fix compilation on HIP SDK 6.3 for Windows - undefined behavior

[apwojcik]

The HIP SDK 6.3 compiler for Windows reports the following warning message, which is treated as an error when compiling MIGraphX on Windows.

C:\...\MIGraphX\src\api\include\migraphx\migraphx.hpp:489:12: error: dereference of type 'T *' that was reinterpret_cast from type 'X *' has undefined behavior [-Werror,-Wundefined-reinterpret-cast]
  489 |         f(*reinterpret_cast<T*>(obj), result, no_out_arg{}, xs...);
      |            ^~~~~~~~~~~~~~~~~~~~~~~~~
C:\...\MIGraphX\src\api/include\migraphx\migraphx.hpp:130:16: note: in instantiation of function template specialization 'migraphx::interface_base<migraphx::handle_base<migraphx::experimental_custom_op, migraphx_experimental_custom_op, migraphx_status (*)(migraphx_experimental_custom_op *), &migraphx_experimental_custom_op_destroy, migraphx_status (*)(migraphx_experimental_custom_op *, const migraphx_experimental_custom_op *), &migraphx_experimental_custom_op_assign_to>>::set_fp(migraphx_status (*)(migraphx_experimental_custom_op *, migraphx_status (*)(migraphx_shape *, void *, char *, unsigned long long, migraphx_shapes *)), (lambda at C:\...\MIGraphX\src\api\include\migraphx\migraphx.hpp:465:13), out_params<1>)::(anonymous class)::operator()<migraphx_shape *, migraphx_shapes *>' requested here
  130 |     auto e = f(std::forward<Ts>(xs)...);
      |                ^
C:\...\MIGraphX\src\api\include\migraphx\migraphx.hpp:438:9: note: in instantiation of function template specialization 'migraphx::call<migraphx_status (*)(migraphx_experimental_custom_op *, migraphx_status (*)(migraphx_shape *, void *, char *, unsigned long long, migraphx_shapes *)), migraphx_experimental_custom_op *, (lambda at C:\...\MIGraphX\src\api\include\migraphx\migraphx.hpp:440:14)>' requested here
  438 |         call(setter,
      |         ^
C:\...\MIGraphX\src\api\include\migraphx\migraphx.hpp:463:16: note: in instantiation of function template specialization 'migraphx::interface_base<migraphx::handle_base<migraphx::experimental_custom_op, migraphx_experimental_custom_op, migraphx_status (*)(migraphx_experimental_custom_op *), &migraphx_experimental_custom_op_destroy, migraphx_status (*)(migraphx_experimental_custom_op *, const migraphx_experimental_custom_op *), &migraphx_experimental_custom_op_assign_to>>::set_fp<sigmoid_custom_op, migraphx_status (*)(migraphx_experimental_custom_op *, migraphx_status (*)(migraphx_shape *, void *, char *, unsigned long long, migraphx_shapes *)), (lambda at C:\...\MIGraphX\src\api\include\migraphx\migraphx.hpp:465:13)>' requested here
  463 |         return set_fp<T>(
      |                ^
C:\...\MIGraphX\src\api\include\migraphx\migraphx.hpp:1568:9: note: in instantiation of function template specialization 'migraphx::interface_base<migraphx::handle_base<migraphx::experimental_custom_op, migraphx_experimental_custom_op, migraphx_status (*)(migraphx_experimental_custom_op *), &migraphx_experimental_custom_op_destroy, migraphx_status (*)(migraphx_experimental_custom_op *, const migraphx_experimental_custom_op *), &migraphx_experimental_custom_op_assign_to>>::set_auto_fp<sigmoid_custom_op, migraphx_status (*)(migraphx_experimental_custom_op *, migraphx_status (*)(migraphx_shape *, void *, char *, unsigned long long, migraphx_shapes *)), (lambda at C:\...\MIGraphX\src\api\include\migraphx\migraphx.hpp:1568:9), migraphx::out_params<1>>' requested here
 1568 |         MIGRAPHX_INTERFACE_LIFT(1, T, experimental_custom_op, compute_shape);
      |         ^
C:\...\MIGraphX\src\api\include\migraphx\migraphx.hpp:565:11: note: expanded from macro 'MIGRAPHX_INTERFACE_LIFT'
  565 |     this->set_auto_fp<T>(                               \
      |           ^
C:\...\MIGraphX\src\api\include\migraphx\migraphx.hpp:1580:28: note: in instantiation of function template specialization 'migraphx::experimental_custom_op::experimental_custom_op<sigmoid_custom_op>' requested here
 1580 |     experimental_custom_op op{obj};
      |                            ^
C:\...\MIGraphX\test\api\test_custom_op.cpp:73:15: note: in instantiation of function template specialization 'migraphx::register_experimental_custom_op<sigmoid_custom_op, void>' requested here
   73 |     migraphx::register_experimental_custom_op(sigmoid_op);

Changing to static_cast resolves the problem.

