor conversion (#2390)

Remove tensor conversions since we will introduce a new way.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[ACL] Update with TopK enhancement (#2375)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 22 Aug 2018 01:32:30 +0000 (10:32 +0900)]
[ACL] Update with TopK enhancement (#2375)

This commit introduces enhanced ACL with TopK.
(https://github.sec.samsung.net/RS7-RuntimeNTools/ComputeLibrary/commit/73fd00be82ebcb3cef7c858770a3aeba4d3e82c7)

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
6 years ago[PureACL] NEON Softmax (#2380)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Wed, 22 Aug 2018 01:32:08 +0000 (10:32 +0900)]
[PureACL] NEON Softmax (#2380)

This commit enables NEON Softmax operation.
Test passed: GeneratedTests.softmax_float_1 and 2, GeneratedTests.softmax_quant8_1 and 2.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
6 years ago[neurun] Simplify namespace for operand (#2381)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 21 Aug 2018 23:36:20 +0000 (08:36 +0900)]
[neurun] Simplify namespace for operand (#2381)

Simplify namespace for operand internal data structure
- Use alias for internal
- Use relative namespace

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agofully connected test fail (#2263)
PRAVEEN DORESWAMY NAIDU/System SW /SRI-Bangalore/Staff Engineer/삼성전자 [Tue, 21 Aug 2018 09:59:08 +0000 (15:29 +0530)]
fully connected test fail (#2263)

Fully connected test for float is failing.
The google test case for fully connected is failing.
The failure for rank 2 operands which might be caused due to improper handling of batches.

Signed-off-by: praveen D N <praveen.dn@samsung.com>
6 years agonnfw: Generated Test fix for reshape (#2262)
TANUJ TEKRIWAL/System SW /SRI-Bangalore/Engineer/삼성전자 [Tue, 21 Aug 2018 09:07:46 +0000 (14:37 +0530)]
nnfw: Generated Test fix for reshape (#2262)

It fixes two generated test case failure
-reshape_weights_as_inputs
-reshape_quant8_weights_as_inputs
This patch adds sanity check if Inputs is not set.

Signed-off-by: Tanuj Tekriwal <tanuj.tekri@samsung.com>
6 years ago[neurun] Introduce Graph::Iterator (#2387)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 21 Aug 2018 08:19:20 +0000 (17:19 +0900)]
[neurun] Introduce Graph::Iterator (#2387)

This commit introduces `Graph::Iterator` class to support various
kinds of iterators. The template parameter `is_const` is used to
support both const and mutable iterator without writing same code
twice.

- Introduce `Graph::Iterator`
- Introduce `Graph::PostDfsIterator` class
- Remove `Graph::iteratePostDfs()` method

Resolve #2386

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[PureACL] NEON Relu, Relu1, Relu6 operations (#2384)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Tue, 21 Aug 2018 08:16:36 +0000 (17:16 +0900)]
[PureACL] NEON Relu, Relu1, Relu6 operations (#2384)

This commit introduces Relu, Relu1, Relu6 operations.
All Relu generated test (except `relu6_quant8_2`) passes.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
6 years agoSet parameter for Floor operation in pureACL (#2383)
윤지영/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 21 Aug 2018 07:28:09 +0000 (16:28 +0900)]
Set parameter for Floor operation in pureACL (#2383)

This commit supports Floor operation for the GPU mode by using OpenCL in acl.
It sets appropriate parameters and invoke the acl floor function.

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
6 years agoUpdate skiplist of GeneratedTests (#2385)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 21 Aug 2018 06:54:49 +0000 (15:54 +0900)]
Update skiplist of GeneratedTests (#2385)

Update skiplist for passed GeneratedTests

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years ago[neurun] Introduce NOP operation (#2382)
김수진/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 21 Aug 2018 06:06:40 +0000 (15:06 +0900)]
[neurun] Introduce NOP operation (#2382)

This commit introduces NOP operation to do nothing.
(Actually it needs the tensor copy into other indexes. but not now.)

In this PR, the NOP operation doesn't anything, it's just the operation for graph manipulation test.

