| 877baf0d0b19e97bfc9a55c90c2d61df7ffb6025 |
|
04-Jun-2018 |
Mika Raento <mikie@google.com> |
Openmp blocktime 1->20ms to keep cores enabled
With 1ms blocktime the benchmark app shows 3x performance degradation on
Pixel 2 in float benchmarks (against no blocktime setting).
In testing, 10ms was enough to give results comparable to no setting.
20ms was chosen as to have a 2x margin of safety while not keeping the
threads busywaiting unnecessarily long.
Bug: 109645291
Test: NeuralNetworksTest_static/*Openmp*
Test: Run benchmark app multiple times
Merged-In: Id2a909f6aefdbe29dd746019c2c9541e31d8c754
(cherry picked from commit fed420cab2704afb2beb69577dbe75392954a57e)
Change-Id: Ie138b11c6713c10489c3aa7e946ac2bf47a3e87b
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 6a2a11dc1edaa61647c35650d7dbdd8893d6ce8b |
|
03-May-2018 |
Mika Raento <mikie@google.com> |
Reduce OpenMP thread busywait time (200 to 1 ms)
Reduce the time OpenMP threads used by Eigen busywait for new work after
completing a piece. This is 200ms by default, which meant that we tried
to continuously run 8 threads for 200ms after completing operations
using Eigen - starving both our other work and the rest of the system.
Bug: 79159165
Test: mm
Test: cherry-picked to systrace branch and look at traces
Test: NeuralNetworksTest_static
Change-Id: I3f21aea6d96c785dd7c47493a739b76876420f7d
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 64031fa26110e6d57896bfdbe1cd58e69cde3dbb |
|
11-Apr-2018 |
Miao Wang <miaowang@google.com> |
Remove pointer values from the log and hide certain verbose logging
unless building with NN_DEBUGGABLE.
Bug: 77605132
Test: mm
Test: NeuralNetworksTest_static on walleye-eng with "debug.nn.vlog 1"
All tests pass and complete verbose logging are seen in logcat.
Test: NeuralNetworksTest_static on walleye-user with "debug.nn.vlog 1"
All tests pass and *no* pointers showing in NNAPI verbose logging.
Change-Id: Icc7357cded3ec5c197e9a4418d327b10c2ac7178
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| aba8381570d7cc4b5431d741132d6f80519201eb |
|
01-Feb-2018 |
Michael Butler <butlermichael@google.com> |
Upgrade NeuralNetworks v1.0 to v1.1 followup CL
This CL addresses some remaining comments from ag/3499102.
Specifically, this CL:
1) Overloads CpuExecutor::run to work with both V1_0::Model and V1_1::Model
2) Documents all VersionedIDevice methods
3) Captures specific variables by reference in lambdas (instead of the general [&])
4) Forward-declares VersionedIDevice instead of including the header.
Bug: 63911257
Test: mm
Test: cts and vts test binaries
Change-Id: I457150346638ea8a2be912046c2bc3c49e95b44e
(cherry picked from commit 8ea26fd92ff697734e5ba690b61f9b0aeda2abf2)
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| b66b5abcef40e88df6ac18ce7890372192102323 |
|
02-Apr-2018 |
Miao Wang <miaowang@google.com> |
Finalize STRIDED_SLICE
- Add documentation for begin_mask, end_mask, shrink_axis_mask.
- Add support for shrink_axis_mask, and corresponding tests.
- Add validation tests for STRIDED_SLICE.
- Updated existing tests and re-generated CTS and VTS tests.
Bug: 73661082
Test: NeuralNetworksTests_static
Change-Id: I49ad2ffcb3b7496ee1225618833ba324bb01c5b5
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| a9ce0001ca1df32fde3f322c385eef8ce7279329 |
|
23-Mar-2018 |
Miao Wang <miaowang@google.com> |
Remove unused / unimplemented parameter "crops" for BATCH_TO_SPACE
Bug: 73661175
Test: mm
Test: NeuralNetworksTests
Change-Id: If4b50c50e0847ce36016f6c257aac3b9d418c7b2
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 4d8c48b3cc55c0748ef87a0ba8bbbd3959680213 |
|
30-Jan-2018 |
Miao Wang <miaowang@google.com> |
Allow CpuExecutor to execute the new ops.
- Add basic tests for all the newly added operations.