[apwojcik]

@causten We also need this PR in the branch release/rocm-rel-6.4.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.35%. Comparing base (27a90b5) to head (6c8ec26).
Report is 8 commits behind head on develop.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #3824   +/-   ##
========================================
  Coverage    92.35%   92.35%           
========================================
  Files          519      519           
  Lines        22311    22311           
========================================
  Hits         20606    20606           
  Misses        1705     1705           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[migraphx-bot]

Test	Batch	Rate new 6c8ec2	Rate old bfc14e	Diff	Compare
torchvision-resnet50	64	3,237.23	3,233.96	0.10%	✅
torchvision-resnet50_fp16	64	6,866.95	6,874.33	-0.11%	✅
torchvision-densenet121	32	2,439.12	2,436.94	0.09%	✅
torchvision-densenet121_fp16	32	4,192.43	4,181.18	0.27%	✅
torchvision-inceptionv3	32	1,613.72	1,612.90	0.05%	✅
torchvision-inceptionv3_fp16	32	2,673.35	2,677.82	-0.17%	✅
cadene-inceptionv4	16	750.82	750.47	0.05%	✅
cadene-resnext64x4	16	810.54	809.72	0.10%	✅
slim-mobilenet	64	6,661.84	6,659.15	0.04%	✅
slim-nasnetalarge	64	196.93	196.77	0.08%	✅
slim-resnet50v2	64	3,431.01	3,430.42	0.02%	✅
bert-mrpc-onnx	8	1,134.75	1,133.14	0.14%	✅
bert-mrpc-tf	1	505.05	500.00	1.01%	✅
pytorch-examples-wlang-gru	1	480.87	470.97	2.10%	✅
pytorch-examples-wlang-lstm	1	447.20	437.32	2.26%	✅
torchvision-resnet50_1	1	793.65	793.65	0.00%	✅
cadene-dpn92_1	1	429.51	430.36	-0.20%	✅
cadene-resnext101_1	1	389.11	389.11	0.00%	✅
onnx-taau-downsample	1	371.29	371.31	-0.00%	✅
dlrm-criteoterabyte	1	31.83	31.79	0.12%	✅
dlrm-criteoterabyte_fp16	1	51.05	51.06	-0.02%	✅
agentmodel	1	8,761.78	8,658.38	1.19%	✅
unet_fp16	2	58.26	58.36	-0.17%	✅
resnet50v1_fp16	1	1,020.41	1,075.27	-5.10%	🔴
resnet50v1_int8	1	793.65	800.00	-0.79%	✅
bert_base_cased_fp16	64	1,159.42	1,159.42	0.00%	✅
bert_large_uncased_fp16	32	360.04	359.96	0.02%	✅
bert_large_fp16	1	217.39	215.98	0.65%	✅
distilgpt2_fp16	16	2,188.78	2,188.78	0.00%	✅
yolov5s	1	538.82	531.08	1.46%	✅
tinyllama	1	49.09	49.12	-0.05%	✅
vicuna-fastchat	1	43.58	43.55	0.07%	✅
whisper-tiny-encoder	1	436.68	436.68	0.00%	✅
whisper-tiny-decoder	1	450.45	448.43	0.45%	✅

This build is not recommended to merge 🔴

[migraphx-bot]

     ✅ bert-mrpc-onnx: PASSED: MIGraphX meets tolerance

     ✅ bert-mrpc-tf: PASSED: MIGraphX meets tolerance

     ✅ pytorch-examples-wlang-gru: PASSED: MIGraphX meets tolerance

     ✅ pytorch-examples-wlang-lstm: PASSED: MIGraphX meets tolerance

     ✅ torchvision-resnet50_1: PASSED: MIGraphX meets tolerance

     ✅ cadene-dpn92_1: PASSED: MIGraphX meets tolerance

     ✅ cadene-resnext101_1: PASSED: MIGraphX meets tolerance

     ✅ dlrm-criteoterabyte: PASSED: MIGraphX meets tolerance

     ✅ agentmodel: PASSED: MIGraphX meets tolerance

     ✅ unet: PASSED: MIGraphX meets tolerance

     ✅ resnet50v1: PASSED: MIGraphX meets tolerance

     ✅ bert_base_cased_fp16: PASSED: MIGraphX meets tolerance

🔴bert_large_uncased_fp16: FAILED: MIGraphX is not within tolerance - check verbose output

     ✅ bert_large: PASSED: MIGraphX meets tolerance

     ✅ yolov5s: PASSED: MIGraphX meets tolerance

     ✅ tinyllama: PASSED: MIGraphX meets tolerance

     ✅ vicuna-fastchat: PASSED: MIGraphX meets tolerance

     ✅ whisper-tiny-encoder: PASSED: MIGraphX meets tolerance

     ✅ whisper-tiny-decoder: PASSED: MIGraphX meets tolerance

     ✅ distilgpt2_fp16: PASSED: MIGraphX meets tolerance