Matthew Brookhart [Wed, 6 May 2020 03:53:07 +0000 (20:53 -0700)]
fix prelu importer and add tests: (#5521)
Samuel [Tue, 5 May 2020 23:47:44 +0000 (05:17 +0530)]
[TFLITE]Nit: Function names made consitent (#5515)
Ramana Radhakrishnan [Tue, 5 May 2020 23:47:10 +0000 (00:47 +0100)]
[PATCH] [ring_buffer.h] Improve commentary for RingBuffer (#5518)
bytes_available refers to the number of bytes used in the ring
buffer. At the same time, fix a typo.
Tianqi Chen [Tue, 5 May 2020 23:47:00 +0000 (16:47 -0700)]
[RPC][BUGFIX][BACKPORT-0.6] Fix bug in rpc ring buffer shrink (#5516)
mbaret [Tue, 5 May 2020 21:38:40 +0000 (22:38 +0100)]
[QNN] Support CallNode inputs in qnn.concatenate (#5360)
* [QNN] Support CallNode inputs in qnn.concatenate
Currently, qnn.concatenate assumes that its 1st arg
(data) is a TupleNode. This may not necessarily be true
if the input is a CallNode which returns a value of
tuple-type. This patch handles the CallNode case by
inserting TupleGetItemNodes.
* Fix lint
* Add test
Change-Id: I40b55517b8b1dabbeca89337f80c0c8e62e34981
* Use isinstance
Change-Id: I731a231113c5214528373ef52b603a9f05ec502a
* isinstance fix
Change-Id: Ib3495532f6e4feb5aae3d3096cedd4dc4676cdb4
* Use elif/else if
Change-Id: Id8123ea2dd9ce3d8267609de7b5602bb84b084fb
* Fix lint
Change-Id: Ib6899bb22260575aa3f5d8b51b5d2a0277ee2b10
* Lint fix
Change-Id: I56cf1930315344e42d956818a6c68e80836ae786
* Spaces
Change-Id: I3edab192e32bafa9ffdc915315791c63279d85dc
Tianqi Chen [Tue, 5 May 2020 19:07:59 +0000 (12:07 -0700)]
[RPC] Call sync in remote cpu to gpu copies (#5512)
Tianqi Chen [Tue, 5 May 2020 00:05:33 +0000 (17:05 -0700)]
[REFACTOR][RPC][PROCOTOL-CHANGE] Modularize the RPC infra (#5484)
* Update dmlc-core which was mistakenly overriden
* [REFACTOR][RPC][PROCOTOL-CHANGE] Modularize the RPC infra.
This PR refactors the RPC protocol to make it more modularized.
- RPCSession: represent a set of features that need to be implemented
- RPCEndPont: End point that forwards the RPCSession requests over a communication channel.
- RPCModule: Exposes an RPCSession as an rpc device in the TVM Runtime API.
In the new design, the local machine is presented as a special case of RPCSession.
The remote is just another client session that calls into RPCEndPoint.
The RPC communication path is as follows.
```
client -> ClientSession -> EndPoint[client@n0]
-> networking[between n0 <=> n1]
-> EndPoint[server@n1] -> LocalSession[@n1]
```
Because of the new modular design, we can now chain more sessions together.
For example, we can now run the following proxy setup (testcase in test_runtime_rpc.test_session_constructor).
```
client -> ClientSession -> Endpoint[client@n0]
-> networking[between n0 <=> n1]
-> Endpoint[server@n1] -> ClientSession -> Endpoint[client@n1]
-> networking[between n1 <=> n2]
-> Endpoint[server@n2] -> LocalSession[@n2]
```
We can also implement other types of Sessions.
As an example, We introduced a PopenSession that communicates with
the another process via a pipe.
We also add more comments about the internal of the RPC.
The communication protocol is simplfied using a similar convention as PackedFunc.
This allows us to further reduce the amount of special remote syscalls.
Due to the major improvement and simplification, we are making a non-compatible update to the RPC protocol.
It means that the client and server needs to be upgraded to together in order for it to function correctly.
This PR also introduces a versioning mechanism to the current RPC procotol,
so that future upgrade will be produce more user friendly with error messages.
