x86/avx_mathfun.h: Remove fallback warnings

[kagurazakakotori]

This patch removes warning messages indicating falling back to SSE2
when AVX2 support is disabled as suggested. Also reorders non-AVX2
macros for readability and faster preprocessing.

Suggested-by: nihui [email protected]
Signed-off-by: Kagurazaka Kotori [email protected]

[kagurazakakotori]

Build test with following configurations:

Non-AVX2: cmake -DCMAKE_BUILD_TYPE=Release -DNCNN_AVX2=OFF -DNCNN_AVX=ON -DNCNN_BUILD_TESTS=ON -DNCNN_BUILD_EXAMPLES=ON
AVX2:     cmake -DCMAKE_BUILD_TYPE=Release -DNCNN_AVX2=ON -DNCNN_AVX=ON -DNCNN_BUILD_TESTS=ON -DNCNN_BUILD_EXAMPLES=ON

Both build passes.

[kagurazakakotori]

Tested only with non-AVX2 build as my build server does not support AVX2 (Intel Ivy Bridge-EP Xeon E5). All unit testes passes with some modification in test_squeezenet.

Seems the relative path of the files to be loaded is wrong, it should be ../examples.

ncnn/tests/test_squeezenet.cpp

Line 170 in 2d46994

#define MODEL_DIR "../../examples"

[codecov-commenter]

Codecov Report

Merging #3527 (08dbd1b) into master (2d46994) will decrease coverage by 0.24%.
The diff coverage is n/a.

❗ Current head 08dbd1b differs from pull request most recent head c3be8e7. Consider uploading reports for the commit c3be8e7 to get more accurate results

@@            Coverage Diff             @@
##           master    #3527      +/-   ##
==========================================
- Coverage   92.44%   92.20%   -0.25%     
==========================================
  Files         551      551              
  Lines      151934   152603     +669     
==========================================
+ Hits       140451   140702     +251     
- Misses      11483    11901     +418     

Impacted Files	Coverage Δ
src/layer/x86/avx_mathfun.h	100.00% <ø> (ø)
src/command.cpp	59.68% <0.00%> (-14.80%)	⬇️
src/layer/vulkan/flatten_vulkan.cpp	96.53% <0.00%> (-2.48%)	⬇️
src/layer/riscv/convolution_packnto1.h	97.72% <0.00%> (-2.28%)	⬇️
src/pipeline.cpp	58.69% <0.00%> (-2.18%)	⬇️
src/layer/vulkan/reshape_vulkan.cpp	91.54% <0.00%> (-2.12%)	⬇️
src/layer/riscv/deconvolution_packnto1.h	97.95% <0.00%> (-2.05%)	⬇️
src/layer/vulkan/packing_vulkan.cpp	79.46% <0.00%> (-1.85%)	⬇️
src/layer/vulkan/permute_vulkan.cpp	97.00% <0.00%> (-1.59%)	⬇️
src/layer/vulkan/reorg_vulkan.cpp	96.35% <0.00%> (-1.57%)	⬇️
... and 110 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2d46994...c3be8e7. Read the comment docs.

[kagurazakakotori]

Tested only with non-AVX2 build as my build server does not support AVX2 (Intel Ivy Bridge-EP Xeon E5). All unit testes passes with some modification in test_squeezenet.

Seems the relative path of the files to be loaded is wrong, it should be ../examples.

ncnn/tests/test_squeezenet.cpp

Line 170 in 2d46994

#define MODEL_DIR "../../examples"

My build setup issue, please ignore this.

[nihui]

Thanks for your contribution !