* Reduce LLC loads, stores and multiplies on MulTransposed - 8% faster on VSX
* Add is_same method so c++11 is not required
* Remove trailing whitespaces.
* Change is_same to DataType depth check
Vectorize minMaxIdx functions
* Updated documentation and intrinsic tests for v_reduce
* Add other files back in from the forced push
* Prevent an constant overflow with v_reduce for int8 type
* Another alternative to fix constant overflow warning.
* Fix another compiler warning.
* Update comments and change comparison form to be consistent with other vectorized loops.
* Change return type of v_reduce_min & max for v_uint8 and v_uint16 to be same as lane type.
* Cast v_reduce functions to int to avoid overflow. Reduce number of parameters in MINMAXIDX_REDUCE macro.
* Restore cast type for v_reduce_min & max to LaneType
Add checks for empty operands in Matrix expressions that don't check properly
* Starting to add checks for empty operands in Matrix expressions that
don't check properly.
* Adding checks and delcarations for checker functions
* Fix signatures and add checks for each class of Matrix Expr operation
* Make it catch the right exception
* Don't expose helper functions to public API
* resize: HResizeLinear reduce duplicate work
There appears to be a 2x unroll of the HResizeLinear against k,
however the k value is only incremented by 1 during the unroll. This
results in k - 1 duplicate passes when k > 1.
Likewise, the final pass may not respect the work done by the vector
loop. Start it with the offset returned by the vector op if
implemented. Note, no vector ops are implemented today.
The performance is most noticable on a linear downscale. A set of
performance tests are added to characterize this. The performance
improvement is 10-50% depending on the scaling.
* imgproc: vectorize HResizeLinear
Performance is mostly gated by the gather operations
for x inputs.
Likewise, provide a 2x unroll against k, this reduces the
number of alpha gathers by 1/2 for larger k.
While not a 4x improvement, it still performs substantially
better under P9 for a 1.4x improvement. P8 baseline is
1.05-1.10x due to reduced VSX instruction set.
For float types, this results in a more modest
1.2x improvement.
* Update U8 processing for non-bitexact linear resize
* core: hal: vsx: improve v_load_expand_q
With a little help, we can do this quickly without gprs on
all VSX enabled targets.
* resize: Fix cn == 3 step per feedback
Per feedback, ensure we don't overrun. This was caught via the
failure observed in Test_TensorFlow.inception_accuracy.
* calib3d: use normalized input in solvePnPGeneric()
* calib3d: java regression test for solvePnPGeneric
* calib3d: python regression test for solvePnPGeneric
* core: disable invalid constructors in C API by default
- C API objects will lose their default initializers through constructors
* samples: stop using of C API
Tests for argument conversion of Python bindings generator
* Tests for parsing elemental types from Python bindings
- Add positive and negative tests for int, float, double, size_t,
const char*, bool.
- Tests with wrong conversion behavior are skipped.
* Move implicit conversion of bool to integer/floating types to wrong
conversion behavior.
Improving VSX performance of integral function
* Adding support for vector get function on VSX datatypes so the
integral function gains a bit of performance.
* Removing get as a datatype member function and implementing a new HAL
instruction v_extract_n to get the n-th element of a vector register.
* Adding SSE/NEON/AVX intrinsics.
* Implement new HAL instruction v_broadcast_element on VSX/AVX/NEON/SSE.
* core(simd): add tests for v_extract_n/v_broadcast_element
- updated docs
- commented out code to repair compilation
- added WASM and MSA default implementations
* core(simd): fix compilation
- x86: avoid _mm256_extract_epi64/32/16/8 with MSVS 2015
- x86: _mm_extract_epi64 is 64-bit only
* cleanup
* Use FlsAlloc/FlsFree/FlsGetValue/FlsSetValue instead of TlsAlloc/TlsFree/TlsGetValue/TlsSetValue to implment TLS value cleanup when thread has been terminated on Windows Vista and above
* Fix 32-bit build
* Fixed calling convention of cleanup callback
* WINAPI changed to NTAPI
* Use proper guard macro
* Vectorize flipHoriz and flipVert functions.