Signed-off-by: sjsujinkim <sjsujin.kim@samsung.com>
6 years agoAdd per-operation instrument code in pure arm compute runtime (#2362)
이상규/동작제어Lab(SR)/Principal Engineer/삼성전자 [Tue, 21 Aug 2018 06:03:10 +0000 (15:03 +0900)]
Add per-operation instrument code in pure arm compute runtime (#2362)

- Put separate copies of profiler under util/profiling
  : profile.h, profile_buffer.h, time.h, time.cc are same to original.
    It is copied to remove circular dependency.

- Add instrument code in pure arm compute runtime
  : See execution.cc

- Move profiling context to util/profiling
  : We will use Profiling Context in other runtime (i.e., neurun) later.

Related Issue: #1696

Signed-off-by: Sanggyu Lee <sg5.lee@samsung.com>
6 years ago[Generated Test] addining test for Fully_Connected (#2370)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Tue, 21 Aug 2018 01:31:54 +0000 (10:31 +0900)]
[Generated Test] addining test for Fully_Connected (#2370)

Currently all of generated tests for FC(Fully_Connected) fail.
This commit adds a test case for FC, which uses same data with nnapi_quickcheck_fully_connected_1.

The purpose of adding this test is to check whether any modification against current FC code works correctly.

(More specifically, to see if NEON FC code works for #2334)

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
6 years agoIntroduce Floor operation into pureACL (#2367)
윤지영/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 21 Aug 2018 01:09:06 +0000 (10:09 +0900)]
Introduce Floor operation into pureACL (#2367)

* Introduce Floor operation into pureACL

This commit adds Floor operation into pureACL runtime.

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
* Throw an exception for unsupported floor operation

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
6 years ago[PureACL] NEON implicit MaxPool2D & AvgPool2D operation (#2361)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Mon, 20 Aug 2018 23:33:45 +0000 (08:33 +0900)]
[PureACL] NEON implicit MaxPool2D & AvgPool2D operation (#2361)

Added NEON implicit MaxPool2D and AvgPool2D operation.
This commits makes 2 `MaxPool2D` generated test passed among 8 tests
and makes 2 `AvgPool2D` generated test passed among 8 tests

- MaxPool2D
```
$ NEON=1 LD_LIBRARY_PATH=Product/out/lib Product/out/unittest/runtime_run_android_nn_test --gtest_filter=GeneratedTests.max_pool*

[       OK ] GeneratedTests.max_pool_float_4 (42 ms)
[       OK ] GeneratedTests.max_pool_quant8_4 (1 ms)
...
[==========] 8 tests from 1 test case ran. (51 ms total)
[  PASSED  ] 2 tests.
[  FAILED  ] 6 tests, listed below:
[  FAILED  ] GeneratedTests.max_pool_float_1
[  FAILED  ] GeneratedTests.max_pool_float_2
[  FAILED  ] GeneratedTests.max_pool_float_3
[  FAILED  ] GeneratedTests.max_pool_quant8_1
[  FAILED  ] GeneratedTests.max_pool_quant8_2
[  FAILED  ] GeneratedTests.max_pool_quant8_3

```

- AvgPool2D
```
$ NEON=1 LD_LIBRARY_PATH=Product/out/lib Product/out/unittest/runtime_run_android_nn_test --gtest_filter=GeneratedTests.avg_pool*

[       OK ] GeneratedTests.avg_pool_float_5 (42 ms)
[       OK ] GeneratedTests.avg_pool_quant8_5 (1 ms)
...
[==========] 10 tests from 1 test case ran. (50 ms total)
[  PASSED  ] 2 tests.
[  FAILED  ] 8 tests, listed below:
[  FAILED  ] GeneratedTests.avg_pool_float_1
[  FAILED  ] GeneratedTests.avg_pool_float_2
[  FAILED  ] GeneratedTests.avg_pool_float_3
[  FAILED  ] GeneratedTests.avg_pool_float_4
[  FAILED  ] GeneratedTests.avg_pool_quant8_1
[  FAILED  ] GeneratedTests.avg_pool_quant8_2
[  FAILED  ] GeneratedTests.avg_pool_quant8_3
[  FAILED  ] GeneratedTests.avg_pool_quant8_4

```

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
6 years ago[neurun] Move and split Plan.h (#2369)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 20 Aug 2018 10:53:50 +0000 (19:53 +0900)]
[neurun] Move and split Plan.h (#2369)

Move and split classes in `internal/Plan.h` into separate files.

