summaryrefslogtreecommitdiff
path: root/nn/runtime/test
AgeCommit message (Collapse)Author
2020-10-15DO NOT MERGE ANYWHERE: temporarily remove libneuralnetworks_fuzzer from Haikuandroid-11.0.0_r46android-11.0.0_r43android-11.0.0_r40android-11.0.0_r39android-11.0.0_r38android11-qpr3-s1-releaseandroid11-qpr3-releaseKris Alder
'make haiku' is failing when this fuzzer is enabled due to b/151102177. This disables the fuzz target from running automatically for now. This also re-arranges some library sections in the build file to make the 'bpfmt' presubmit happy. Bug: 151238901 Bug: 170512504 Test: 'make haiku' and verify the fuzz target isn't built Change-Id: I2e11bd04bb1be136f7b3b0cc1ac2f80ad08a0b03
2020-08-20Fix mixed-build CTS failures.android-cts-11.0_r2Xusong Wang
The tests are newly added in R and should not be enforced for devices with only a system upgrade. Bug: 162195407 Bug: 162395335 Test: CTS with a mixed build of R + P Test: CTS with a mixed build of R + Q Change-Id: Ia0761b0b0ed9c663c262388f4ad36bdc0c2f40d1 Merged-In: Ia0761b0b0ed9c663c262388f4ad36bdc0c2f40d1 (cherry picked from commit 34ca86b85dbe2fa954f1412b3f4f81dcf2797b5b)
2020-06-25Fix bad mergeSlava Shklyaev
Change I514b78ee was supposed to replace *dynamic* tests with *unknown_rank* and *unknown_dimension* test but the old tests did not get removed for some reason. Bug: 132458982 Bug: 154597673 Test: m Change-Id: I19ed85add6d0e5a57a0b6415af6dbf63e2cc17b6
2020-06-23Merge "Add control flow tests with unknown dimensions" into rvc-devSlava Shklyaev
2020-06-22Merge "Do not schedule unknown size control flow operations to real devices" ↵Slava Shklyaev
into rvc-dev
2020-06-22Do not schedule unknown size control flow operations to real devicesSlava Shklyaev
This feature would likely have very limited vendor support in Android R. It's too late to add IF or WHILE tests to the 1.3 VTS where an inner or outer input or output operand of a control flow operation has a type that is not fully specified. To avoid exercising untested behaviour, we have decided to disallow this at the HAL level. This change ensures that we do not schedule such operations for execution on any device other than the CPU device. See http://b/159076604#comment5 and http://b/132458982#comment63. Bug: 159076604 Bug: 132458982 Test: NNT_static Change-Id: Ic5f864e6129acc4208b7751b5b182ae30a39f0a4
2020-06-22Merge "Sets bias quant scale in QLSTM test to 0 as it's unused." into rvc-devViet Dang
2020-06-22Sets bias quant scale in QLSTM test to 0 as it's unused.Viet Dang
Bug: 155421116 Test: NeuralNetworksTest_static Change-Id: I434d70e30b137a37febfab83127d79f9a6ecbae6
2020-06-17Add control flow tests with unknown dimensionsSlava Shklyaev
Bug: 132458982 Bug: 154597673 Test: NNT_static --gtest_filter='*unknown_*' Change-Id: I514b78ee3cf2efd86a47412e413a6873341a59d0
2020-06-17Merge "Remove unknown rank control flow tests from VTS" into rvc-devSlava Shklyaev
2020-06-16Remove temporary test filtersMiao Wang
Bug: 153876253 Test: mm Change-Id: I4d36ed2aeaa7666fa349b50c93397792e34fad12
2020-06-16Remove unknown rank control flow tests from VTSSlava Shklyaev
Fix: 159076604 Test: m Change-Id: I44f31a004309594f03ef7eadf10c2ff51097e954
2020-06-15Merge "Change TEST_MAPPING configuration of NeuralNetworksTest_static" into ↵Michael Butler
rvc-dev
2020-06-12Change TEST_MAPPING configuration of NeuralNetworksTest_staticMichael Butler
This change restricts presubmit to only run a single pass of NeuralNetworksTest_static ("pass 10"), corresponding to: * useCpuOnly = 0 * computeMode = ComputeMode::ASYNC * allowSyncExecHal = 1 Bug: 131770421 Test: mma Test: atest Change-Id: Ifd316ea87151ae80001208a38cb7af321f7cf9ec
2020-06-12Merge changes I8da1aa92,Ia112e46d into rvc-devSlava Shklyaev
* changes: Support WHILE with growing output tensor in CpuExecutor Relax control flow boundary operand dimension constraint
2020-06-10Support WHILE with growing output tensor in CpuExecutorSlava Shklyaev
1. Adds a WHILE test that produces a tensor of dynamic size. The loop body model has inputs and outputs of unknown rank. 2. Adds an IF test where the branch models have unkown input and output ranks. 3. Modifies the partitioner to avoid trying to delegate referenced models if an IF or WHILE or their associated referenced models have unknown dimensions. 4. Adds dynamic loop output shape support to CpuExecutor by allocating new temporaries on each iteration. Bug: 132458982 Bug: 154597673 Test: NNT_static --gtest_filter='*_dynamic*' Change-Id: I8da1aa92abf090adbdcbf647c113fe27f4c002d8
