Age | Commit message (Collapse) | Author |
|
Refactor supported operators by breaking out model semantics
into its own class. Model semantics checked right after model
read.
Signed-off-by: Jonas Ohlsson <jonas.ohlsson@arm.com>
Change-Id: If442b189efcd91dda01af60b2b3adedfacdf2fad
|
|
Remove quant_values attribute from Tensor class.
It only needs a single values attribute, holding either
quantized or unquantized values as appropriate.
Change-Id: Ie96f80ac58061b6077e0f7048dc60209fdfbcafa
Signed-off-by: James Peet <james.peet@arm.com>
|
|
Added basic TOSA support, enabling Vela to
read and compile a .tosa file corresponding to
CONV2D + Rescale + Clamp, and writing it to an
optimized .tflite file.
The optimized .tflite file, will in this case, hold
a commandstream where the Rescale and Clamp has been
fused into the CONV2D.
The optimized tflite file is not output from Vela.
-Added support to read .tosa file into Vela
internal structure.
- Added tosa_reader.py, tosa_mapper.py and
helper files stored under tosa/
- Support for this limited to ~10 ops
-Added reader_util.py for functions common
for TOSA and TFLite
-Added tosa_graph_optimiser.py
-Added support to fuse Rescale into convolution
-Modified handling for padding
-Added support to fuse Clamp to previous op
-Added graph_optimiser_util.py
-Moved functions common for TOSA/TFLite graph
optimization to this file.
-Renamed graph_optimiser.py to tflite_graph_optmiser.py
-Added separate tosa_supported_operators.py
-Added supported_operator_util.py
-For functions in common for TOSA/TFLite
Signed-off-by: Patrik Gustavsson <patrik.gustavsson@arm.com>
Change-Id: Ic3c540504ec8c5eb4771397fdc6882050ecf33ab
|
|
When a MEAN operator with a single reduction axis
specifies the axis index attribute as an array with
a single element rather than a scalar index, the
operator is placed on the CPU even though it is
technically supported.
This commit fixes this issue and also adds some new
tests for the axis constraints.
Signed-off-by: Dwight Lidman <dwight.lidman@arm.com>
Change-Id: Ia287f3b9cc80a805e972cd4b2962e52526a8dc16
|
|
This is a small commit which changes one of
the four MEAN implementations to a simpler
one, using an AvgPool instead of a
DepthwiseConv.
Signed-off-by: Dwight Lidman <dwight.lidman@arm.com>
Change-Id: I9e8af071e8b820796577ee4792b4812a1212602b
|
|
This commit adds support for emulating the behavior
of the QuantizedMeanOrSum implementation of MEAN in
TensorFlow Lite.
Signed-off-by: Dwight Lidman <dwight.lidman@arm.com>
Change-Id: Ifd24e0e678e2f85cd66ab82deeaaf010d5351b1e
|
|
- Added full support for PAD operator
- Hardware padding is still used whenever possible
- Bug fix Pad followed by max pool if IFM contains negative values
Change-Id: Ifc64d1943737d94466f5e2821009dab12a49a965
Signed-off-by: Louis Verhaard <louis.verhaard@arm.com>
|
|
This commit adds support for the MEAN operator,
with some caveats.
Signed-off-by: Dwight Lidman <dwight.lidman@arm.com>
Change-Id: I165cb26cb5aefd68e70d2cfc68291ccf7b778921
|
|
Previously the keep_dims or keep_num_dims attribute was not supported for Sum and Fully Connected operators and would thus crash for certain tests. With this update, the attribute is extracted correctly and saved to the optimised tflite file.
Signed-off-by: erik.andersson@arm.com <erik.andersson@arm.com>
Change-Id: If33487f6d299bb99788bb3d13332b842ba961641
|
|
PAD followed by max/average pool is run on NPU if NPU
padding can be used. Average pool is converted to depthwise.