* Change v_load_mirror_1 to use vec_revb for VSX
* Only use vec_revb in ISA3.0
* Removing vec_revb code since some of the older compilers don't fully support it.
* Use new v_reverse intrinsic and cleanup code.
* Ensure there are no alignment issues with copies
- move TLS & instrumentation code out of core/utility.hpp
- (*) TLSData lost .gather() method (to dispose thread data on thread termination)
- use TLSDataAccumulator for reliable collecting of thread data
- prefer using of .detachData() + .cleanupDetachedData() instead of .gather() method
(*) API is broken: replace TLSData => TLSDataAccumulator if gather required
(objects disposal on threads termination is not available in accumulator mode)
Fixing bug with comparison of v_int64x2 or v_uint64x2
* Casting v_uint64x2 to v_float64x2 and comparing does NOT work in all cases. Rewrite using epi64 instructions - faster too.
* Fix bad merge.
* Fix equal comparsion for non-SSE4.1. Add test cases for v_int64x2 comparisons.
* Try to fix merge conflict.
* Only test v_int64x2 comparisons if CV_SIMD_64F
* Fix compiler warning.
* New v_reverse HAL intrinsic for reversing the ordering of a vector
* Fix conflict.
* Try to resolve conflict again.
* Try one more time.
* Add _MM_SHUFFLE. Remove non-vectorize code in SSE2. Fix copy and paste issue with NEON.
* Change v_uint16x8 SSE2 version to use shuffles
* Cuda + OpenGL on ARM
There might be multiple ways of getting OpenCV compile on Tegra (NVIDIA Jetson) platform, but mainly they modify CUDA(8,9,10...) source code, this one fixes it for all installations.
( https://devtalk.nvidia.com/default/topic/1007290/jetson-tx2/building-opencv-with-opengl-support-/post/5141945/#5141945 et al.).
This way is exactly the same as the one proposed but the code change happens in OpenCV.
* Updated,
The link provided mentions: cuda8 + 9, I have cuda 10 + 10.1 (and can confirm it is still defined this way).
NVIDIA is probably using some other "secret" backend with Jetson.
* core: rework and optimize SIMD implementation of dotProd
- add new universal intrinsics v_dotprod[int32], v_dotprod_expand[u&int8, u&int16, int32], v_cvt_f64(int64)
- add a boolean param for all v_dotprod&_expand intrinsics that change the behavior of addition order between
pairs in some platforms in order to reach the maximum optimization when the sum among all lanes is what only matters
- fix clang build on ppc64le
- support wide universal intrinsics for dotProd_32s
- remove raw SIMD and activate universal intrinsics for dotProd_8
- implement SIMD optimization for dotProd_s16&u16
- extend performance test data types of dotprod
- fix GCC VSX workaround of vec_mule and vec_mulo (in little-endian it must be swapped)
- optimize v_mul_expand(int32) on VSX
* core: remove boolean param from v_dotprod&_expand and implement v_dotprod_fast&v_dotprod_expand_fast
this changes made depend on "terfendail" review
- renamed Cascade Lake AVX512_CEL => AVX512_CLX (align with Intel SDE tool)
- fixed CLX instruction sets (no IFMA/VBMI)
- added flag to bypass CPU baseline check: OPENCV_SKIP_CPU_BASELINE_CHECK
[GSoC 2019] Improve the performance of JavaScript version of OpenCV (OpenCV.js)
* [GSoC 2019]
Improve the performance of JavaScript version of OpenCV (OpenCV.js):
1. Create the base of OpenCV.js performance test:
This perf test is based on benchmark.js(https://benchmarkjs.com). And first add `cvtColor`, `Resize`, `Threshold` into it.
2. Optimize the OpenCV.js performance by WASM threads:
This optimization is based on Web Worker API and SharedArrayBuffer, so it can be only used in browser.
3. Optimize the OpenCV.js performance by WASM SIMD:
Add WASM SIMD backend for OpenCV Universal Intrinsics. It's experimental as WASM SIMD is still in development.