Bug: 63911257
Test: mm
Test: NeuralNetworksTest (CTS) including new tests pass
Change-Id: I78124e5d5dac6824da678342a02792439d1c725e
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 4cb4fa2fe863a8f404c0448da160dd5bb4187db4 |
|
01-Dec-2017 |
David Gross <dgross@google.com> |
"Canonicalize" class RunTimePoolInfo.
This class manages an important resource (an mmap() memory) region.
Rather than treating the class as a bundle of state with methods to
explicitly manipulate that resource, give it the full C++ concrete data
type treatment: Constructors, destructor, and move assignment, with all
data private. Now the lifetime of the managed resource matches the
lifetime of the class instance that manages the resource: The resource
is freed when the instance is destroyed or assigned to.
Test: nn/runtime/test
Bug: 69685100
Bug: 69929024
Change-Id: Ib53a4c70d8db496d69a584d501f0aa6f74ddea99
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 06b5a0a85abe10b618a3a5ed023ac5c74678647e |
|
30-Nov-2017 |
David Gross <dgross@google.com> |
Fix RunTimePoolInfo leak of shared memory regions.
SamplePreparedModel and StepExecutor call RunTimePoolInfo::set(), which
calls mmap(), but there is no corresponding call to munmap(). The fix
is to add a new method runTimePoolInfo::release(), and call it from the
appropriate points.
Without this fix, even if we don't run out of virtual address space,
we can still run up against a kernel limit in the number of mapped regions
("/proc/sys/vm/max_map_count").
ALSO: Add a missing check for HIDL transport error (isOk()).
Test: nn/runtime/test
Bug: 69685100
Change-Id: If6ac1396e89cd738738a6ae3397450bd1d80f506
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 1da8fed77c5c296afa18f754ec3616e7f02a4cfd |
|
12-Oct-2017 |
Jean-Luc Brouillet <jeanluc@google.com> |
Pass large model operand values in shared memory.
Because of a limitation in HIDL, large tensor values (e.g. weights)
need to pass in shared memory rather than as HIDL arguments. This
prevented large models from running on a driver.
Separated the handling of memory pools so that request and model
pools are not comingled.
Also improve logging so we see more details about the models when
logging.
Bug: 67603060
Test: Ran system tests and VTS tests.
Change-Id: I760e31275699f9306c4b21945935dc3a4ca68754
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 915cca2dab8ee7a05fec2905f11ec3664be49c6e |
|
06-Oct-2017 |
Yang Ni <yangni@google.com> |
NEON support for LSTM
Bug: 67716699
Pulled in the latest implementation of tensor utils from TF Lite,
which has enabled NEON for tensor maths.
Test: NeuralNetworksTest
Change-Id: Ic1abbb40c1a8f9c218ea20e58108c6f88e044369
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| bee07f73a5f998a2dd6dc581e7776557c21f9684 |
|
09-Oct-2017 |
Miao Wang <miaowang@google.com> |
Force CpuExecutor validating user-provided model output operands.
- For operands with OperandLifeTime::MODEL_OUTPUT, the dimensions,
type, and other meta-data must match the output Shape calculated
from the operation preparation step.
- Fix the ill-defined tests caught by the added validation.
- Incidental changes: generated more tests from tests specs.
Bug: 67390841
Test: NeuralNetworksTests pass
Change-Id: I40d35db0f7a868feae773dbf7e12cf4bf5f5e275
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 820215d28bed6c90f696cde0f282445d16da432e |
|
05-Oct-2017 |
Miao Wang <miaowang@google.com> |
Allow fine grain verbose logging control through system property.
- Existing DEBUG and INFO change to the new mechanism, existing ERROR and
WARNING unchanged.
- By default, all verbose logging are disabled.
- setprop debug.nn.vlog 1 enables all verbose logging.
- setprop debug.nn.vlog "tag1 tag2 ..." only enable the selected tags.
- Available tags:
all: same as 1.
model: enable logging with tag MODEL.
compilation: enable logging with tag COMPILATION.
execution: enable logging with tag EXECUTION.
cpuexe: enable logging with tag CPUEXE.
manager: enable logging with tag MANAGER.
driver: enable logging with tag DRIVER.
