Fix nonzero for non-standard input layouts

[ikalinic]

Motivation

Fix the reference nonzero operator so it accepts and correctly reads non-standard input layouts such as transposed and broadcasted tensors.

Technical Details

Removed the standard-layout requirement from nonzero shape checking and changed compute-time element access to use the logical index from shape_for_each. Added reference and verify tests for transposed inputs, plus reference coverage for broadcasted inputs.

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.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #5046      +/-   ##
===========================================
+ Coverage    92.78%   92.90%   +0.12%     
===========================================
  Files          596      603       +7     
  Lines        32011    32524     +513     
===========================================
+ Hits         29700    30214     +514     
+ Misses        2311     2310       -1     

Files with missing lines	Coverage Δ
src/include/migraphx/op/nonzero.hpp	95.45% <100.00%> (-4.55%)	⬇️

... and 29 files with indirect coverage changes

🚀 New features to boost your workflow:

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

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

Test	Batch	New Rate (791bc4)	Old Rate (e1e03e)	Diff	Status
torchvision-resnet50	64	1,428.98	2,693.41	-46.95%	🔴
torchvision-resnet50_fp16	64	6,568.04	6,415.71	2.37%	✅
torchvision-densenet121	32	2,684.86	2,702.20	-0.64%	✅
torchvision-densenet121_fp16	32	4,504.53	4,553.39	-1.07%	✅
torchvision-inceptionv3	32	1,797.70	950.94	89.04%	🔆
torchvision-inceptionv3_fp16	32	2,820.20	2,269.97	24.24%	🔆
cadene-inceptionv4	16	821.87	538.17	52.72%	🔆
cadene-resnext64x4	16	783.58	459.28	70.61%	🔆
slim-mobilenet	64	8,382.68	8,429.89	-0.56%	✅
slim-nasnetalarge	64	228.85	89.65	155.28%	🔆
slim-resnet50v2	64	3,162.27	1,563.17	102.30%	🔆
bert-mrpc-onnx	8	1,171.48	1,165.41	0.52%	✅
bert-mrpc-tf	1	488.76	486.44	0.48%	✅
pytorch-examples-wlang-gru	1	489.30	487.62	0.34%	✅
pytorch-examples-wlang-lstm	1	384.51	392.17	-1.95%	✅
torchvision-resnet50_1	1	753.22	771.94	-2.43%	✅
cadene-dpn92_1	1	441.26	442.37	-0.25%	✅
cadene-resnext101_1	1	366.08	364.35	0.48%	✅
onnx-taau-downsample	1	401.04	400.76	0.07%	✅
dlrm-criteoterabyte	1	32.42	32.68	-0.78%	✅
dlrm-criteoterabyte_fp16	1	51.80	52.60	-1.51%	✅
agentmodel	1	9,290.57	9,162.71	1.40%	✅
unet_fp16	2	57.03	57.35	-0.55%	✅
resnet50v1_fp16	1	933.86	1,017.75	-8.24%	🔴
resnet50v1_int8	1	943.46	937.30	0.66%	✅
bert_base_cased_fp16	64	1,097.63	1,102.37	-0.43%	✅
bert_large_uncased_fp16	32	345.62	346.79	-0.34%	✅
bert_large_fp16	1	204.35	203.94	0.20%	✅
distilgpt2_fp16	16	2,095.43	2,094.08	0.06%	✅
yolov5s	1	559.88	560.76	-0.16%	✅
tinyllama	1	45.83	45.87	-0.10%	✅
vicuna-fastchat	1	44.02	44.20	-0.40%	✅
whisper-tiny-encoder	1	413.32	415.10	-0.43%	✅
whisper-tiny-decoder	1	410.23	409.89	0.08%	✅
llama2_7b	1	20.82	20.96	-0.65%	✅
qwen1.5-7b	1	23.58	23.66	-0.34%	✅
phi3-3.8b	1	26.72	26.82	-0.37%	✅
llama3-8b	1	21.74	21.88	-0.67%	✅
whisper-large-encoder	1	10.17	10.21	-0.34%	✅
whisper-large-decoder	1	106.63	103.92	2.61%	✅
mistral-7b	1	22.51	23.89	-5.80%	🔴
FLUX.1-schnell	1	762.32	757.77	0.60%	✅

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	✅	PASSED: MIGraphX meets tolerance
whisper-large-decoder	✅	PASSED: MIGraphX meets tolerance
mistral-7b	✅	PASSED: MIGraphX meets tolerance
FLUX.1-schnell	✅	PASSED: MIGraphX meets tolerance

[Copilot]

Pull request overview

Fixes the reference implementation of the nonzero operator to correctly handle non-standard input layouts (e.g., transposed and broadcasted tensors), aligning compute-time element access with logical indexing.

Changes:

• Removed the standard-layout constraint from nonzero shape validation and updated compute-time access to use logical indices from shape_for_each.
• Added new reference + verify tests covering transposed inputs and reference coverage for broadcasted inputs.
• Added a changelog entry under “Resolved issues”.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
src/include/migraphx/op/nonzero.hpp	Removes standard-layout requirement and updates indexing logic to support non-standard layouts.
test/ref/nonzero.cpp	Adds reference tests for transposed and broadcasted inputs.
test/verify/test_nonzero.cpp	Adds verify coverage for transposed inputs.
CHANGELOG.md	Documents the fix under resolved issues.

[TedThemistokleous]

Just need clarification on your test data. I think you have extra values that aren't checked / dont make sense.

Possible to also clean up the output assignment with a transform using migraphx ranges for front().size instead of keeping the raw loop.