* Address review comments
* Remove ld library path
MORITA Kazutaka [Mon, 4 May 2020 20:51:36 +0000 (05:51 +0900)]
[RUST][RUNTIME] Fix workspace (#5503)
* [RUST][RUNTIME] Fix workspace
* use ok_or_else instead of ok_or
Yao Wang [Mon, 4 May 2020 15:26:17 +0000 (08:26 -0700)]
Fix Canonical Simplifier (#5505)
Thomas Viehmann [Sun, 3 May 2020 15:34:14 +0000 (17:34 +0200)]
bump tophub rocm version (#5504)
Andrew Reusch [Sat, 2 May 2020 17:52:40 +0000 (10:52 -0700)]
[uTVM] Reset target and wait for runtime initialization on connect. (#5499)
* This ensures a clean runtime environment for tuning and evaluating
tasks.
* This patch assumes that the code flashed to target executes the ARM
BKPT or RISC-V EBREAK instructions after startup.
MORITA Kazutaka [Sat, 2 May 2020 00:06:43 +0000 (09:06 +0900)]
[Rust] Fixes for wasm32 target (#5489)
* [Rust] Fixes for wasm32 target
* [Rust] Add test for wasm32
* allow cargo config to be into repo
* Disable wasm tests in CI
Thierry Moreau [Sat, 2 May 2020 00:02:24 +0000 (17:02 -0700)]
[TFLite] Model importer to be compatible with tflite 2.1.0 (#5497)
Krzysztof Parzyszek [Fri, 1 May 2020 20:55:34 +0000 (15:55 -0500)]
Link necessary libraries when building runtime for Android (#5496)
- Link libgcc to enable use of thread-local storage (ThreadLocalStore
requires emutls).
- Link liblog when building with Hexagon support.
Zhi [Fri, 1 May 2020 20:27:44 +0000 (13:27 -0700)]
[REFACTOR][BOYC] Non recursive partitioning (#5493)
* non recursive partitioning
* refactor maps
* rebase upstream
* refactor shared output
* address comments
Co-authored-by: Cody Yu <comaniac0422@gmail.com>
Samuel [Fri, 1 May 2020 13:03:45 +0000 (18:33 +0530)]
[MXNET]broadcast and logical op support (#5461)
* [MXNET]broadcast and logical op support
* Review comment fixed
Krzysztof Parzyszek [Fri, 1 May 2020 08:43:22 +0000 (03:43 -0500)]
[Hexagon] Change "scalar" and "stack" in IDL from "inrout" to "in" (#5487)
Co-authored-by: Abhikrant Sharma <quic_abhikran@quicinc.com>
Krzysztof Parzyszek [Fri, 1 May 2020 05:06:06 +0000 (00:06 -0500)]
Make "none" DataType explicit (#5491)
* Make "none" DataType explicit
The None data type is created when converting an empty string to DataType.
Add functions to create it and recognize it. Convert it to the "void" LLVM
type in LLVM codegen.
* Rename "none" to "void"
* Map VoidType:Type -> Void:DataType in GetRuntimeDataType
* Map Void:DataType -> VoidType:Type in GetType
Tianqi Chen [Thu, 30 Apr 2020 23:11:38 +0000 (16:11 -0700)]
[IR] Initial stab at std::string->String upgrade (#5438)
Animesh Jain [Thu, 30 Apr 2020 19:21:27 +0000 (12:21 -0700)]
Removing older Object detection TFlite test (#5477)
Andrew Reusch [Thu, 30 Apr 2020 17:59:33 +0000 (10:59 -0700)]
[RUNTIME][uTVM] AutoTVM + uTVM for Cortex-M7 (#5417)
* Prototype for micro TVM.
* Cleanup and sync micro tvm prototype.
* Use /std:c++14 with MSVC.
* Per tqchen: project has already moved to C++14
* Presubmit failed for code that built locally on gcc.
* fix ASF lint, and fix add_asf_header too
* Compiles with USE_MICRO=OFF.
* Cleanup TargetPtr and word size representations.
