Gitiles
Code Review Sign In
review.mlplatform.org / ml / android-nn-driver / acf479a1c3841b3036dc63dd84f1d57d29f94f64 / . / NnapiSupport.txt
blob: 7dfb6dd07ce41f3491bd990aaf19aa80585c4915 [file] [log] [blame]
surmeh0149b9e102018-05-17 14:11:25 +0100[diff] [blame]1------ ArmNN for Android NNAPI supported operations ------
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]2
Matthew Benthama3e23ca2019-05-13 12:49:59 +0100[diff] [blame]3This release of ArmNN for Android supports use as a driver for the Android Neural Networks API. It implements the
4android.hardware.neuralnetworks@1.0 and android.hardware.neuralnetworks@1.1 HAL interfaces.
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]5
6For more information on the Android Neural Networks API, see https://developer.android.com/ndk/guides/neuralnetworks/index.html
7
8For integration and usage documentation, please see README.md.
9
10--- Support for Android Neural Networks HAL operations ---
11
12The following AndroidNN operations are currently supported.
13
14AndroidNN operator Tensor type supported
nikraj01c6494cc2019-05-22 16:30:44 +0100[diff] [blame]15ADD (FLOAT32,QUANT8_ASYMM,QUANT16_SYMM)
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]16AVERAGE_POOL_2D (FLOAT32,QUANT8_ASYMM)
Sadik Armagan82fbeb32018-12-20 17:12:26 +0000[diff] [blame]17BATCH_TO_SPACE_ND (FLOAT32,QUANT8_ASYMM)
nikraj01c6494cc2019-05-22 16:30:44 +0100[diff] [blame]18CONCATENATION (FLOAT32,QUANT8_ASYMM,QUANT16_SYMM)
nikraj019972b402019-05-24 14:13:09 +0100[diff] [blame]19CONV_2D (FLOAT32,QUANT8_ASYMM,QUANT16_SYMM)
20DEPTHWISE_CONV_2D (FLOAT32,QUANT8_ASYMM,QUANT16_SYMM)
nikraj01c6494cc2019-05-22 16:30:44 +0100[diff] [blame]21DIV (FLOAT32,QUANT8_ASYMM,QUANT16_SYMM)
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]22FLOOR (FLOAT32)
Matthew Bentham378333d2018-10-01 16:07:27 +0100[diff] [blame]23FULLY_CONNECTED (FLOAT32,QUANT8_ASYMM)
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]24L2_NORMALIZATION (FLOAT32)
nikraj01c6494cc2019-05-22 16:30:44 +0100[diff] [blame]25L2_POOL_2D (FLOAT32,QUANT8_ASYMM)
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]26LOCAL_RESPONSE_NORMALIZATION (FLOAT32)
nikraj01c6494cc2019-05-22 16:30:44 +0100[diff] [blame]27LOGISTIC (FLOAT32,QUANT8_ASYMM,QUANT16_SYMM)
Ferran Balaguerd04c0432018-11-15 14:48:05 +0000[diff] [blame]28LSTM (FLOAT32)
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]29MAX_POOL_2D (FLOAT32,QUANT8_ASYMM)
Ferran Balaguerd04c0432018-11-15 14:48:05 +0000[diff] [blame]30MEAN (FLOAT32,QUANT8_ASYMM)
nikraj01c6494cc2019-05-22 16:30:44 +0100[diff] [blame]31MUL (FLOAT32,QUANT8_ASYMM,QUANT16_SYMM)
Ferran Balaguerd04c0432018-11-15 14:48:05 +0000[diff] [blame]32PAD (FLOAT32,QUANT8_ASYMM)
nikraj01c6494cc2019-05-22 16:30:44 +0100[diff] [blame]33RELU (FLOAT32,QUANT8_ASYMM,QUANT16_SYMM)
34RELU1 (FLOAT32,QUANT8_ASYMM,QUANT16_SYMM)
35RELU6 (FLOAT32,QUANT8_ASYMM,QUANT16_SYMM)
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]36RESHAPE (FLOAT32,QUANT8_ASYMM)
nikraj01c6494cc2019-05-22 16:30:44 +0100[diff] [blame]37RESIZE_BILINEAR (FLOAT32,QUANT8_ASYMM)
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]38SOFTMAX (FLOAT32,QUANT8_ASYMM)
Sadik Armagan82fbeb32018-12-20 17:12:26 +0000[diff] [blame]39SPACE_TO_BATCH_ND (FLOAT32,QUANT8_ASYMM)
saoste01b8471482018-10-10 09:44:51 +0100[diff] [blame]40SQUEEZE (FLOAT32,QUANT8_ASYMM)
Sadik Armagan82fbeb32018-12-20 17:12:26 +0000[diff] [blame]41STRIDED_SLICE (FLOAT32,QUANT8_ASYMM)
nikraj01c6494cc2019-05-22 16:30:44 +0100[diff] [blame]42SUB (FLOAT32,QUANT8_ASYMM,QUANT16_SYMM)
43TANH (FLOAT32,QUANT8_ASYMM,QUANT16_SYMM)
saoste01fe463152018-10-18 17:49:56 +0100[diff] [blame]44TRANSPOSE (FLOAT32,QUANT8_ASYMM)
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]45
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]46
47--- Unsupported operators ---
48
telsoa01ce3e84a2018-08-31 09:31:35 +0100[diff] [blame]49The following AndroidNN 1.0 operations are currently not supported.
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]50
51DEPTH_TO_SPACE
52DEQUANTIZE
53EMBEDDING_LOOKUP
54HASHTABLE_LOOKUP
55LSH_PROJECTION
telsoa015307bc12018-03-09 13:51:08 +0000[diff] [blame]56RNN
57SPACE_TO_DEPTH
58SVDF
59
60Where operations are not supported by the ArmNN Android NN Driver, the driver indicates this to the framework appropriately and the framework implements those operations using a CPU implementation.