2020-06-10Relax control flow boundary operand dimension constraintSlava Shklyaev
Also updates the NDK spec to mention the constraints and adds some validation tests. Bug: 132458982 Bug: 156918813 Test: NNT_static Change-Id: Ia112e46da065a623a52ac1c402d28dcb963e5580
2020-06-02Relaxed LSH_PROJECTION: do not convert the hash tensor as internalXusong Wang
LSH_PROJECTION is very sensitive to the value of the hash tensor. Prior to this CL, AllInputsAsInternalCoverter will convert the hash tensor to internal by introducing a dummy ADD operation. Under relaxed execution mode, the small precision loss in ADD will result in a significant difference in the final result. This CL prevents the hash tensor from being converted to internal in relaxed precision tests. Additionally, this CL removes a redundant variation in lsh_projection_float16. Fixes: 155962587 Test: NNT_static Change-Id: Id5522b4949a4e3ab4801537e8eb747a25f0cd0e8
2020-05-27Merge "Use scales representable in fp16 for fp16 casting ops in RGG." into ↵Xusong Wang
rvc-dev
2020-05-26Add shape check to CAST validationLev Proleev
The generated test is added only to CTS since VTS would fail on some 1.2 drivers. Fix: 156284111 Test: NeuralNetworksTest_static Change-Id: I4f3c6cdb9f546501e4ca375d7900431c384d6885
2020-05-21Use scales representable in fp16 for fp16 casting ops in RGG.Xusong Wang
The scale in NNAPI is always a fp32 data type. As there are accelerators that only support fp16 data type, there is precision loss when executing a casting operation that converts a fp16 buffer to/from a quantized buffer. In the worst case, the precision loss in scale could be up to 0.5 ULP, and results in 0.5 * 255 = 127.5 ULP difference in the final result. This is much larger than the tolerance of these operations in RGG. This CL sets the scale of such operations to be always representable in fp16 to avoid potential precission loss. Additionally, relax the MSE requirement on DEQUANTIZE. Bug: 155842363 Test: NNT_static_fuzzing Change-Id: I511ddb6753be1dabac36e1423820a03a4f28641e
2020-05-18Merge "Fix null ptr dereference in cpu implementation of operations" into ↵Lev Proleev
rvc-dev
2020-05-16Merge "Fix incorrect (and inverted) comparison." into rvc-devStephen Hines
2020-05-15Fix null ptr dereference in cpu implementation of operationsLev Proleev
Operations fixed: * MEAN * ARGMIN/ARGMAX * STRIDED_SLICE The operations would crash when provided with inputs that resulted in an empty output shape. The change fixes the bug by making the operations output a tensor of size [1] in this case. Also, update the documentation to clarify this behaviour and move squeeze operation test to appropriate spec directory. Bug: 155508675 Bug: 155660285 Bug: 155508675 Bug: 155238914 Test: NNTest_static Change-Id: Ia865c26021dd4d781659957049dd567beeaeae99
2020-05-15Merge "Fix null ptr dereference in REDUCE_* cpu implementation" into rvc-devLev Proleev
2020-05-14Fix incorrect (and inverted) comparison.Stephen Hines
This test was incorrectly passing this comparison before because the uninitialized value in `that.channelQuant->scales` was 0xaaaa... instead of not needing to be compared at all when `channelQuant` isn't set. Now we check whether to examine `scales` at all. Bug: http://b/156464649 Bug: http://b/156514991 Test: atest CtsNNAPITestCases:TensorRankConstraint Change-Id: I1f4db99bdd63e5dc1c95f574e6eca3adfffd138d (cherry picked from commit 63deac05d3a1d53430966a49c6979554f128383d)
2020-05-13Merge "Fix CPU fallback bug with SIMPLE execution plan" into rvc-devSlava Shklyaev
2020-05-13Fix CPU fallback bug with SIMPLE execution planSlava Shklyaev
Fix: 155923033 Test: m Change-Id: Ia701c6097695fd452c408d4423998c55d823a52f
2020-05-12Merge "Avoid sending ahwb requests to 1.0 and 1.1 drivers." into rvc-devXusong Wang
2020-05-11Avoid sending ahwb requests to 1.0 and 1.1 drivers.Xusong Wang
AHWB buffer support is a feature introduced in 1.2. Prior to this CL, requests with AHWB as memory pool will be sent to 1.0 and 1.1 drivers. This CL modifies the compliance check and avoids sending ahwb to 1.0 and 1.1 drivers. Using AHWBs on compilations with 1.0 and 1.1 drivers will result in a CPU fallback. Bug: 155686276 Test: NNTS Change-Id: Ib00a2d0d24d4a8c385b5992a1168e20c4a6bb786
2020-05-11Add 1.3 support to TestPartitioning.cppPrzemysław Szczepaniak
Bug: 141704706 Test: NeuralNetworksTest_static on coral Change-Id: Icd2ee06877f6790a9dd610c501908dc110959972