Change-Id: Icc3652e6d9ecff5ac3dc7d92080313d90c245404
Signed-off-by: Louis Verhaard <louis.verhaard@arm.com>
|
|
- Squeeze is no longer listed as supported operator
- Added missing doc-string for a Pad constraint
Change-Id: Ifd5e493acb0eb28bc4f104df74b3491589db8c29
Signed-off-by: Louis Verhaard <louis.verhaard@arm.com>
|
|
Added supported operator check that 32-bit fused activation functions
are not supported.
Change-Id: I01fdafeff8fdb13c71eae4f63be7e6f81b9223df
Signed-off-by: Louis Verhaard <louis.verhaard@arm.com>
|
|
- Added checks for unsupported pad sizes in PAD operator
- Bug fix right pad/bottom pad calculation when replacing PAD operator
by hardware padding
Change-Id: Ib84be711277d987052f14352ab386e0e0b774987
Signed-off-by: Louis Verhaard <louis.verhaard@arm.com>
|
|
Change-Id: If49abc31f093f1bd3393bee86f821fd35972086f
Signed-off-by: Diqing Zhong <diqing.zhong@arm.com>
|
|
Fixed a bug where PAD having no consumers would result in a crash.
Now the constraint doesn't crash and thus the intended error message is shown, resulting in easier debugging.
Change-Id: I1e4403d47a6152e7adbf7bc065db86d4217d39cc
Signed-off-by: erik.andersson@arm.com <erik.andersson@arm.com>
|
|
- Added operator check that OFM scale > smallest float32 number
- Generalized the restriction that IFM/OFM scale must not be infinite
Change-Id: I918f5ea3d8fdec6e8f6bd6780ed13a19d1234ed6
Signed-off-by: Louis Verhaard <louis.verhaard@arm.com>
|
|
Constraints and unit tests were added to check the new pad operator.
Change-Id: Id6d4cf2c4da486928c8f46ba1fa124eec66895a6
Signed-off-by: erik.andersson@arm.com <erik.andersson@arm.com>
|
|
Use an Enum instead of a bytestring to specify VALID or SAME padding
Signed-off-by: Michael McGeagh <michael.mcgeagh@arm.com>
Change-Id: I4e87f8c32b3bfac176d822a68de061e85a558fce
|
|
Minor fix in SPLITV tensor indexing for supported operators check.
Signed-off-by: Jacob Bohlin <jacob.bohlin@arm.com>
Change-Id: If8fa702bfbb25a4a7e5bdb136a19ef72eec7e1c2
|
|
This commit adds a constraint to FullyConnected
ops in supported_operators.py that puts any
such op on the CPU if tensor dimensions of the
output(s) are not 2D.
Signed-off-by: Dwight Lidman <dwight.lidman@arm.com>
Change-Id: I8c898a780b40fc4a1383c09213f0696ea6699b7d
|
|
When printing a set in the docstrings for the SUPPORTED_OPS.md file, the
order is random.
Reuse existing sorted string repr for the operator list and apply to
other printed sets (data types)
Signed-off-by: Michael McGeagh <michael.mcgeagh@arm.com>
Change-Id: I2ac12ea91c2637219e5c24f9a863aa0fc2086e77
|
|
Change-Id: Ifbd6c053ac618bedce0f56fe5c4c647a71d9cc46
Signed-off-by: Andreas Nevalainen <andreas.nevalainen@arm.com>
|
|
Vela only supports per-channel scaling for
convolution ops. This commit adds a check that
puts ops with per-channel scaling on the CPU.
A caveat worth mentioning is that neither
TensorFlow Lite or TensorFlow Lite Micro support
per-channel scaling for the CPU placed op,
however the problem is moved away from Vela.
This commit also changes a small utility function
in supported_operators.py used for docstring
formatting.
Signed-off-by: Dwight Lidman <dwight.lidman@arm.com>
Change-Id: I9ed090592f1d05dd4566d3e54dba1ef405299383
|
|
Fixed DepthwiseConv2D fails when bias tensor quant_values are None.
Also fixed DepthwiseConv2D fails with implicit depth multiplier.