- `internal::Plan` to `neurun::codegen::Plan`
- `internal::operand::Context` to `neurun::codegen::operand::Context`
- `internal::op::Sequence` to `neurun::codegen::operation::Sequence`

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[neurun] Change namespace for internal operand (#2368)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Mon, 20 Aug 2018 10:22:24 +0000 (19:22 +0900)]
[neurun] Change namespace for internal operand (#2368)

Change namespace: internal::tflite::operand -> neurun::internal::operand

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years ago[Generated Test] TC for Concat([4D, 4D, 4D], axis=3) (#2364)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Mon, 20 Aug 2018 09:37:57 +0000 (18:37 +0900)]
[Generated Test] TC for Concat([4D, 4D, 4D], axis=3) (#2364)

Currently all of generated test for concat fail.
This commit adds a test case for concat with 4D inputs and axis=3 and passes without error.
The purpose of adding this test is to check whether any modification against current concat code works correctly.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
6 years ago[neurun] Move internal Object class into internal/operand (#2365)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Mon, 20 Aug 2018 09:23:21 +0000 (18:23 +0900)]
[neurun] Move internal Object class into internal/operand (#2365)

Move internal Object class from internal/Model.h(cc) to internal/operand/Object.h(cc)
Remove internal/Model.cc

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoRemove depthwise-xxx tests from skiplist (#2336)
이춘석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Mon, 20 Aug 2018 09:09:21 +0000 (18:09 +0900)]
Remove depthwise-xxx tests from skiplist (#2336)

Removed tests :

-GeneratedTests.depthwise_conv2d_float_2
-GeneratedTests.depthwise_conv2d_float_large_2_weights_as_inputs
-GeneratedTests.depthwise_conv2d_float
-GeneratedTests.depthwise_conv2d_float_weights_as_inputs
-GeneratedTests.depthwise_conv2d_quant8_2
-GeneratedTests.depthwise_conv2d_quant8_large
-GeneratedTests.depthwise_conv2d_quant8_large_weights_as_inputs
-GeneratedTests.depthwise_conv2d_quant8
-GeneratedTests.depthwise_conv2d_quant8_weights_as_inputs

Signed-off-by: Chunseok Lee <chunseok.lee@samsung.com>
6 years ago[neurun] Move internal Shape class into internal/operand (#2363)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Mon, 20 Aug 2018 07:57:45 +0000 (16:57 +0900)]
[neurun] Move internal Shape class into internal/operand (#2363)

Move internald Shape class from internal/Model.h(cc) to internal/operand/Shape.h(cc)

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years ago[neurun] Move internal Data class into internal/operand (#2360)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Mon, 20 Aug 2018 07:01:05 +0000 (16:01 +0900)]
[neurun] Move internal Data class into internal/operand (#2360)

Move internal Data class from internal/Model.h to internal/operand/Data.h

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoIntroduce RNN operation into pureACL (#2359)
장지섭/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 20 Aug 2018 06:08:10 +0000 (15:08 +0900)]
Introduce RNN operation into pureACL (#2359)

This commit introduces RNN operation into purACL.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
6 years agotflite_benchmark_model supports `use_nnapi` parameter. (#2337)
이상규/동작제어Lab(SR)/Principal Engineer/삼성전자 [Mon, 20 Aug 2018 05:29:44 +0000 (14:29 +0900)]
tflite_benchmark_model supports `use_nnapi` parameter. (#2337)

Previously tflite_benchmark_model always run models with interpreter only.
Now, it uses nnapi delegate on `use_nnapi=true`

```
LD_LIBRARY_PATH=Product/out/lib Product/out/bin/tflite_benchmark_model \
--graph=../models/mobilenet_quant_v1_224.tflite \
--input_layer="Placeholder" --input_layer_shape="1,224,224,3" \
--num_threads=1 --use_nnapi=true \
```

Note that it is still under development.
Instrument code for each operation is not inserted yet.