* fix compile warning
* address logan's comments
* address logan and liangfu comments
* address thierry's comments
* address u99127, liangfu, tmoreau89 comments
Co-authored-by: Logan Weber <weberlo@cs.washington.edu>
Zhi [Thu, 30 Apr 2020 17:00:27 +0000 (10:00 -0700)]
[Fix] Add ConstantNode to IsAtomic (#5457)
* add constantnode to atomic
* Add ToANormalForm to FoldConstant
Hua Jiang [Thu, 30 Apr 2020 15:07:23 +0000 (08:07 -0700)]
[VTA] Fix VTA compile issue (#5481)
* [VTA] Fix Pynq driver build issue.
Issue:
When doing vta compile in xilinx FPGA, the pynqdriver.cc report can
not find <vta/driver.h>
Solution:
add related path.
* Fix libvta load fail issue.
issue:
run vta on pynq board libvta.so load failed.
solution:
fixed VTA.make logic issue
Tianqi Chen [Thu, 30 Apr 2020 15:07:00 +0000 (08:07 -0700)]
[RUNTIME] Improved Packed FFI for optional. (#5478)
Allows Optional<NDArray> and module to be passed with the right type code.
Yizhi Liu [Thu, 30 Apr 2020 15:06:41 +0000 (08:06 -0700)]
[team] add reviewer kparzysz-quic (#5482)
Samuel [Thu, 30 Apr 2020 08:07:49 +0000 (13:37 +0530)]
[FRONTEND][TFLITE]Logical not op support (#5475)
Bing Xu [Thu, 30 Apr 2020 04:07:18 +0000 (21:07 -0700)]
[intrin] a few more math functions (#5468)
ZHANG Hao [Thu, 30 Apr 2020 00:54:57 +0000 (08:54 +0800)]
[VTA][Runtime] clear sram index in reset (#5470)
Co-authored-by: Zhang Hao <zhanghao@4paradigm.com>
mbaret [Wed, 29 Apr 2020 18:23:15 +0000 (19:23 +0100)]
[BYOC] Bind constant tuples in graph partitioner (#5476)
* Bind constant tuples in the graph partitioner
Change-Id: I815b32b5445a536c1837369b04f67dbbb0aed900
* Add partitioning test
Change-Id: I3a492ec8d1beab4830214e3bc8da2a7c80771ca4
* Rename test target
Change-Id: Ie32f37c1395ff597c0047ad3a93ed04ce3f3125d
Tianqi Chen [Wed, 29 Apr 2020 17:45:26 +0000 (10:45 -0700)]
[CI] update ci-gpu to the latest (#5469)
boh_inspur [Wed, 29 Apr 2020 17:22:25 +0000 (12:22 -0500)]
[CODEGEN][CUDA] Fix a bug when vectorized load&store was involved for… (#5428)
* [CODEGEN][CUDA] Fix a bug when vectorized load&store was involved for "char2"
* Add unittest for char2
* vector element load support char2&add some unittest for vector element load
* Merge common up logic&Support char3&Add unittest for char3
mbaret [Wed, 29 Apr 2020 16:13:16 +0000 (17:13 +0100)]
[TFLITE] Match TFLite shape for SSD custom op (#5473)
This patch ensures that the output shape from TVM's
Detection_PostProcess is the same as TFLite's and
expands the unit test to confirm this.
Change-Id: If5db95741533f131241dfebbaa7708dbd528fe70
Mahesh Ambule [Wed, 29 Apr 2020 10:51:57 +0000 (16:21 +0530)]
[Frontend][TFLite] support for FILL and SPLIT_V operators (#5330)
* tflite spliv ops
* TFLITE fill and splitv ops
* TFLITE fill and splitv ops
* TFLITE fill and splitv ops
* remove unnecessary operator check
Mahesh Ambule [Wed, 29 Apr 2020 05:35:54 +0000 (11:05 +0530)]
[Frontend][TFLite] L2_POOL_2D operator (#5452)
* TFLITE fill and splitv ops
* l2_pool_2d op changes in comment
* TFLite l2_pool_2d op added test case in main
* TFLite L2_POOL_2D added check for quantized input
Thierry Moreau [Wed, 29 Apr 2020 05:06:30 +0000 (22:06 -0700)]
[COMMUNITY] @liangfu -> committer (#5460)
Leandro Nunes [Tue, 28 Apr 2020 23:32:38 +0000 (00:32 +0100)]
Improve Docker cache reuse by pointing to the current version of the image, (#5466)
on top of another image to be used as reference.
