add migraphx load and save functions, allowing the hiding of an error message

[apwojcik]

The MIGraphX EP uses the migraphx_load and migraphx_save functions for MXR caching. If a model is not yet compiled (file missing from cache), a message in the console states an error occurred, which misled users. The v2 version of the functions allows error message reporting to be turned off.

The PR introduces the migraphx_load_v2 and migraphx_save_v2 functions. The original functions were replaced with macros, emulating the old behavior with the help of v2 functions.

[TedThemistokleous]

Did you test this with a rebuild and the latest stuff off repo.radeon.com?

Concerned since we're removing the old signature and Cpp name mangles based on arguments you'll have a runtime issue between libraries if a user updates MIGraphX package with this change but doesn't rebuild Onnxruntime with the new macro.

@apwojcik can you try the following with this change

1. Install the latest Onnxruntime from repo.radeon on your system with MIGraphX latest 2 . Build this change as a package and install it over the existing MIGraphX in the system
2. Try to run a model with the Onnxruntime MIGraphX EP

If you hit an error you may need to put the old signature back in for now.

[codecov[bot]]

Codecov Report

Attention: Patch coverage is 88.46154% with 3 lines in your changes missing coverage. Please review.

Project coverage is 92.04%. Comparing base (e230c02) to head (eaac578). Report is 1 commits behind head on develop.

Files with missing lines	Patch %	Lines
src/api/api.cpp	81.25%	3 Missing :warning:

Additional details and impacted files

```diff @@ Coverage Diff @@ ## develop #3473 +/- ## ======================================== Coverage 92.04% 92.04% ======================================== Files 506 506 Lines 20872 20874 +2 ======================================== + Hits 19212 19214 +2 Misses 1660 1660 ``` | [Flag](https://app.codecov.io/gh/ROCm/AMDMIGraphX/pull/3473/flags?src=pr&el=flags&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=ROCm) | Coverage Δ | | |---|---|---| | [](https://app.codecov.io/gh/ROCm/AMDMIGraphX/pull/3473/flags?src=pr&el=flag&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=ROCm) | `92.04% <88.46%> (+<0.01%)` | :arrow_up: | Flags with carried forward coverage won't be shown. [Click here](https://docs.codecov.io/docs/carryforward-flags?utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=ROCm#carryforward-flags-in-the-pull-request-comment) to find out more.

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[migraphx-bot]

Test	Batch	Rate new eaac57	Rate old ab5da3	Diff	Compare
torchvision-resnet50	64	3,250.98	3,250.20	0.02%	:white_check_mark:
torchvision-resnet50_fp16	64	6,984.86	6,981.72	0.04%	:white_check_mark:
torchvision-densenet121	32	2,433.90	2,434.35	-0.02%	:white_check_mark:
torchvision-densenet121_fp16	32	4,096.08	4,104.80	-0.21%	:white_check_mark:
torchvision-inceptionv3	32	1,637.02	1,637.98	-0.06%	:white_check_mark:
torchvision-inceptionv3_fp16	32	2,743.11	2,743.88	-0.03%	:white_check_mark:
cadene-inceptionv4	16	778.86	778.58	0.04%	:white_check_mark:
cadene-resnext64x4	16	807.68	808.70	-0.13%	:white_check_mark:
slim-mobilenet	64	7,458.23	7,457.95	0.00%	:white_check_mark:
slim-nasnetalarge	64	208.35	208.52	-0.08%	:white_check_mark:
slim-resnet50v2	64	3,437.13	3,435.85	0.04%	:white_check_mark:
bert-mrpc-onnx	8	1,152.49	1,153.18	-0.06%	:white_check_mark:
bert-mrpc-tf	1	310.58	305.52	1.66%	:white_check_mark:
pytorch-examples-wlang-gru	1	415.99	384.89	8.08%	:high_brightness:
pytorch-examples-wlang-lstm	1	386.88	386.28	0.16%	:white_check_mark:
torchvision-resnet50_1	1	755.10	768.57	-1.75%	:white_check_mark:
cadene-dpn92_1	1	434.34	437.45	-0.71%	:white_check_mark:
cadene-resnext101_1	1	374.14	383.50	-2.44%	:white_check_mark:
onnx-taau-downsample	1	344.12	344.79	-0.20%	:white_check_mark:
dlrm-criteoterabyte	1	35.05	35.03	0.06%	:white_check_mark:
dlrm-criteoterabyte_fp16	1	58.06	58.06	-0.00%	:white_check_mark:
agentmodel	1	8,260.98	7,924.68	4.24%	:high_brightness:
unet_fp16	2	58.00	57.94	0.11%	:white_check_mark:
resnet50v1_fp16	1	930.73	929.40	0.14%	:white_check_mark:
resnet50v1_int8	1	941.96	960.30	-1.91%	:white_check_mark:
bert_base_cased_fp16	64	1,153.83	1,153.43	0.03%	:white_check_mark:
bert_large_uncased_fp16	32	355.95	355.83	0.03%	:white_check_mark:
bert_large_fp16	1	211.79	210.44	0.64%	:white_check_mark:
distilgpt2_fp16	16	2,158.66	2,159.20	-0.03%	:white_check_mark:
yolov5s	1	533.17	539.05	-1.09%	:white_check_mark:
tinyllama	1	43.38	43.40	-0.06%	:white_check_mark:
vicuna-fastchat	1	175.08	170.95	2.41%	:white_check_mark:
whisper-tiny-encoder	1	418.14	417.95	0.04%	:white_check_mark:
whisper-tiny-decoder	1	424.17	423.97	0.05%	:white_check_mark:

Check results before merge :high_brightness:

[migraphx-bot]

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

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

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

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

     :white_check_mark: torchvision-resnet50_1: PASSED: MIGraphX meets tolerance

     :white_check_mark: cadene-dpn92_1: PASSED: MIGraphX meets tolerance

     :white_check_mark: cadene-resnext101_1: PASSED: MIGraphX meets tolerance

     :white_check_mark: dlrm-criteoterabyte: PASSED: MIGraphX meets tolerance

     :white_check_mark: agentmodel: PASSED: MIGraphX meets tolerance

     :white_check_mark: unet: PASSED: MIGraphX meets tolerance

     :white_check_mark: resnet50v1: PASSED: MIGraphX meets tolerance

     :white_check_mark: bert_base_cased_fp16: PASSED: MIGraphX meets tolerance

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

     :white_check_mark: bert_large: PASSED: MIGraphX meets tolerance

     :white_check_mark: yolov5s: PASSED: MIGraphX meets tolerance

     :white_check_mark: tinyllama: PASSED: MIGraphX meets tolerance

     :white_check_mark: vicuna-fastchat: PASSED: MIGraphX meets tolerance

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

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

     :white_check_mark: distilgpt2_fp16: PASSED: MIGraphX meets tolerance

[pfultz2]

This breaks binary compatibility. Why cant we just check in onnxruntime if the file exists first before we load it?

[pfultz2]

Concerned since we're removing the old signature and Cpp name mangles based on arguments you'll have a runtime issue between libraries if a user updates MIGraphX package with this change but doesn't rebuild Onnxruntime with the new macro.

There is no name mangling since this a C API, but you are correct that binary compatibility is broken. A simpler solution would be to just check if the file exists before loading it.

[apwojcik]

This breaks binary compatibility. Why cant we just check in onnxruntime if the file exists first before we load it?

That would be too easy :-) But you have a valid point. I am closing this PR. Thanks!

[apwojcik]

Did you test this with a rebuild and the latest stuff off repo.radeon.com?

Concerned since we're removing the old signature and Cpp name mangles based on arguments you'll have a runtime issue between libraries if a user updates MIGraphX package with this change but doesn't rebuild Onnxruntime with the new macro.

@apwojcik can you try the following with this change

1. Install the latest Onnxruntime from repo.radeon on your system with MIGraphX latest 2 . Build this change as a package and install it over the existing MIGraphX in the system
2. Try to run a model with the Onnxruntime MIGraphX EP

If you hit an error you may need to put the old signature back in for now.

Ted, your concern is justified. I have not considered such a scenario. Anyway, I am closing the PR and using Paul's suggested more straightforward solution.

[apwojcik]

Closing the PR.