Signed-off-by: Fredrik Svedberg <fredrik.svedberg@arm.com>
Change-Id: I799a565eefa498ccf7ac626fcd472b8cbd908931
|
|
EXPAND_DIMS is not yet supported by vela, and so should not be in the
list of supported ops.
Signed-off-by: Michael McGeagh <michael.mcgeagh@arm.com>
Change-Id: I5eca13eb52eb9b40ecc6592cda978614c71db99d
|
|
A new CLI has been added that allows the generation of a report
containing a summary table of all TFLite ops that can be placed on the
NPU, and what the constraints are for that operator to be successfully
scheduled on the NPU.
This option will generate a new file, SUPPORTED_OPS.md containing this
information, in the current working directory.
Signed-off-by: Michael McGeagh <michael.mcgeagh@arm.com>
Change-Id: I6a7e2a49f251b76b2ea1168fff78e00da1910b25
|
|
Put softmax on CPU if beta < 0
Signed-off-by: Patrik Gustavsson <patrik.gustavsson@arm.com>
Change-Id: I4ec866dd44d14e2737c4cd96474e54bb770bfb3e
|
|
Added external API to generate register command streams.
Existing code generation has been refactored to make
use of this API.
Change-Id: Ibb4c2b167809869f16470b14da24f08a65c82b7b
Signed-off-by: Louis Verhaard <louis.verhaard@arm.com>
|
|
This commit removes the constraint on all tensor
shapes matching the OFM shape.
The motivation is that this constraint essentially
only checks that the fixup function has run.
This means that it removes the possibility for the
fixup function to run after the supported operator
check and this effectively means that any
StridedSlice operator that would be placed on the
CPU is still modified by the fixup function.
Because the fixup function is moved to after the
supported operators check, some unreachable cases
are removed from the fixup function.
Signed-off-by: Dwight Lidman <dwight.lidman@arm.com>
Change-Id: I7a82126b7de73bd67873b4e6daf53a6767e33d16
|
|
Previously the internal operator type was printed when checking the
supported operator checks. This now converts that back to the external
type name.
Additionally removed dead code and changed the message for cpu-only ops
Signed-off-by: Michael McGeagh <michael.mcgeagh@arm.com>
Change-Id: Ib2b0cbcb49fdf63edb835828e266b079e63bae37
|
|
All existing constraints have now been refactored using the new
framework.
Signed-off-by: Michael McGeagh <michael.mcgeagh@arm.com>
Change-Id: Ic9ba0d7040cb9f114b959a949bfdf777f86752c7
|
|
Added a supported_operators check for Relu activation functions. If the
scaling value overflows to infinity, it will be placed on the CPU.
Signed-off-by: Jacob Bohlin <jacob.bohlin@arm.com>
Change-Id: I66b7bec062599609aadcbb7531caebbc45a7451f
Signed-off-by: Jacob Bohlin <jacob.bohlin@arm.com>
|
|
- Fixed and documented both tensor and quant params scaling checks
- Added quant params validity check and tensor quantisation check
- Added valid tensor checks to some graph optimisation functions
Signed-off-by: Tim Hall <tim.hall@arm.com>
Change-Id: I8d6e8f03a603d28886dde511672c8399c85b794c
|
|
Using a new system to report constraints, replaced existing
functionality for checking conv-like ops.
This new system will allow reporting of all constraints regardless of
any input network.
Signed-off-by: Michael McGeagh <michael.mcgeagh@arm.com>
Change-Id: If81177deca2a3b57c9dd9a3a08868cbc9cef0c23
|
|
Suppress info print that Const/Placeholder/SubgraphInput are not supported
on the NPU.
Change-Id: I6f323b64185b01b619b584c1473ae61d010ab3a4
Signed-off-by: Louis Verhaard <louis.verhaard@arm.com>
|
|
This reverts commit 04986c0016e59993563490fe67052371fc0e1ad2.
Reason for revert: Merged by mistake
Change-Id: I150ad9ba7074ad1e80f21180aeba56a454d9f748
|
|
Suppress info print that Const/Placeholder/SubgraphInput are not supported
on the NPU.