Signed-off-by: Sanggyu Lee <sg5.lee@samsung.com>
6 years ago[PureACL] Add NEON Operations (#2343)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Mon, 20 Aug 2018 04:43:04 +0000 (13:43 +0900)]
[PureACL] Add NEON Operations (#2343)

PureACL was modified to run CL or NEON operations. (Provide "NEON=1" as env variable to run NEON operations)
Three NEON operations were tested: `Add`, `Sub`, `Mul` (8 generated tests were passed).

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
6 years agoFix Shape Constraints of depthwise_conv2d_explicit for quant8 (#2326)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 20 Aug 2018 03:05:57 +0000 (12:05 +0900)]
Fix Shape Constraints of depthwise_conv2d_explicit for quant8 (#2326)

Shape constraints of depthwise_conv2d_explicit was not correct
This commit will apply correct constraints for quant8 and enable below tests

`GeneratedTests.depthwise_conv2d_quant8_large`
`GeneratedTests.depthwise_conv2d_quant8_large_weights_as_inputs`
`GeneratedTests.depthwise_conv2d_quant8`
`GeneratedTests.depthwise_conv2d_quant8_weights_as_inputs`

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years agoMake TopK support 1D, 2D tensor (#2323)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Mon, 20 Aug 2018 02:55:30 +0000 (11:55 +0900)]
Make TopK support 1D, 2D tensor (#2323)

Previous TopK supports 1D only.
This commit makes TopK support 2D also in runtime layer.
(ACL layer PR : https://github.sec.samsung.net/RS7-RuntimeNTools/ComputeLibrary/pull/41)

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
6 years ago[neurun] Encapsulate Linear IR (#2350)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 20 Aug 2018 02:55:08 +0000 (11:55 +0900)]
[neurun] Encapsulate Linear IR (#2350)

So far Linear IR was a vector of Node elements but this commit
encapsulates it into a class. Also this commit introduces a new phase
for Graph `LINEARIZED` which indicates that the Graph IR is no longer
valid and all data structures are moved to `Linear` object. For now
nothing is moved yet, but will in the future.

Resolve #2320

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[neurun] Fix typo in model.cc (#2351)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 20 Aug 2018 02:51:02 +0000 (11:51 +0900)]
[neurun] Fix typo in model.cc (#2351)

Fix typo for `or` operation in `src/frontend/model.cc`

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years agoAdd logistic_quan_1 quicktest (#2345)
윤지영/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 20 Aug 2018 02:22:36 +0000 (11:22 +0900)]
Add logistic_quan_1 quicktest (#2345)

To verify logistic quantizied operation,
add the logistic_quan_1 test in nnapi_quicktest

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
6 years agoDo not return fail for test all unittest (#2348)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Mon, 20 Aug 2018 02:20:23 +0000 (11:20 +0900)]
Do not return fail for test all unittest (#2348)

Do not return fail for test all unittest
Setting exit code makes CI job fail without report

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years ago[neurun] Add assertion for Graph building methods (#2340)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 20 Aug 2018 02:09:24 +0000 (11:09 +0900)]
[neurun] Add assertion for Graph building methods (#2340)

Add assertions for Graph building method which must be called only in
`BUILDING` phase.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[neurun] Add lower phase skeleton (#2335)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 20 Aug 2018 02:08:29 +0000 (11:08 +0900)]
[neurun] Add lower phase skeleton (#2335)

Add `lower()` method to Graph which changes Graph Phase from MODEL to
LOWERED. However it does not do actual lowering work yet.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years agoSupport quantization for Logistic activate function (#2344)
윤지영/동작제어Lab(SR)/Engineer/삼성전자 [Mon, 20 Aug 2018 02:08:17 +0000 (11:08 +0900)]
Support quantization for Logistic activate function (#2344)

This patch allows to pass the scale and zeroPoint to backend

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
6 years agoModify testcase data for TopK (#2324)
윤현식/동작제어Lab(SR)/Principal Engineer/삼성전자 [Fri, 17 Aug 2018 06:10:55 +0000 (15:10 +0900)]
Modify testcase data for TopK (#2324)

Previous testcases data was already sorted and made some TopK tests passed.
With this commit, testcase uses shuffled (not sorted) data.

