[AIRADSW-468] Fix Windows hang by draining the output pipe before waiting

[ivarusic-amd]

Motivation

Multiple models hanging on migraphx-driver compile step WINDOWS.

Technical Details

When migraphx-hiprtc-driver produced enough stdout/stderr to fill the Windows pipe buffer, the parent process would deadlock: child blocked writing to a full pipe, parent blocked in WaitForSingleObject waiting for the child to exit.
Fixed by draining the output pipe after closing stdin and before WaitForSingleObject

Linux should not be affected.

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.

[causten]

How do you want to handle the warning? Empty While Loop. Without something the code will hold the CPU in a tight loop

[ivarusic-amd]

@causten I change it, can you take a look. Thanks

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

Test	Batch	New Rate (e49c31)	Old Rate (413902)	Diff	Status
torchvision-resnet50	64	2,092.75	2,180.39	-4.02%	✅
torchvision-resnet50_fp16	64	6,529.97	2,618.95	149.34%	🔆
torchvision-densenet121	32	2,676.32	2,590.80	3.30%	✅
torchvision-densenet121_fp16	32	4,540.30	4,556.87	-0.36%	✅
torchvision-inceptionv3	32	1,712.17	1,797.28	-4.74%	✅
torchvision-inceptionv3_fp16	32	2,814.20	2,826.73	-0.44%	✅
cadene-inceptionv4	16	812.22	825.60	-1.62%	✅
cadene-resnext64x4	16	550.17	784.52	-29.87%	🔴
slim-mobilenet	64	8,411.36	8,431.73	-0.24%	✅
slim-nasnetalarge	64	225.73	229.17	-1.50%	✅
slim-resnet50v2	64	1,952.86	2,839.08	-31.21%	🔴
bert-mrpc-onnx	8	508.39	1,170.04	-56.55%	🔴
bert-mrpc-tf	1	486.53	485.19	0.28%	✅
pytorch-examples-wlang-gru	1	336.31	334.54	0.53%	✅
pytorch-examples-wlang-lstm	1	466.80	460.96	1.27%	✅
torchvision-resnet50_1	1	784.46	767.87	2.16%	✅
cadene-dpn92_1	1	447.49	447.01	0.11%	✅
cadene-resnext101_1	1	363.99	363.58	0.12%	✅
onnx-taau-downsample	1	400.11	401.61	-0.37%	✅
dlrm-criteoterabyte	1	28.78	32.69	-11.98%	🔴
dlrm-criteoterabyte_fp16	1	52.69	52.37	0.60%	✅
agentmodel	1	10,435.74	9,867.77	5.76%	🔆
unet_fp16	2	54.39	21.15	157.17%	🔆
resnet50v1_fp16	1	941.28	922.03	2.09%	✅
resnet50v1_int8	1	940.79	940.01	0.08%	✅
bert_base_cased_fp16	64	1,103.54	1,103.36	0.02%	✅
bert_large_uncased_fp16	32	347.74	347.72	0.01%	✅
bert_large_fp16	1	124.75	204.50	-39.00%	🔴
distilgpt2_fp16	16	2,096.95	1,575.45	33.10%	🔆
yolov5s	1	573.38	466.21	22.99%	🔆
tinyllama	1	45.98	35.55	29.35%	🔆
vicuna-fastchat	1	44.04	43.74	0.68%	✅
whisper-tiny-encoder	1	252.95	418.56	-39.57%	🔴
whisper-tiny-decoder	1	419.47	377.94	10.99%	🔆
llama2_7b	1	20.45	9.87	107.18%	🔆
qwen1.5-7b	1	23.68	23.61	0.28%	✅
phi3-3.8b	1	26.83	26.82	0.04%	✅
llama3-8b	1	21.83	21.78	0.21%	✅
whisper-large-encoder	1	10.32	10.33	-0.08%	✅
whisper-large-decoder	1	104.15	105.04	-0.85%	✅
mistral-7b	1	23.82	13.41	77.63%	🔆
FLUX.1-schnell	1	746.65	793.08	-5.86%	🔴

Regressions detected 🔴

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

Test	Status	Result
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
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
llama2_7b	✅	PASSED: MIGraphX meets tolerance
qwen1.5-7b	✅	PASSED: MIGraphX meets tolerance
phi3-3.8b	✅	PASSED: MIGraphX meets tolerance
llama3-8b	✅	PASSED: MIGraphX meets tolerance
whisper-large-encoder	❌	ERROR - check error output traceback Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 377, in main() File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 224, in main model = migraphx.parse_onnx(model_name, default_dim_value=batch) RuntimeError: /data/src/include/migraphx/op/convolution.hpp:103: normalize_compute_shape: CONVOLUTION: mismatched channel numbers
whisper-large-decoder	✅	PASSED: MIGraphX meets tolerance
mistral-7b	✅	PASSED: MIGraphX meets tolerance
FLUX.1-schnell	✅	PASSED: MIGraphX meets tolerance