* [GSoC2019]
1. use short license header
2. fix documentation node issue
3. remove the unused `hasSIMD128()` api
* [GSoC2019]
1. fix emscripten define
2. use fallback function for f16
* [GSoC2019]
Fix rebase issue
* Added MSA implementations for mips platforms. Intrinsics for MSA and build scripts for MIPS platforms are added.
Signed-off-by: Fei Wu <fwu@wavecomp.com>
* Removed some unused code in mips.toolchain.cmake.
Signed-off-by: Fei Wu <fwu@wavecomp.com>
* Added comments for mips toolchain configuration and disabled compiling warnings for libpng.
Signed-off-by: Fei Wu <fwu@wavecomp.com>
* Fixed the build error of unsupported opcode 'pause' when mips isa_rev is less than 2.
Signed-off-by: Fei Wu <fwu@wavecomp.com>
* 1. Removed FP16 related item in MSA option defines in OpenCVCompilerOptimizations.cmake.
2. Use CV_CPU_COMPILE_MSA instead of __mips_msa for MSA feature check in cv_cpu_dispatch.h.
3. Removed hasSIMD128() in intrin_msa.hpp.
4. Define CPU_MSA as 150.
Signed-off-by: Fei Wu <fwu@wavecomp.com>
* 1. Removed unnecessary CV_SIMD128_64F guarding in intrin_msa.hpp.
2. Removed unnecessary CV_MSA related code block in dotProd_8u().
Signed-off-by: Fei Wu <fwu@wavecomp.com>
* 1. Defined CPU_MSA_FLAGS_ON as "-mmsa".
2. Removed CV_SIMD128_64F guardings in intrin_msa.hpp.
Signed-off-by: Fei Wu <fwu@wavecomp.com>
* Removed unused msa_mlal_u16() and msa_mlal_s16 from msa_macros.h.
Signed-off-by: Fei Wu <fwu@wavecomp.com>
ISA 2.07 (aka POWER8) effectively extended the expanding multiply
operation to word types. The altivec intrinsics prior to gcc 8 did
not get the update.
Workaround this deficiency similar to other fixes.
This was exposed by commit 33fb253a66
which leverages the int -> dword expanding multiply.
This fixes Issue #15506
Use 4x FMA chains to sum on SIMD 128 FP64 targets. On
x86 this showed about 1.4x improvement.
For PPC, do a full multiply (32x32->64b), convert to DP
then accumulate. This may be slightly less precise for
some inputs. But is 1.5x faster than the above which
is about 1.5x than the FMA above for ~2.5x speedup.
Implement cvRound using inline asm. No compiler support
exists today to properly optimize this. This results in
about a 4x speedup over the default rounding. Likewise,
simplify the growing number of rounding function overloads.
For P9 enabled targets, utilize the classification
testing instruction to test for Inf/Nan values. Operation
speedup is about 1.2x for FP32, and 1.5x for FP64 operands.
For P8 targets, fallback to the GCC nan inline. It provides
a 1.1/1.4x improvement for FP32/FP64 arguments.
Add a new macro definition OPENCV_USE_FASTMATH_GCC_BUILTINS to enable
usage of GCC inline math functions, if available and requested by the
user.
Likewise, enable it for POWER. This is nearly always a substantial
improvement over using integer manipulation as most operations can
be done in several instructions with no branching. The result is a
1.5-1.8x speedup in the ceil/floor operations.
1. As tested with AT 12.0-1 (GCC 8.3.1) compiler on P9 LE.
Add a basic sanity test to verify the rounding functions
work as expected.
Likewise, extend the rounding performance test to cover the
additional float -> int fast math functions.
Due to the explicitly declared copy constructor Vec<T, n>::Vec(Vec <T,n>&)
GCC 9 warns if there is no assignment operator, as having one typically
requires the other (rule-of-three, constructor/desctructor/assginment).
As the values are just a plain array the default assignment operator does
the right thing. Tell the compiler explicitly to default it.
Signed-off-by: Stefan Brüns <stefan.bruens@rwth-aachen.de>
* core: improve AVX512 infrastructure by adding more CPU features groups
* cmake: use groups for AVX512 optimization flags
* core: remove gap in CPU flags enumeration
* cmake: restore default CPU_DISPATCH