Bug: 63905942
Test: mm
Test: NeuralNetworksTests pass
Test: manually set property to allow individual module to logging
verbosely
Test: tested with debug.nn.partition default -> 0 -> 1 to make
sure getProp still works fine
Change-Id: Iaa0ffa62176dabcdc35058748597df720fd6e47e
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| eefb1e60444afd08a4350e11f281ac7064ebba63 |
|
04-Oct-2017 |
Yang Ni <yangni@google.com> |
Change stateful Ops to stateless ones
Bug: 63905942
Updated Ops RNN, LSTM, and SVDF.
Split outputs used for states into inputs and outputs.
Test: NeuralNetworksTest
Change-Id: Ia3d11f640cba4cab1b94d0b9746c46d347c024a4
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 62cc2758c1c2d303861e209f26bddcf4d7564b73 |
|
28-Sep-2017 |
Jean-Luc Brouillet <jeanluc@google.com> |
Fix problem passing null optional operands.
Allow optional operands to be specified in the model as having no value.
Modify the HAL to allow passing missing operands. Change the validation
of most operations to verify that none of their operands are missing.
Bug: 63905942
Test: Compiled and ran runtime/test.
Change-Id: I378bc51847a8fb30cb2d36c9bf707ebadc79289b
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 93e086fbc0f1577eabdffd0d3420589f2788bd95 |
|
02-Oct-2017 |
Miao Wang <miaowang@google.com> |
Remove activation parameter from CONCATENATION
Bug: 67185615
Test: mm
Change-Id: I3555259305761691ea65d2fff406d7fcd8ee8eb5
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 8c689bdfc391e47854ec27bad0f9d685135af253 |
|
27-Sep-2017 |
Yang Ni <yangni@google.com> |
Add preparation code missed in some OPs
Bug: 63905942
This updates speech model related OPs: Embedding Lookup,
Hashtable Lookup, LSH Projection, LSTM, RNN, and SVDF.
Deduce the shape of an output or intermediate tensor.
Allocate memory if needed (i.e., not set by user).
Validate inputs (dimensions, etc.)
Minor cleanups related to the utility functions shared by these ops.
Also made necessary fixes and implementations as discovered while
doing this.
Test: NeuralNetworksTest
Change-Id: I7c0b789edb4d3565526d58905609d2023c1a5351
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 850d1eb131d62b07438ff5869bdb80e09fada766 |
|
30-Sep-2017 |
Miao Wang <miaowang@google.com> |
Allow implicit padding
- Upadte the documentation to support implicit padding.
- Removed error logging for implicit padding input.
Bug: 63905942
Test: mm
Test: NeuralNetworksTest pass
Change-Id: I5149433ccb393ed390ad9e24013249ef37c6aba8
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 25f0f2dcd7049cd9f8a60133697dff114bbbf00e |
|
28-Sep-2017 |
Miao Wang <miaowang@google.com> |
Allow ADD and MUL taking quantized input.
- Implement the quant8 path for ADD and MUL.
- Add tests for quant8 ADD and MUL, both normal and broadcast version.
- Add validation for quantization parameters, forcing scale >= 0 and
zeroPoint in range [0, 255] for QUANT8_ASYMM type operands.
Bug: 63905942
Test: mm
Test: NeuralNetworksTest pass
Change-Id: Ib643217737af9624a079369e0aea792025223fbb
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 93ea4104cfd9108b67bb0a641cee2b8cbbe01129 |
|
25-Sep-2017 |
TreeHugger Robot <treehugger-gerrit@google.com> |
Merge "Remove operationTuple." into oc-mr1-dev
|
| 7d6ac906f000f3afe418e92c0a4ae36b2ea1143e |
|
23-Sep-2017 |
Jean-Luc Brouillet <jeanluc@google.com> |
Remove operationTuple.
Removed operationTuple from Model in the HAL, as the data type was
redundant information.
Removed supportedOperationTuples from Capabilities, as real drivers need
more complex restrictions than this provided. For the OC-MR1, we'll just
rely on getSupportedNodes.
Also removed the unused cachesCompilation.
Bug: 63905942
Test: Compiled and ran tests, incl. vts.
Change-Id: I84433d1e0e5211ac2f77317f3752171d85a5ab04
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 874d039215516aebdaba2e242609199897fe80c0 |
|
23-Sep-2017 |
Miao Wang <miaowang@google.com> |
Fix sigmoid and softmax tests and implementation.