SXM-inspur [Tue, 28 Apr 2020 20:44:03 +0000 (04:44 +0800)]
[Topi][Cuda]Optimizations of global_ave_pool for NHWC layout (#5450)
* Optimizations of global_ave_pool for NHWC layout
* Optimize the code format to pass inspection of pylint
Co-authored-by: Shawn-Inspur <wushaohua@inspur.com>
Matthew Brookhart [Tue, 28 Apr 2020 01:20:23 +0000 (18:20 -0700)]
Add RoiAlign to Onnx frontend (#5454)
Michal Piszczek [Mon, 27 Apr 2020 15:14:33 +0000 (08:14 -0700)]
[TFLite Runtime] Add TFLite Runtime dependencies to CI CPU docker build (#5437)
Nikolay Nez [Mon, 27 Apr 2020 11:34:58 +0000 (20:34 +0900)]
[Pytorch] fix translation of transpose when axis argument is as a list (#5451)
notoraptor [Mon, 27 Apr 2020 07:36:35 +0000 (03:36 -0400)]
[relay][topi] Add operation relay.nn.dilate() which calls topi.nn.dilate() (#5331)
* Add operation relay.nn.dilate() which calls topi.nn.dilate().
* Fix typo
* Set op pattern to injective
Samuel [Mon, 27 Apr 2020 06:53:10 +0000 (12:23 +0530)]
[ONNX]GatherNd, Round, IsNaN, IsInf (#5445)
yongfeng-nv [Mon, 27 Apr 2020 00:00:52 +0000 (20:00 -0400)]
Improve IntervalSet's floormod (#5367)
Thierry Moreau [Sun, 26 Apr 2020 23:49:05 +0000 (16:49 -0700)]
[Docs] VTA install doc migration from md to rst (#5442)
Samuel [Sun, 26 Apr 2020 02:58:02 +0000 (08:28 +0530)]
[KERAS]Embedding layer (#5444)
mbaret [Sat, 25 Apr 2020 16:18:30 +0000 (17:18 +0100)]
[RELAY] Move frontend utils (#5345)
* [RELAY] Move frontend utils
The util file currently under frontend is used from
outside of frontend (in qnn/op/legalizations). This suggests
that the file should be pushed up to a higher level.
The benefit from this change is that importing qnn no longer
also imports all the frontends.
* Inline get_scalar_from_constant
Change-Id: I1cc64e9ecb0eadb6ac0f7b62e6ea174644af4ad4
* Remove util.py from Relay
Change-Id: If9cd7cf3fc0bd1861a3a9b5604f338e084d8db96
* Shorten functions
Change-Id: Ieb537d82e6ee52421ff05a90cd00a03679ffebf2
* Line length
Change-Id: I1d216b7e73a060c4f118f5da50ce58b18eba907f
Wei Pan [Sat, 25 Apr 2020 08:20:27 +0000 (01:20 -0700)]
[CodeGen] Cleanup generated code (#5424)
- remove unnecessary white spaces from storage kind
- do not start a new scope for vectorization as temporary
variables are alll uniquely generated.
The above two changes make vectorized code much cleaner.
Signed-off-by: Wei Pan <weip@nvidia.com>
Matthew Brookhart [Sat, 25 Apr 2020 06:37:09 +0000 (23:37 -0700)]
Add TopK to ONNX Frontend (#5441)
* Add TopK to ONNX Frontend
* respond to review comments
Thomas Viehmann [Sat, 25 Apr 2020 01:32:18 +0000 (03:32 +0200)]
fix miopen pad (#5433)
Samuel [Sat, 25 Apr 2020 01:23:35 +0000 (06:53 +0530)]
[PYTORCH]Rsub, Embedded, OneHot ops support (#5434)
Samuel [Sat, 25 Apr 2020 01:22:10 +0000 (06:52 +0530)]
[RELAY][PYTORCH]cosh,sinh,log2,log10,log1p op support (#5395)
* [RELAY][PYTORCH]cosh,sinh,log2,log10,log1p op support
* Review comment fixed
* Gradient testcase added
Tianqi Chen [Fri, 24 Apr 2020 19:49:05 +0000 (12:49 -0700)]
[RUNTIME][OBJECT] Introduce static slots for common objects. (#5423)
The _type_child_slots can be used to enable quick type checking optimization
by checking the whether the type index is within the bound.