Signed-off-by: Hyun Sik Yoon <hyunsik.yoon@samsung.com>
6 years agoUnify the order of inputs in STRIDED_SLICE operation & Update ACL (#2193)
장지섭/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 17 Aug 2018 06:03:08 +0000 (15:03 +0900)]
Unify the order of inputs in STRIDED_SLICE operation & Update ACL (#2193)

* Unify the order of inputs in STRIDED_SLICE operation

This commit unifies the order of inputs in STRIDED_SLICE operation.
  - add reordering bits of a scalar as the order of inputdata's shape.
  - add reordering data of a vector as the order of inputdata's shape.
  - unify the order of inputs in STRIDED_SLICE.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
* Update ACL

This commit updates acl.
  - Extend the range in which StridedSlice is supported by cl.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
6 years agoModify for testing all unittest (#2288)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 17 Aug 2018 05:56:57 +0000 (14:56 +0900)]
Modify for testing all unittest (#2288)

Fix `runall` argument type and defualt value
Fix typo `unittestall` to `unittestall_on`
Add running all unittests in `run_unittest.py`
Revise the structure of `run_unittest.py` and `run_unittest.sh` for running all unittests

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years ago[neurun] Enable operand value setting validation unittest (#2330)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Fri, 17 Aug 2018 05:35:21 +0000 (14:35 +0900)]
[neurun] Enable operand value setting validation unittest (#2330)

Enable operand value setting validation unittest
- SetOperandValue
- SetOperandValueFromMemory

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoSkip initialization for kernel and bias if their data is null (#2327)
김용섭/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 17 Aug 2018 05:21:32 +0000 (14:21 +0900)]
Skip initialization for kernel and bias if their data is null (#2327)

Skip initialization for kernel and bias if their data is null. This
situation happens when kernel's value is set at execution time.

Signed-off-by: Yongseop Kim <yons.kim@samsung.com>
6 years ago[neurun] Check invalid offset for setOperandValueFromMemory (#2328)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Fri, 17 Aug 2018 05:00:23 +0000 (14:00 +0900)]
[neurun] Check invalid offset for setOperandValueFromMemory (#2328)

Check invalid offset value for setOperandValueFromMemory
- offset is memory border over
- offset+length is memory border over

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoEnable multiplier of depthwise_conv2d (#2322)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 17 Aug 2018 04:13:16 +0000 (13:13 +0900)]
Enable multiplier of depthwise_conv2d (#2322)

Since arm compute library 18.05, multiplier of depthwise_conv2d is enabled.
By applying this feature, below tests pass
`GeneratedTests.depthwise_conv2d_float_2`
`GeneratedTests.depthwise_conv2d_float_large_2_weights_as_inputs`
`GeneratedTests.depthwise_conv2d_float`
`GeneratedTests.depthwise_conv2d_float_weights_as_inputs`
`GeneratedTests.depthwise_conv2d_quant8_2`

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years ago[neurun] Introduce codegen::Dumper for debugging (#2325)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Fri, 17 Aug 2018 04:09:44 +0000 (13:09 +0900)]
[neurun] Introduce codegen::Dumper for debugging (#2325)

Introduce `codegen::Dumper` for debugging which dumps the LIR info in
the order of execution.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[neurun] Remove legacy operand index (#2315)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 16 Aug 2018 22:19:35 +0000 (07:19 +0900)]
[neurun] Remove legacy operand index (#2315)

* [neurun] Remove legacy operand index

Remove legacy operand index
Use operand index in graph

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
* Simplify input/output index access

6 years ago[neurun] Check operand index and size before setting value (#2305)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 16 Aug 2018 09:27:12 +0000 (18:27 +0900)]
[neurun] Check operand index and size before setting value (#2305)

* [neurun] Check operand index and size before setting value

Check operand index is exist
Check operand size before setting value
Define operand object method to get operand size
Return error when index is invalid or length is different with operand size