Change-Id: I689d25481df0cd10487484c9f639e4253df081ee
Signed-off-by: Louis Verhaard <louis.verhaard@arm.com>
|
|
Keeping the constraint functions consistent with each other
Added specific tensor names in the extra info
Added operator name to the warning generated
This should help easily identify specific problematic nodes in a graph
and give a good enough explanation as to why they are placed on the CPU
Signed-off-by: Michael McGeagh <michael.mcgeagh@arm.com>
Change-Id: Ie5bbdd31e5e75fe37e3d8bb8fee1d260080bce83
|
|
Added info print for unsupported operator
Signed-off-by: Patrik Gustavsson <patrik.gustavsson@arm.com>
Change-Id: I1002d1c2249661bff17ef86d9500d1aeb2a1e38e
|
|
Vela supports batching of FC, restriction removed.
Signed-off-by: Patrik Gustavsson <patrik.gustavsson@arm.com>
Change-Id: Ica56738f1b2676628644fc44f2039a24807f5ccb
|
|
This commit changes and amends some parts of the
restriction functions in order to make sure
operators are correctly placed.
Signed-off-by: Dwight Lidman <dwight.lidman@arm.com>
Change-Id: I336cf33a874c9078a5bbf81ce129ff917dbc5e9a
|
|
- op.type is now an enum instead of a string
- Removed unused operator codes
- Refactored some attributes like npu_block_type, fused_activation_function
- Refactored operator index calculation
- Refactored a number of operator sets
Change-Id: I641f65ee375794b7aec42abc0664251ae37d78e8
Signed-off-by: Louis Verhaard <louis.verhaard@arm.com>
|
|
Fix issue for checking axis in concat, now allowing 0.
Signed-off-by: Patrik Gustavsson <patrik.gustavsson@arm.com>
Change-Id: I85a5fc3dacdfc66dc01b0e05048dd100254fddff
|
|
A new mechanism to report generic restrictions/constraints for
operators has been implemented.
Each check is its own defined function, and has a general reason for
the constraint defined as its docstring.
This allows us to query all reasons up front and report this without
having to run through real data to trigger the checks.
This is part of a larger refactoring and the specific restrictions will
be replaced by a similar mechanism.
Signed-off-by: Michael McGeagh <michael.mcgeagh@arm.com>
Change-Id: Id3fb2639f91cfac5fc5b8c14f7620de1a85972b2
|
|
Part of larger refactoring. The sets of operators do not need to be
instance attributes and are not expected to be modified at runtime.
This in turn allows almost all functions to become class methods.
Signed-off-by: Michael McGeagh <michael.mcgeagh@arm.com>
Change-Id: I7dc24d65cdd6c4bda641b3d6133b3134302a552f
|
|
Min and max operations was not passed through
the checking of elementwize OPs in the supported
operator checking.
Changed so they are passed through this check as well.
Signed-off-by: Patrik Gustavsson <patrik.gustavsson@arm.com>
Change-Id: I358a121de33882802415d97d9ed5dbee53233f77
|
|
Fixed crash in networks with 5D tensors.
Fixed crash for (int32) tensors without quantization.
Added validity checks for concatenation.
Moved unfusing of activation function from tflite_reader to graph_optimiser.
Signed-off-by: Fredrik Svedberg <fredrik.svedberg@arm.com>
Change-Id: Ib9ba8891dc95ef5491e15d0feedef44331a26393
|
|
Added check for inf numbers for all scales.
Signed-off-by: Andreas Nevalainen <andreas.nevalainen@arm.com>
Change-Id: I84fcae429be4869d8489f66bef26863c254104cd
|
|
Updated supported operator checks for StridedSlice:
- allow negative indices in begin/end values
- added more checks on shapes
Change-Id: I3ac76bfa6b313f0e2250f0749f152fb0e3aa033c
Signed-off-by: Louis Verhaard <louis.verhaard@arm.com>
|
|
Change-Id: I750ec63a0e37b38feaf4cbdcc883fdbef92bccdf
Signed-off-by: Andreas Nevalainen <andreas.nevalainen@arm.com>
|