This PR enables these static slots:
- Introduce a static assert to avoid the scenario when a developer forget to
_type_child_slots when the field is set for the type's parent.
- Revamp and assign static type index to common runtime objects
- Add a DumpTypeTable call to allow developer monitor the current situation
of type table and offers suggestions for the slots(ideally the slots equals
the number of children so there is no overflow.
JishinMaster [Fri, 24 Apr 2020 16:50:01 +0000 (18:50 +0200)]
Corrected TVM autotuning on GPU (#5432)
Added missing "tir" in tvm.tir.analysis.verify_gpu_code(f, kwargs)
Thomas Viehmann [Fri, 24 Apr 2020 15:56:52 +0000 (17:56 +0200)]
misc fixes for ROCm (pointer lifetime, runtime::String refactor) (#5431)
Samuel [Fri, 24 Apr 2020 10:53:38 +0000 (16:23 +0530)]
[FRONTEND][TFLITE]Gather, StridedSlice op support added (#4788)
* [FRONTEND][TFLITE]Gather, StridedSlice op added
* Review comments fixed
Samuel [Fri, 24 Apr 2020 10:49:26 +0000 (16:19 +0530)]
[PYTORCH]where, addcdiv, addcmul op support (#5383)
* [PYTORCH]Where, addcdiv, addcmul op support
* Review comments fixed
Tianqi Chen [Fri, 24 Apr 2020 05:06:05 +0000 (22:06 -0700)]
[PY][FFI] Introduce PyNativeObject, enable runtime.String to subclass str (#5426)
To make runtime.String to work as naturally as possible in the python side,
we make it sub-class the python's str object. Note that however, we cannot
sub-class Object at the same time due to python's type layout constraint.
We introduce a PyNativeObject class to handle this kind of object sub-classing
and updated the FFI to handle PyNativeObject classes.
MORITA Kazutaka [Fri, 24 Apr 2020 02:54:26 +0000 (11:54 +0900)]
[FRONTEND][MXNET] support elemwise logic ops (#5361)
Michal Piszczek [Thu, 23 Apr 2020 23:25:46 +0000 (16:25 -0700)]
Add option to specify flatbuffers location (#5425)
Samuel [Thu, 23 Apr 2020 22:05:25 +0000 (03:35 +0530)]
[MXNET]DepthToSpace & SpaceToDepth Operator (#5408)
Ramana Radhakrishnan [Thu, 23 Apr 2020 22:05:03 +0000 (23:05 +0100)]
[RFC] Pytest environment improvements (#5421)
* [RFC] Pass pytest options globally.
In many places having a global pytest flag is useful . For me with the
build and test of tvm , I would like to be able to globally pass in
pytest options as part of development flow or CI flows where one would
like to measure other things regularly that need measurements including
pytest coverage data that I would like to experiment with across the stack.
This has been achieved with an additional setup-pytest-env.sh file in
tests/scripts rather than putting in something in every single task test
script and something I would like to avoid.
This now means the -v option to pytest is superfluous. I did consider
having a pytest.ini file but that doesn't allow me to pass any old
environment variable in and this seems to be the compromise.
* Improve other use case documentation
* Rationalize pytest environment.
* Remove the setting from docker/with_same_user.
* Take the opportunity to migrate common PYTHONPATH and
TVM_PATH into the common environment setting.
* Fixup vta fsim
* Be more explicit with common PYTHONPATH
* Fix python path for task_python_vta_fsim.sh properly
* Fix nit in documentation.
Cody Yu [Thu, 23 Apr 2020 20:56:43 +0000 (13:56 -0700)]
[BYOC] Use Non-Recursive Visitor/Mutator (#5410)
* Non-Recursive AnnotatedTarget and MergeAnnotation
* Non-Recursive AnnotatedRegionSet and RegionMerger
Tianqi Chen [Thu, 23 Apr 2020 19:40:11 +0000 (12:40 -0700)]
[DOCS] Migrate some markdowns to rst, fix sphinx3 warnings (#5416)
* [DOCS] Migrate some markdowns to rst, fix sphinx3 warnings
* Add note block
Ramana Radhakrishnan [Thu, 23 Apr 2020 19:13:42 +0000 (20:13 +0100)]
[CI] Migrate Tensorflow and Tensorflow lite in CI to 2.1.0 (#5392)
* Migrate Tensorflow and TFLite in the CI up to 1.15.2
The latest stable version of Tensorflow and Tensorflow lite
in the 1.x series is 1.15.2. The tflite frontend is receiving
support for versions of tflite > 1.14 but there is no consistent
testing.