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
* Add comment for type

6 years ago[neurun] Remove `internal::operand::Set` (#2314)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 16 Aug 2018 09:10:12 +0000 (18:10 +0900)]
[neurun] Remove `internal::operand::Set` (#2314)

Remove unused class `internal::operand::Set` which has been replaced
by `neurun::graph::operand::Set`.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years agoSupport initializing constant matrix(2d type) at compilation time (#2307)
김용섭/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 16 Aug 2018 08:33:41 +0000 (17:33 +0900)]
Support initializing constant matrix(2d type) at compilation time (#2307)

- while initializing matrix at execute time as input/output
is supported, initializing constant matrix(2d type) at
compilation time is not done so far in nnfw.
- initialize constant matrix by using matrix's View, Reader and
Iterator classes at compile time.

Signed-off-by: Yongseop Kim <yons.kim@samsung.com>
6 years ago[neurun] Implement DAG Checker (#2312)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 16 Aug 2018 08:10:42 +0000 (17:10 +0900)]
[neurun] Implement DAG Checker (#2312)

A checker that tests if given graph is a directed-acyclic-graph.
If there is a cycle the checker fails, otherwise passes.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years agoFix test data of depthwise_conv2d_float_large_2_weights_as_inputs (#2300)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 16 Aug 2018 08:10:02 +0000 (17:10 +0900)]
Fix test data of depthwise_conv2d_float_large_2_weights_as_inputs (#2300)

`depthwise_conv2d_float_large_2_weights_as_inputs` is another test version, using same data of `depthwise_conv2d_float_large_2`
This commit will fix the test data to be same

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years ago[neurun] Remove finish field in ANeuralNetworksModel (#2308)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Thu, 16 Aug 2018 07:46:21 +0000 (16:46 +0900)]
[neurun] Remove finish field in ANeuralNetworksModel (#2308)

Remove finish field in ANeuralNetworksModel
Use graph phase to check model building is finished

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years agoIntroduce testcases for reduce mean (#2302)
최성진/동작제어Lab(SR)/Principal Engineer/삼성전자 [Thu, 16 Aug 2018 07:46:02 +0000 (16:46 +0900)]
Introduce testcases for reduce mean (#2302)

* Introduce testcases for reduce mean

This commit introduces testcases for reduce_mean operation.
-4D input
-axises (width-height)
-keep dims

These are test cases for RS7-RuntimeNTools/ComputeLibrary#34

related issue:#2003

Signed-off-by: SungJin Choi lotieye.choi@samsung.com
* Modify mean_axis01_2 test case

This commit modifies mean_axis01_2 test case.

Signed-off-by: SungJin Choi <lotieye.choi@samsung.com>
6 years agoIntroduce View, Reader and IndexIterator of Matrix (#2309)
김용섭/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 16 Aug 2018 07:06:15 +0000 (16:06 +0900)]
Introduce View, Reader and IndexIterator of Matrix (#2309)

Introduce View, Reader and IndexIterator of Matrix. These classes would
be used for initializing constant matrix.

Signed-off-by: Yongseop Kim <yons.kim@samsung.com>
6 years agoAdd python script of run_frameworktest (#2085)
남궁석/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 16 Aug 2018 04:41:39 +0000 (13:41 +0900)]
Add python script of run_frameworktest (#2085)

This commit will add python script of run_frameworktest

Signed-off-by: Seok NamKoong <sk.namkoong@samsung.com>
6 years agoUpdate ACL (#2301)
장지섭/동작제어Lab(SR)/Engineer/삼성전자 [Thu, 16 Aug 2018 02:26:58 +0000 (11:26 +0900)]
Update ACL (#2301)

- Enable QASYMM8 in the CLArithmeticAdditionKernel.
- Extend the range in which StridedSlice is supported for CPU.
- Bugfix of subtensor validation.
- Unify the order of input data in StridedSliceCPU.

Signed-off-by: jiseob.jang <jiseob.jang@samsung.com>
6 years ago[neurun] Introduce Graph Phase (#2298)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 14 Aug 2018 10:14:11 +0000 (19:14 +0900)]
[neurun] Introduce Graph Phase (#2298)

`graph::Graph::Phase` is an enum class that represents which phase the
Graph is at. Since we are going to use `Graph` class for multiple
phases we need this with verification infra.

