Update nn-hal and integrate remote inference support #8
+557
−176
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This patch add validation checks for the operation which were causing segfault like: Dequantize, Div, L2Pooling2D, Less, Logistic, MaxPool2d, Mul, Quantize, ReduceMin, ReduceSum, Relu, Relu1, Relu6, Softmax, Sub, Tanh Also validation checks of few operations were modified to check for zero tensor operands, and operand lifetime Add, AveragePool2D, FullyConnected, SpaceToBatch, Squeeze, Transpose, TransposeConv2D All the above operations were causing segfault while running run cts -m CtsNNAPITestCases -t TestGenerated/GeneratedTest* Change-Id: I300d230dad1ab2f919a097d4db7a3865d7f91e11 Tracked-On: OAM-102888 Signed-off-by: Ratnesh Kumar Rai <ratnesh.kumar.rai@intel.com>
This patch disables the following operations MAX_POOL_2D, L2_POOL_2D, DEPTHWISE_CONV_2D, CONV_2D, AVERAGE_POOL_2D, L2_NORMALIZATION, RESIZE_BILINEAR RESIZE_NEAREST_NEIGHBOR Above operations were causing segfault while running CtsNNAPITestCases -t TestRandomGraph/RandomGraphTest* This operations doesn't have validation check for operand with zero values which was causing segfault in mkldnn plugin while loading the model generated by nnhal Following operartions behaviour/parameters is changed in V1.3, thus disbaling them for now L2_NORMALIZATION, RESIZE_BILINEAR , RESIZE_NEAREST_NEIGHBOR Change-Id: I4a1e230dad1ab2f919a097d4db7a3865d7f91e11 Tracked-On: OAM-102888 Signed-off-by: Ratnesh Kumar Rai <ratnesh.kumar.rai@intel.com>
moved manifest changes which enable nnhal within nnhal. Thus to disable nnhal one needs to modify only mixin.spec, by setting neuralnetworks: false Change-Id: I79ec230dad1ab2f919a097d4db7a3865d7f91e11 Tracked-On: OAM-103820 Signed-off-by: Ratnesh Kumar Rai <ratnesh.kumar.rai@intel.com>
- Changed neural networks default paths in Android.bp Tracked-On: OAM-104044 Signed-off-by: Nagamani Chennuboina <nagamani.chennuboina@intel.com> Signed-off-by: Anoob Anto K <anoob.anto.kodankandath@intel.com>
This commit reverts the disabled ops and const life operand check for the operations required for object detection Also validation check is added for DepthToSpace op Change-Id: I87da230dad1ab2f919a097d4db7a3865d7f91e11 Tracked-On: OAM-104082 Signed-off-by: Ratnesh Kumar Rai <ratnesh.kumar.rai@intel.com>
Certain operations were disabled and limitations were applied to a few other operations as part of OAM-102888 : e19bead, 0722c03. These changes were needed due to the CTS failures observed on Android_S while using NNHAL1.3. Here, we are re-enabling these operations since it is blocking many usecases like Image Classification, etc. Tracked-On: OAM-105480 Signed-off-by: Anoob Anto K <anoob.anto.kodankandath@intel.com>
Tracked-On: OAM-102337 Signed-off-by: Ratnesh Kumar Rai <ratnesh.kumar.rai@intel.com> Signed-off-by: akodanka <anoob.anto.kodankandath@intel.com>
Introducing unix socket for AI Dispatcher grpc connection along with TCP socket. Using additional vendor property vendor.nn.hal.grpc_socket_path. Tracked-On: OAM-105559 Signed-off-by: Anoob Anto K <anoob.anto.kodankandath@intel.com>
Tracked-On: OAM-105572 Signed-off-by: Anoob Anto K <anoob.anto.kodankandath@intel.com>
Tracked-On: OAM-105568 Signed-off-by: Anoob Anto K <anoob.anto.kodankandath@intel.com>
Tracked-On: OAM-105568 Signed-off-by: Anoob Anto K <anoob.anto.kodankandath@intel.com>
Tracked-On: OAM-105569 Signed-off-by: Anoob Anto K <anoob.anto.kodankandath@intel.com>
|
Jira issue not in ['bug','improvement'], make sure jira issue is of type bug or improvement. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.