- CPU executor now checks both the scale and offset for the
output Shape.
- The golden references and output range for the tests are updated.
Bug: 63905942
Test: mm
Test: NeuralNetworksTest pass
Change-Id: I9e892ae0de8ea17298dbb7edb96036e1d30c84fb
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| feb29cb5007bd968e8850e51fab95ea24eccfd90 |
|
18-Sep-2017 |
Miao Wang <miaowang@google.com> |
Fix implementation bug in Dequantize
Bug: 65826935
Test: dequantize in CTS test and NeuralNetworkTests pass
Change-Id: I13d32f9f83c22b0ba342b46dfe466bbba4831d98
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| a5c16e4e73a34e6776a8f071ab884444c28b6893 |
|
14-Sep-2017 |
Miao Wang <miaowang@google.com> |
Rename ANeuralNetworksOperandType.offset to zeroPoint
Bug: 63905942
Test: NeuralNetworksTest
Change-Id: I15464657dc4eed9adbfc444df306cd0062993be3
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 05292dc440641a38fab6a07c1372ce6780816c1c |
|
13-Sep-2017 |
TreeHugger Robot <treehugger-gerrit@google.com> |
Merge "Update ANeuralNetworksMemory related API" into oc-mr1-dev
|
| 42dc6a6cd68877cd85e3bc475b41bda0fd946c41 |
|
13-Sep-2017 |
Miao Wang <miaowang@google.com> |
Update ANeuralNetworksMemory related API
- Remove ANeuralNetworksMemory_createShared.
- Remove ANeuralNetworksMemory_getPointer.
- Allow a offset specified for
ANeuralNetworksMemory_createFromFd.
- Update the tests.
Bug: 63905942
Test: mm
Test: updated NeuralNetworksTests pass.
Change-Id: I1fd7362ec379f5f855c71c6758e0b647be9aaa76
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 33cbb4260ee16456a9c698c87aabb0f048f9cd88 |
|
13-Sep-2017 |
TreeHugger Robot <treehugger-gerrit@google.com> |
Merge "Add implicit padding support" into oc-mr1-dev
|
| 3cc685b32dee7f28504fb1bec633c3cead137538 |
|
12-Sep-2017 |
Miao Wang <miaowang@google.com> |
Add implicit padding support
- Transient change, will be removed.
Bug: 63905942
Test: mm
Test: Converting the test with implicit padding pass.
Change-Id: I888bf3daa42ec2394b0fed1522c0aa353f5ddbaa
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 24bb62867d702f8d4ea42ae794a96d7383c510ac |
|
12-Sep-2017 |
Miao Wang <miaowang@google.com> |
Fix the bug caused by recent refactor that zeroPoint is not correctly
set.
Bug: 63905942
Test: mm
Test: NeuralNetworksTest pass
Change-Id: Ic5c49e8299c43dca8fe36754dd4c2dea517562d5
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 7c9cca3ae67378313c586e010f32fb29e648da95 |
|
12-Sep-2017 |
TreeHugger Robot <treehugger-gerrit@google.com> |
Merge "Add documentation for ops" into oc-mr1-dev
|
| 85e34644847dfc0c210520a21217aeb6269f3d3d |
|
10-Sep-2017 |
Miao Wang <miaowang@google.com> |
Add documentation for ops
This CL adds/updates documentation for the following ops:
ADD
AVERAGE_POOL
CONCATENATION
CONV_2D
DEPTHWISE_CONV_2D
DEPTH_TO_SPACE
DEQUANTIZE
FLOOR
FULLY_CONNECTED
L2_NORMALIZATION
L2_POOL_2D
LOCAL_RESPONSE_NORMALIZATION
LOGISTIC
MAX_POOL_2D
MUL
RELU
RELU1
RELU6
RESHAPE
RESIZE_BILINEAR
SOFTMAX
SPACE_TO_DEPTH
TANH
Bug: 63905942
Test: mm
Change-Id: I23818067bebd1c510ed08b9ac466b7cfb28de039
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| b17915aedf050e189a6cfbed85529edf3cbdbd74 |
|
11-Sep-2017 |
TreeHugger Robot <treehugger-gerrit@google.com> |
Merge "Making padding explicit, instead of SAME / VALID" into oc-mr1-dev
|
| 6cd685f64bd82c003b8d0943fc6b7b8e0730b939 |
|
10-Sep-2017 |
Miao Wang <miaowang@google.com> |
Making padding explicit, instead of SAME / VALID
- Use explicit padding {padding_left, padding_right, padding_top,
padding_bottom} for applicable ops.