- Introduce an enum class `graph::Graph::Phase`
- Phase transition when `ANeuralNetworksModel_finish()` is called
- Call verifier from the appropriate place

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[neurun] Introduce IVerifier interface (#2296)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 14 Aug 2018 09:00:21 +0000 (18:00 +0900)]
[neurun] Introduce IVerifier interface (#2296)

Introduce IVerifier interface to support various verifications, each
verification can be reused from different graph phases.

This commit also contains a skeleton of `DAGChecker`.

Part of #2247

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[neurun] Extract TensorMarker to codegen (#2294)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 14 Aug 2018 07:54:53 +0000 (16:54 +0900)]
[neurun] Extract TensorMarker to codegen (#2294)

Extract TensorMarker into a separate file and move it to codegen
directory.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[neurun] Move operand IObject to backend (#2293)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 14 Aug 2018 07:40:26 +0000 (16:40 +0900)]
[neurun] Move operand IObject to backend (#2293)

Move `internal::Object` to `neurun::backend::operand::Object`.

Part of #2286

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[neurun] Move cpu operand Object to backend (#2289)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 14 Aug 2018 04:52:09 +0000 (13:52 +0900)]
[neurun] Move cpu operand Object to backend (#2289)

Move `internal::cpu::operand::Object` to
`neurun::backend::cpu::operand::Object`.

Part of #2286

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[neurun] Fix feature to ACL TensorShape conversion (#2290)
오형석/동작제어Lab(SR)/Staff Engineer/삼성전자 [Tue, 14 Aug 2018 04:39:28 +0000 (13:39 +0900)]
[neurun] Fix feature to ACL TensorShape conversion (#2290)

Fix feature to ACL TensorShape conversion to set batch value

Signed-off-by: Hyeongseok Oh <hseok82.oh@samsung.com>
6 years ago[neurun] Extract Planner to codegen (#2291)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 14 Aug 2018 03:47:16 +0000 (12:47 +0900)]
[neurun] Extract Planner to codegen (#2291)

Extract Planner into a separate file and move it to codegen directory.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years agoFix for Max_pool_float_* Gtest Failure (#2283)
Shubham Gupta/System SW /SRI-Bangalore/Engineer/삼성전자 [Tue, 14 Aug 2018 02:58:10 +0000 (08:28 +0530)]
Fix for Max_pool_float_* Gtest Failure (#2283)

This is fix for max_pool_float_*.Batches are now set correctly in arm_compute::tensorshape

Signed-off-by: Shubham <shub98.gupta@samsung.com>
6 years ago[neurun] Extract asXX helpers into seperate file (#2278)
김수진/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 14 Aug 2018 02:08:25 +0000 (11:08 +0900)]
[neurun] Extract asXX helpers into seperate file (#2278)

* [neurun] Extract asXX casting helpers into Cast.h

This commit extracts asXX casting helpers into Cast.h likes PACL(https://github.sec.samsung.net/STAR/nnfw/blob/master/runtimes/pure_arm_compute/src/internal/arm_compute/Cast.h)

Signed-off-by: sjsujinkim <sjsujin.kim@samsung.com>
* Change file name/add namespace/seperate source file

* Add empty line

6 years agoRename `kernel/cpufallback` to `kernel/cpu` (#2282)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 14 Aug 2018 02:04:07 +0000 (11:04 +0900)]
Rename `kernel/cpufallback` to `kernel/cpu` (#2282)

Rename the directory to remove inconsistency with namespace name in
the code and backend module(cpu backend is in `backend/cpu`).

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>
6 years ago[neurun] Remove unused internal::Model (#2284)
이한종/동작제어Lab(SR)/Engineer/삼성전자 [Tue, 14 Aug 2018 02:03:32 +0000 (11:03 +0900)]
[neurun] Remove unused internal::Model (#2284)

Now that we replace Model with Graph IR, Model is unused anymore.

Signed-off-by: Hanjoung Lee <hanjoung.lee@samsung.com>