Loop subgraph support

[weizhu12-amd]

Motivation

MIGraphX cannot run the YOLOv3 model because the model contains loop subgraphs. This PR resolves the parsing and compiling issues associated with this subgraph.

Technical Details

1, Handle duplicate loop subgraphs during parsing
YOLOv3 may contain duplicate loop subgraphs, causing a duplicate-key insertion into the subgraph map and resulting in a crash. Preserve the original parent subgraph during parsing and restore it after parsing is complete.

2, Insert copy and allocation instructions for identity IR
Loop operations are lowered to identity IR. Copy and allocation instructions need to be inserted between consecutive identity instructions. However, the current implementation skips this.
This change ensures that the required copy and allocation instructions are generated correctly for identity IR.

Changelog Category

Add a CHANGELOG.md entry for any option other than Not Applicable

• • Added: New functionality.
• • Changed: Changes to existing functionality.
• • Removed: Functionality or support that has been removed. (Compared to a previous release)
• • Optimized: Component performance that has been optimized or improved.
• • Resolved Issues: Known issues from a previous version that have been resolved.
• • Not Applicable: This PR is not to be included in the changelog.

Follow the LLVM AI Tool Use Policy for contributions using AI.

[gh-app-migraphx-bot-pr-write]

Test	Batch	New Rate (4cca90)	Old Rate (0c63af)	Diff	Status
torchvision-resnet50	64	3,266.36	3,266.06	0.01%	✅
torchvision-resnet50_fp16	64	7,545.37	7,546.56	-0.02%	✅
torchvision-densenet121	32	2,484.11	2,482.18	0.08%	✅
torchvision-densenet121_fp16	32	5,010.03	4,976.34	0.68%	✅
torchvision-inceptionv3	32	2,060.58	2,057.16	0.17%	✅
torchvision-inceptionv3_fp16	32	4,416.28	4,415.87	0.01%	✅
cadene-inceptionv4	16	nan	818.77	nan	❌
cadene-resnext64x4	16	783.27	783.71	-0.06%	✅
slim-mobilenet	64	8,388.23	8,356.38	0.38%	✅
slim-nasnetalarge	64	nan	229.42	nan	❌
slim-resnet50v2	64	3,180.60	3,185.54	-0.16%	✅
bert-mrpc-onnx	8	1,169.69	1,167.46	0.19%	✅
bert-mrpc-tf	1	496.02	486.41	1.97%	✅
pytorch-examples-wlang-gru	1	476.54	480.59	-0.84%	✅
pytorch-examples-wlang-lstm	1	378.71	390.19	-2.94%	✅
torchvision-resnet50_1	1	1,040.89	1,037.04	0.37%	✅
cadene-dpn92_1	1	444.89	475.84	-6.50%	🔴
cadene-resnext101_1	1	365.14	365.49	-0.09%	✅
onnx-taau-downsample	1	842.01	843.97	-0.23%	✅
dlrm-criteoterabyte	1	nan	32.42	nan	❌
dlrm-criteoterabyte_fp16	1	nan	52.07	nan	❌
agentmodel	1	9,344.31	9,645.99	-3.13%	✅
unet_fp16	2	58.75	58.92	-0.28%	✅
resnet50v1_fp16	1	1,372.04	1,412.17	-2.84%	✅
resnet50v1_int8	1	nan	1,602.24	nan	❌
bert_base_cased_fp16	64	1,097.40	1,098.36	-0.09%	✅
bert_large_uncased_fp16	32	345.69	344.75	0.27%	✅
bert_large_fp16	1	205.21	204.70	0.25%	✅
distilgpt2_fp16	16	2,091.74	2,090.27	0.07%	✅
yolov5s	1	nan	560.47	nan	❌
tinyllama	1	nan	45.79	nan	❌
vicuna-fastchat	1	44.15	44.12	0.07%	✅
whisper-tiny-encoder	1	412.46	412.35	0.03%	✅
whisper-tiny-decoder	1	409.09	413.51	-1.07%	✅
llama2_7b	1	nan	20.84	nan	❌
qwen1.5-7b	1	nan	23.52	nan	❌
phi3-3.8b	1	nan	26.73	nan	❌
llama3-8b	1	nan	21.66	nan	❌
whisper-large-encoder	1	10.18	10.18	-0.07%	✅
whisper-large-decoder	1	103.69	106.43	-2.58%	✅
mistral-7b	1	nan	23.62	nan	❌
FLUX.1-schnell	1	774.45	756.55	2.37%	✅

Regressions detected 🔴

[gh-app-migraphx-bot-pr-write]

Test	Status	Result
bert-mrpc-onnx	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
bert-mrpc-tf	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
pytorch-examples-wlang-gru	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
pytorch-examples-wlang-lstm	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
dlrm-criteoterabyte	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
agentmodel	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
unet	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
resnet50v1	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
bert_base_cased_fp16	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
bert_large_uncased_fp16	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
bert_large	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
yolov5s	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
tinyllama	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
vicuna-fastchat	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
whisper-tiny-encoder	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
whisper-tiny-decoder	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
distilgpt2_fp16	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
llama2_7b	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
qwen1.5-7b	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
phi3-3.8b	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
llama3-8b	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
whisper-large-encoder	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
whisper-large-decoder	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
mistral-7b	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'
FLUX.1-schnell	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 27, in import onnxruntime as ort ModuleNotFoundError: No module named 'onnxruntime'

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #5103      +/-   ##
===========================================
+ Coverage    92.90%   93.26%   +0.36%     
===========================================
  Files          603      623      +20     
  Lines        32526    32969     +443     
===========================================
+ Hits         30217    30746     +529     
+ Misses        2309     2223      -86     

Files with missing lines	Coverage Δ
src/onnx/onnx_parser.cpp	88.31% <ø> (-0.03%)	⬇️
src/replace_allocate.cpp	98.89% <100.00%> (-1.11%)	⬇️

... and 42 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.