- Updated the corresponding tests.
Bug: 63905942
Test: mm
Test: updated NeuralNetworkTest pass
Change-Id: Ifb8ccf59fbe5766dffe1537aa5357192a7c88a70
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| a4e2ee81d015b5e26897d18b18dd50044faba62f |
|
10-Sep-2017 |
Jean-Luc Brouillet <jeanluc@google.com> |
Refactor how arguments are tracked.
Use an explicit description of the lifetime of an operand
rather than relying on poolIndex to do the job.
Bug: 63905942
Test: Ran unit tests and test.py
Change-Id: I8ce007c439ccc99b1e04cce412ba6c93453c7a02
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 55cef2193279a646292fdeb355399627dd8990c8 |
|
09-Sep-2017 |
Miao Wang <miaowang@google.com> |
Explicitly add "_2D" to applicable ops, and add fused activation
function enum.
- Add the fused activation enum to explicitly set the expectation that
only the included ones are supported for fused ops.
- Add "_2D" to CONV, DEPTHWISE_CONV, AVG_POOL, MAX_POOL, and L2_POOL,
since they are only supposed to do 2D computation.
- Updated the tests and the stack to plumbing the changes through.
Bug: 63905942
Test: mm
Test: NeuralNetworksTests pass for both CPU path and sample hidl driver.
Change-Id: I60aeabab1d193d26d938c0d1d3ffdd519740d222
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| fce2d202e5dd6f4b2193d84d8726ff4ba86e2fc8 |
|
09-Sep-2017 |
TreeHugger Robot <treehugger-gerrit@google.com> |
Merge "Add support for mmaped file descriptor" into oc-mr1-dev
|
| 105807d963d969197fe78185ed588bfad3dc0ea5 |
|
05-Sep-2017 |
Miao Wang <miaowang@google.com> |
Add support for mmaped file descriptor
- create new API to create ANeuralNetworks_Memory from a mmaped fd.
- add the corresponding Memory constructor to NeuralNetworksWrapper.h
- Add unit test TestFd for the new API.
Bug: 63905942
Test: mm
Test: TestFd pass with CPU path.
Test: TestFd pass on sample HIDL driver, with ag/2588539
Change-Id: I1fcb0f0bd01e58bd592e18eff11c27034ee4a4c3
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 0831219b2763e7906d0f42bd15351a8e05c6f0ab |
|
01-Sep-2017 |
Yang Ni <yangni@google.com> |
Move operation files into subdirectory
Bug: 63905942
The convention seems to have all CPU implmentation of OPs in the
subdirectory nn/common/operations.
This CL moves recently added speech/text related OPs to that
directory.
Test: unit tests of all moved ops on marlin
Change-Id: I026aa94c58f6ee4e57e519781ee92d5792106de6
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| f23352521cb0834f33f3802b7b2626db64d0eea3 |
|
22-Aug-2017 |
Miao Wang <miaowang@google.com> |
Making Add and Mul able to handle broadcast Add/Mul.
Bug: 63905942
Test: mm
Test: BroadcastAddTwo, BroadcastMulTwo
Change-Id: I41460be25ab812069fff31abfbd1f6ab66b7aba2
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 9b8414da39fe772ec9d84c5bc5b101ceef001567 |
|
31-Aug-2017 |
TreeHugger Robot <treehugger-gerrit@google.com> |
Merge "Add OEM_OPERATION enum." into oc-mr1-dev
|
| 681ee0e3a438d807f34e97bc5a49047d5b7463a0 |
|
31-Aug-2017 |
Yang Ni <yangni@google.com> |
Merge "Implement LSTM cell on CPU" into oc-mr1-dev
|
| a67d8a0e28baef2fab57a2a3255461869abed74d |
|
30-Aug-2017 |
Miao Wang <miaowang@google.com> |
Add OEM_OPERATION enum.