There are 2 failures already in the source base with tf 1.15
and I'm concerned this will just get exacerbated over time
if we don't have CI picking this up and I view this as a stepping
stone towards stepping CI to TF2.x.
The test failures that I have commented will get issues raised
for them as issues to be fixed.
* Comment out run of qnn_mobilenet_v3_net
This is another test that fails with TFlite 1.15.2
* Skip the qnn_mobilenet_v3 test in the pytest fashion.
* Switch docker versions to support Tensorflow 2.1.0
* Fix up pytest imports and usage.
* Skip these tests currently for Tensorflow 2.1.0
Wei Pan [Thu, 23 Apr 2020 18:59:03 +0000 (11:59 -0700)]
[cuDNN] Add cuDNN grouped convolutions support (#5319)
Signed-off-by: Wei Pan <weip@nvidia.com>
Zhao Wu [Thu, 23 Apr 2020 17:10:02 +0000 (01:10 +0800)]
[Frontend] Asymmetric padding of convolution support (#4803)
samwyi [Thu, 23 Apr 2020 15:05:42 +0000 (08:05 -0700)]
fix [RUNTIME][VULKAN] vkBuffer released before memory copy command send to GPU (#5388) (#5418)
MORITA Kazutaka [Thu, 23 Apr 2020 15:05:08 +0000 (00:05 +0900)]
[DOCS] Migrate HLS documents from md to rst (#5419)
MORITA Kazutaka [Thu, 23 Apr 2020 07:35:43 +0000 (16:35 +0900)]
[RUNTIME][CONTRIB] CoreML Runtime (#5283)
* [RUNTIME][CONTRIB] CoreML Runtime
* fix lint
* fix CI
* use xcrun to compile coreml model
Tianqi Chen [Thu, 23 Apr 2020 03:44:25 +0000 (20:44 -0700)]
[TIR][REFACTOR] Remove ir_pass in favor of analysis/transform. (#5415)
This PR removes ir_pass(old style pass functions) in favor
of analysis/transform(new style pass manager).
Andrew Reusch [Thu, 23 Apr 2020 01:30:11 +0000 (18:30 -0700)]
Don't remove() TempDirectory in __del__ after atexit hook runs. (#5414)
* Use atexit to remove TempDirectory before interpreter shutdown.
* Can't rely on complex functions from __del__ anyway.
* Fixes warning message on my box:
Exception ignored in: <function TempDirectory.__del__ at 0x12be10680>
Traceback (most recent call last):
File ".../tvm/python/tvm/contrib/util.py", line 55, in __del__
File ".../tvm/python/tvm/contrib/util.py", line 51, in remove
File "/usr/local/opt/python/Frameworks/Python.framework/Versions/3.7/lib/python3.7/shutil.py", line 509, in rmtree
AttributeError: 'NoneType' object has no attribute 'path'
Krzysztof Parzyszek [Wed, 22 Apr 2020 22:58:42 +0000 (17:58 -0500)]
[LLVM] Replace calls to Type::getVectorNumElements (#5398)
This function has recently been removed from LLVM 11. Use alternative
way to obtain vector element count (VectorType::getNumElements) which
works for all LLVM versions.
Andrew Reusch [Wed, 22 Apr 2020 22:35:21 +0000 (15:35 -0700)]
Customize SI prefix in logging (#5411)
* Customize SI prefix in logging
* Include unit test
Haichen Shen [Wed, 22 Apr 2020 21:59:54 +0000 (14:59 -0700)]
[Relay] Fix memory leak when accessing NDArray (#5413)
Tianqi Chen [Wed, 22 Apr 2020 20:25:36 +0000 (13:25 -0700)]
[TIR] Enhance Substitute, python bindings for Substitute/PostOrderVisit/IRTransform. (#5400)
Substitute now takes a std::function to customize more replacing behaviors.