2020-05-07Merge "Add AndroidTest.xml for NeuralNetworksTest_static and blacklist ↵Miao Wang
failing tests" into rvc-dev
2020-05-07Fix null ptr dereference in REDUCE_* cpu implementationLev Proleev
When asked to reduce across all dimensions, reduce would produce a zero-sized tensor without dimensions and cause segmentation fault in the implementation. The change fixes the bug by making the op output a tensor of size [1] in this case. Also, updated the bug to clarify this behaviour. Bug: 155508675 Test: NNTest_static Change-Id: Ie98d8fa2e508255fd50f6bd8184dc323ba90fac8
2020-05-07Add AndroidTest.xml for NeuralNetworksTest_static and blacklist failing testsMiao Wang
Test: mm Test: atest NeuralNetworksTest_static Fixes: 155942515 Fixes: 155942747 Fixes: 155942575 Fixes: 155942378 Fixes: 155942301 Fixes: 155942510 Fixes: 155942697 Fixes: 155942566 Change-Id: I940dda440c958fcef527d532d57b01586e070fb7
2020-05-07Merge "Fix caching bug in quantization coupling test" into rvc-devLev Proleev
2020-05-07Merge "Skip FailingDriverTest when fallback is disabled" into rvc-devSlava Shklyaev
2020-05-07Merge "Improve shape validation in FULLY_CONNECTED" into rvc-devSlava Shklyaev
2020-05-07Skip FailingDriverTest when fallback is disabledSlava Shklyaev
The test does not make sense when partitioning or CPU fallback is disabled. Bug: 155849908 Test: adb shell setprop debug.nn.partition 0 && \ NNT_static --gtest_filter=FailingDriverTest.FailAfterInterpretedWhile Test: adb shell setprop debug.nn.partition 2 && \ NNT_static --gtest_filter=FailingDriverTest.FailAfterInterpretedWhile Change-Id: Ie81eef414de3f767f4dfcf282d39ba63e366057f
2020-05-06Improve shape validation in FULLY_CONNECTEDSlava Shklyaev
1. Adds bias dimension count validation. 2. Does more validation in validate(). Bug: 155575142 Bug: 155261461 Test: NNT_static --gtest_filter="*while_fib*" Test: NNT_static --gtest_filter=ValidationTestDimensionProductOverflow2.DynamicOutputShapeOverflow Change-Id: I7b70a29e76fdf99e656ee6e3867cfd97675cfeec
2020-05-05Fix caching bug in quantization coupling testLev Proleev
Execute function in our tests among other things makes sure that the models work with caching enabled. Caching token and directory are constant and therefore the same for all tests. This is usually not a problem since the tests check only one model with a set of different configs but this was not the case for QuantizationCouplingTest suite. This suite ran signed and unsigned models in one test making them share the same caching token and directory which is an incorrect use of caching API. This CL solves this problem by removing the execution of an unsigned model from the test. This is safe since the unsigned model is still verified in GeneratedTests suite. Bug: 155604227 Test: NNTest_static Change-Id: I6aa2c8cd3fa41e37a5de53259d1fcc784f327093
2020-05-01Merge "Add more NN priority validation tests" into rvc-devMichael Butler
2020-05-01Merge "Dump RGG spec under all failures." into rvc-devXusong Wang
2020-05-01Merge "Fix null ptr dereference in squeeze cpu implementation." into rvc-devPrzemyslaw Szczepaniak
2020-04-30Add more NN priority validation testsMichael Butler
Bug: 147925145 Test: mma Test: NeuralNetworksTest_static Change-Id: Ie636c7cdc84ff7a262925f8ff8ebbd4d4f45201b
2020-05-01Merge "Add internal NNT_static tests for buffer copying." into rvc-devXusong Wang
2020-05-01Merge changes I3e1eb543,I012eb4df into rvc-devMichael Butler
* changes: Create tests for VersionedInterfaces errors Simplify IDevice reboot logic
2020-04-30Add internal NNT_static tests for buffer copying.Xusong Wang
This helps cover the code path of data copying between hidl memory and IBuffer, as well as data copying between IBuffers. These code paths may not be covered by CTS because of the lack of driver support. This CL additionally extracts TestAshmem class from TestGenerated to a common TestUtils. Bug: 152209365 Test: NNT_static Change-Id: I617bfbc391c7d0ada0c32b31ee2e6e493d5dc6a2
2020-04-30Dump RGG spec under all failures.Xusong Wang
Fixes: 155195637 Test: NNT_static_fuzzing and inspect the dumped spec Change-Id: I11b50433b17afeceedae954f2832c1c2193b2c7f
2020-04-30Merge "Fix NNAPI TimingTests" into rvc-devMichael Butler
2020-04-30Merge changes I3e037df0,Ifd1f45a2 into rvc-devXusong Wang
* changes: Regenerate tests. Re-enable tests that are previously disabled.
generated by cgit v1.2.3 (git 2.25.1) at 2024-07-05 23:28:26 +0000