Bug: 63905942
Test: mm
Test: All unit tests passing on sailfish
Change-Id: I391ca039620be6525d95ad581acdb6e11d091af2
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 8dcf8b31f39001b8463f1512c1c7de568361b963 |
|
31-Aug-2017 |
Yang Ni <yangni@google.com> |
Merge "Implement SVDF on CPU" into oc-mr1-dev
|
| 3c35ae1db59fedf4b4be0ee6e9ac3ce586820d7c |
|
29-Jul-2017 |
Yang Ni <yangni@google.com> |
Implement LSTM cell on CPU
Bug: 63905942
Adapted previous TF Lite implementation and unit tests.
Modified the API implmentation to allow input tensor to have a null
buffer pointer to indicate an optional input.
Test: adb shell /data/nativetest64/lstm_test/lstm_test
Change-Id: Iab89fe6c7731dc68c8460f30630174c55e3b24df
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 7a5429d307673470c2ee5940d5a3d8f483b5331e |
|
11-Aug-2017 |
Yang Ni <yangni@google.com> |
Implement SVDF on CPU
Bug: 63905942
Adapted previous TF lite implementation and unit test for
implementation of NN API on CPU.
Test: adb shell /data/nativetest64/svdf_test/svdf_test
Change-Id: I86c39c228f1075686fd3e7303f472a3a4b1409ca
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 15b59e721efe1a75b0e8cde8bd44bbb532446e13 |
|
12-Aug-2017 |
Yang Ni <yangni@google.com> |
Implement EMBEDDING_LOOKUP and HASHTABLE_LOOKUP
Bug: 63905942
Adapted previous TF Lite implementation and unit tests for NN API on
CPU.
Test: adb shell /data/nativetest64/embedding_lookup_test/embedding_lookup_test
Test: adb shell /data/nativetest64/hashtable_lookup_test/hashtable_lookup_test
Change-Id: I6e4a9d8c5d0d47e66e3fa771967da3526d19ac4b
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| f1817c663af4f22bc089ef82cd50df4186422c42 |
|
23-Aug-2017 |
Yang Ni <yangni@google.com> |
Implement LSH Projection for NN API on CPU
Bug: 63905942
Adapted previous TF Lite implementation and unit tests.
Test: adb shell /data/nativetest/lsh_projection_test/lsh_projection_test
Change-Id: Id90192becfec11a658d57495831327dbb380fc0a
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| b8810af67d4efbd966d6788c56cd01c5aa06a71e |
|
30-Aug-2017 |
TreeHugger Robot <treehugger-gerrit@google.com> |
Merge "Add full support for Memory pools." into oc-mr1-dev
|
| 8b99bb1d98a42b67ba1c00e12c7abb3708cf7c05 |
|
21-Aug-2017 |
Jean-Luc Brouillet <jeanluc@google.com> |
Add full support for Memory pools.
Weights can now be passed via memory pools.
Test: Compiled and ran the unit test.
Bug: 63905942
Change-Id: I997a10f6102dcae7c6876c96a4f6f913eaf54c74
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 39ac3cd6199ba1abcacc5781a3c9f60eac321bc8 |
|
29-Aug-2017 |
Yang Ni <yangni@google.com> |
Merge "Implement RNN in NN API" into oc-mr1-dev
|
| 9f41362ea2d250b89e59c73cc0194c6a6720cc9d |
|
21-Aug-2017 |
Miao Wang <miaowang@google.com> |
Implement quantized RELU, RELU1, RELU6 and SOFTMAX.
Bug: 63905942
Test: mm
Test: InceptionV3 quantized end to end test pass
Change-Id: Ibfdfa5d2476f6ee5f2f96aab78c522eeebcc5920
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 7743646307dbcdfc3ed5188b46f1dfe34b69272c |
|
21-Aug-2017 |
Yang Ni <yangni@google.com> |
Implement RNN in NN API
Bug: 63905942
Adapted previous TF lite implementation and unit test for
implementation of NN API on CPU.
Test: adb shell /data/nativetest64/rnn_test/rnn_test
Change-Id: Iae724391f902c806e987a6dff50c2c4b24c993ce
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| f6e79096ad149a9557c6f5c7753ac75e05155a79 |
|
11-Aug-2017 |
Miao Wang <miaowang@google.com> |
Implement the following ops reshaping a tensor.
- RESHAPE, both FLOAT32 and QUANT8.