Co-authored-by: Siyuan Feng <hzfengsy@sjtu.edu.cn>
Co-authored-by: Siyuan Feng <hzfengsy@sjtu.edu.cn>
Tianqi Chen [Wed, 22 Apr 2020 20:07:35 +0000 (13:07 -0700)]
Update dmlc-core to latest (#5401)
Haichen Shen [Wed, 22 Apr 2020 14:55:15 +0000 (07:55 -0700)]
[Fix] Remove the duplicate PrintIR pass in Relay (#5403)
Ramana Radhakrishnan [Wed, 22 Apr 2020 06:09:11 +0000 (07:09 +0100)]
Factor out import of common tflite.Operator in tflite frontend. (#5355)
* Restructure imports in tflite frontend.
These python modules are needed for every tflite file parsed.
Factorize out imports of the common most ones.
Now that the import of operator is common, asserts can be commonized.
Loses 473 lines of duplication.
* Only restrict to tflite.Operator
Samuel [Wed, 22 Apr 2020 03:57:06 +0000 (09:27 +0530)]
[KERAS]Minimum & AlphaDropout op support (#5380)
Krzysztof Parzyszek [Wed, 22 Apr 2020 02:11:28 +0000 (21:11 -0500)]
[LLVM] Use ArrayRef<int> in calls to CreateShuffleVector (#5399)
This switch was made in LLVM 11. Previously this function was expecting
mask indices of type uint32_t. This variant is now deprecated.
Tianqi Chen [Tue, 21 Apr 2020 21:23:18 +0000 (14:23 -0700)]
[PTYTHON] Migrate VTA TIR passes to the new pass manager. (#5397)
Ramana Radhakrishnan [Tue, 21 Apr 2020 14:49:41 +0000 (15:49 +0100)]
Tf2 test fixups (#5391)
* Fix oversight in importing tf.compat.v1 as tf.
* Actually disable test for lstm in TF2.1
Since the testing framework actually uses pytest, the version
check needs to be moved.
Andrew Reusch [Tue, 21 Apr 2020 14:41:52 +0000 (07:41 -0700)]
Fix test_ir_type. (#5390)
* The void return type is not None/nullptr, it's VoidType or
TupleType([]).
Animesh Jain [Tue, 21 Apr 2020 11:06:13 +0000 (04:06 -0700)]
[Topi, ARM] Disbale Winograd for quantized tensors. (#5363)
* [Topi, ARM] Disbale Winograd for quantized tensors.
* Relaxing float
Josh Fromm [Tue, 21 Apr 2020 10:57:13 +0000 (03:57 -0700)]
Add ability to have multiple copies of same input to onnx_inputs. (#5389)
Tianqi Chen [Tue, 21 Apr 2020 03:53:32 +0000 (20:53 -0700)]
[ARITH] Remove legacy const pattern functions (#5387)
Tianqi Chen [Tue, 21 Apr 2020 02:28:08 +0000 (19:28 -0700)]
[ARITH] Remove the legacy Simplify, migrate to Analyzer. (#5385)
The legacy Simplify/CanonicalSimplify are now a thin wrapper around the Analyzer.
This PR removes these functions and migrated every place that requires
simplification to enforce Analyzer creation.
The new API would encourage more Analyzer sharing and potentially enable
context-aware analyzer-based simplification.
Tianqi Chen [Tue, 21 Apr 2020 02:28:00 +0000 (19:28 -0700)]
[REFACTOR][TE] Inline -> te/schedule/operation_inline.h (#5386)
Rationale: inline is a transformation used in te to
rewrite its internal expressions. It is not a formal IRModule->IRModule transform pass.
Also removed the python test as the test is covered by stage.compute_inline.
Bing Xu [Mon, 20 Apr 2020 05:15:37 +0000 (22:15 -0700)]
[Blocksparse] Pipeline for lowering dense model to sparse-dense (#5377)
Tianqi Chen [Mon, 20 Apr 2020 02:57:25 +0000 (19:57 -0700)]
[TIR][REFACTOR] RewriteForTensorCore -> te/schedule (#5379)
* [TIR][REFACTIR] RewriteForTensorCore -> te/schedule
RewriteForTensor depends on the schedule information, which makes it differ
from a typical pass(which should get all the information from the input TIR).
As a result, we refactor it as a SchedulePostProc step for now.
We should revisit it later as we introduce more support for tensor core patterns in the TIR.
* Fix VTA to fit the new IR Pattern
Samuel [Mon, 20 Apr 2020 01:48:51 +0000 (07:18 +0530)]
[PYTORCH]Unary Ops (#5378)
Tianqi Chen [Sun, 19 Apr 2020 22:26:51 +0000 (15:26 -0700)]
[TIR][REFACTOR] Remove te::Tensor dependencies from TIR passes. (#5372)
* [TIR][REFACTOR] Remove te::Tensor dependencies from TIR passes.
te::Tensor is an useful object for tensor expression, but brings
un-necessary reverse dependency in TIR nodes such as Provide and Realize.
This PR is a first step to remove this dependency. We will use Buffer in all the places
where the te::Tensor was used. The rough correspondence are:
- Provide -> BufferStore
- Realize -> BufferRealize
- HalideCall -> BufferLoad.
After this change, we can not use IRModule of PrimFuncs cleanly to represent TIR
at any point of the optimizations. Buffer will serve as the abstraction for the TIR data
models to represent the intermediate storages and their constraints.
We still keep Realize/HalideCall and Provide as TIR nodes for now to make the change minimum.
Right after ScheduleOps, we call SchedulePostProcToPrimFunc to canonicalize the temporary IR
generated by TE(which contains these nodes) to the TIR.
The TIR optimizations are now mostly migrated to to the pass manager.
Followup PRs are needed to migrate the remaining few passes.
* Fix dev tutorial
shoubhik [Sun, 19 Apr 2020 04:09:32 +0000 (21:09 -0700)]
Remove developer facing api from frontend exports. (#5375)
Josh Fromm [Sun, 19 Apr 2020 04:09:14 +0000 (21:09 -0700)]
Add cuda target check to dense tensorcore schedule. (#5376)
Tang, Shizhi [Sun, 19 Apr 2020 03:11:51 +0000 (11:11 +0800)]
[TIR] Fix lower_warp_memory when there are >1 warp buffers (#5368)
* fix recursion in lower_warp_memory
* post-order mutation
Tianqi Chen [Sat, 18 Apr 2020 19:33:58 +0000 (12:33 -0700)]
[TIR][REFACTOR] Migrate low-level passes in tvm.lower to the Unified IR pass manager. (#5364)
- Migrate BoundCheckers and Simplify
- Migrate RewriteUnsafeSelect and RemoveNoOp
- Migrate UnrollLoop and StorageRewrite
- Migrate InjectDoubleBuffer and InjectVirtualThread
- Migrate LoopPartition and Vectorize
- Migrate CoProcSync, LiftAttrScope, InjectCopyIntrin
We still keep ir_pass registerations for now.
Need a separate PR to refactor the parts before the StorageFlatten.
Krzysztof Parzyszek [Sat, 18 Apr 2020 11:56:17 +0000 (06:56 -0500)]
[RUNTIME] FastRPC interface for Hexagon runtime (#5353)
* [RUNTIME] FastRPC interface for Hexagon runtime
Co-authored-by: Ravishankar Kolachana <quic_rkolacha@quicinc.com>
Co-authored-by: Krzysztof Parzyszek <kparzysz@quicinc.com>
* Explain store offset in a comment in launcher
Co-authored-by: Abhikrant Sharma <quic_abhikran@quicinc.com>
Co-authored-by: Ravishankar Kolachana <quic_rkolacha@quicinc.com>
Zhi [Sat, 18 Apr 2020 06:38:59 +0000 (23:38 -0700)]
fix fuse over functions that are handled by external codegen (#5365)
Marcus Shawcroft [Fri, 17 Apr 2020 18:05:42 +0000 (19:05 +0100)]
docker: Drop caffe2 download progess bars (#5359)
Change-Id: Ia15c3c8f41f75423814e559f6fdb062098f19464
Samuel [Fri, 17 Apr 2020 14:54:24 +0000 (20:24 +0530)]
[RELAY][PYTORCH]GroupNorm op support added (#5358)