- DEPTH_TO_SPACE, both FLOAT32 and QUANT8.
- SPACE_TO_DEPTH, both FLOAT32 and QUANT8.
- RESIZE_BILINEAR, FLOAT32.
Bug: 63905942
Test: mm
Change-Id: Ie44e9e6877e511f4c28671161f8d5a7420a7a35d
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 15b1e5b3ef28dc43cfd68143119e3a2967d2c557 |
|
10-Aug-2017 |
Miao Wang <miaowang@google.com> |
Implement ADD, MUL, FLOOR and DEQUANTIZE.
Bug: 63905942
Test: mm
Change-Id: Idd40cee44b491a395e6cb417f3ae490e29017725
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| d9bef406eb022b143bb2fa7d98bc4b4cdc79de06 |
|
10-Aug-2017 |
Miao Wang <miaowang@google.com> |
Implement L2_NORMALIZATION and LOCAL_RESPONSE_NORMALIZATION.
Bug: 63905942
Test: mm
Change-Id: I2c3d41232279a64169519bf97e6d2bacb20fd2a0
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| bbfd239e43526ff969699d3fc6110395edd2108b |
|
26-Jul-2017 |
Miao Wang <miaowang@google.com> |
Implement Softmax, FullyConnected, and Concatenation.
- FLOAT32 and QUANT8 versions of FullyConnected and Concatenation are
covered.
- This change only covers FLOAT32 version of Softmax
Bug: 63905942
Test: mm
Test: end-to-end test with Inception V3 pass.
Test: end-to-end test with StripOCR quantized pass.
Change-Id: I9e001265cc31df406fdb3d685d10a1c61216c700
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 27e9be3904b034e422ee9b6ab70b35ea994d2b39 |
|
03-Aug-2017 |
Miao Wang <miaowang@google.com> |
Initial implementation of the following quantized ops.
- CONV_QUANT8
- DEPTHWISE_CONV_QUANT8
- AVERAGE_POOL_QUANT8
- MAX_POOL_QUANT8
- LOGISTIC_QUANT8
Additionally, added functions to plumb through quantization
parameters.
Bug: 63905942
Test: mm
Test: end-to-end MobileNet quantized test pass
Change-Id: Ib2753c68bf2c51467ae1c158b45541bcfdf10789
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| eb1f88846f147d1d80ee0d688fe4635b89a40ffa |
|
26-Jul-2017 |
Miao Wang <miaowang@google.com> |
Implement the following operations for Android NN runtime.
- CONV_FLOAT32
- DEPTHWISE_CONV_FLOAT32
- AVERAGE_POOL_FLOAT32
- L2_POOL_FLOAT32
- MAX_POOL_FLOAT32
- RELU_FLOAT32
- RELU6_FLOAT32
- TANH_FLOAT32
- LOGISTIC_FLOAT32
Bug: 63905942
Test: mm
Test: End-to-end test with MobileNet pass
Change-Id: I3eaa9794c7cdffd01792d26c5d6497c8d56d8605
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 707dbd2d55f5dacf78ffb3ad7c8b3f37c2e9d758 |
|
25-Jul-2017 |
Jean-Luc Brouillet <jeanluc@google.com> |
Pass the model as an HIDL component.
Major rework where instead of serializing the model and
passing it in a shared memory, we pass it via HIDL.
The runtime/test code does two run, one goes through the
default CPU path, the other through any available driver.
Note: The code in hardware/../sample was written by Michael Buttler.
Bug: 63905942
Test: Runs both the CPU path and through the sample driver.
Change-Id: Ie3ee975d33056ba299895b13193f4698a690dd04
/frameworks/ml/nn/common/CpuExecutor.cpp
|
| 96775128e3bcfdc5be51b62edc50309c83861fe8 |
|
12-Jul-2017 |
Jean-Luc Brouillet <jeanluc@google.com> |
First implementation of the Neural Networks API.
This first version can run a simple query on the CPU either
via the fallback path or through a simulated driver.
This code has many deficiencies: single threaded, not all
validation are done, not going through HIDL, and not
enough unit tests. Expect more changes!
Test: Compiled and ran the unit tests
Change-Id: I9f6a485a2e7207aeb5f91a2904dcb4b7fd8a6f65
/frameworks/ml/nn/common/CpuExecutor.cpp
|