Skip fuse_horizontal pass on dynamic shaped inputs

[CharlieL7]

Motivation

• Hitting an error when compiling dynamic shapes model with fuse_horizontal calling lens() on a dynamic shape.

Technical Details

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.

[Copilot]

Pull request overview

Skips fuse_horizontal gather fusion candidates when any input shape is dynamic, avoiding compilation-time errors from calling lens() on dynamic shapes during horizontal fusion.

Changes:

• Add a dynamic-shape guard in same_table_gather_horizontal_fusion::is_candidate.
• Add the same guard in gather_horizontal_fusion::is_candidate.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #5017      +/-   ##
===========================================
+ Coverage    92.69%   92.72%   +0.03%     
===========================================
  Files          596      596              
  Lines        31603    31739     +136     
===========================================
+ Hits         29292    29427     +135     
- Misses        2311     2312       +1     

Files with missing lines	Coverage Δ
src/fuse_horizontal.cpp	99.28% <100.00%> (+0.03%)	⬆️

... and 1 file 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 (1656cf)	Old Rate (90b175)	Diff	Status
torchvision-resnet50	64	2,976.90	886.14	235.94%	🔆
torchvision-resnet50_fp16	64	5,731.84	1,735.02	230.36%	🔆
torchvision-densenet121	32	1,595.41	700.76	127.67%	🔆
torchvision-densenet121_fp16	32	4,530.26	1,439.55	214.70%	🔆
torchvision-inceptionv3	32	1,778.18	538.77	230.04%	🔆
torchvision-inceptionv3_fp16	32	2,770.97	182.21	1420.72%	🔆
cadene-inceptionv4	16	819.22	205.30	299.03%	🔆
cadene-resnext64x4	16	783.36	179.33	336.82%	🔆
slim-mobilenet	64	4,691.70	1,806.11	159.77%	🔆
slim-nasnetalarge	64	226.18	115.20	96.34%	🔆
slim-resnet50v2	64	2,413.30	1,029.11	134.50%	🔆
bert-mrpc-onnx	8	570.39	137.82	313.87%	🔆
bert-mrpc-tf	1	231.65	17.79	1202.14%	🔆
pytorch-examples-wlang-gru	1	181.32	35.57	409.70%	🔆
pytorch-examples-wlang-lstm	1	250.66	145.48	72.30%	🔆
torchvision-resnet50_1	1	241.07	45.24	432.86%	🔆
cadene-dpn92_1	1	414.17	27.50	1406.06%	🔆
cadene-resnext101_1	1	81.88	28.35	188.84%	🔆
onnx-taau-downsample	1	137.69	103.47	33.06%	🔆
dlrm-criteoterabyte	1	19.66	23.55	-16.51%	🔴
dlrm-criteoterabyte_fp16	1	47.88	28.67	67.01%	🔆
agentmodel	1	12,372.99	640.03	1833.18%	🔆
unet_fp16	2	56.60	12.84	340.89%	🔆
resnet50v1_fp16	1	1,001.77	186.82	436.23%	🔆
resnet50v1_int8	1	938.84	150.02	525.82%	🔆
bert_base_cased_fp16	64	953.68	424.43	124.70%	🔆
bert_large_uncased_fp16	32	346.52	191.64	80.81%	🔆
bert_large_fp16	1	202.97	203.46	-0.24%	✅
distilgpt2_fp16	16	2,090.55	791.60	164.09%	🔆
yolov5s	1	176.68	157.53	12.15%	🔆
tinyllama	1	45.99	28.67	60.43%	🔆
vicuna-fastchat	1	44.13	33.14	33.16%	🔆
whisper-tiny-encoder	1	417.60	418.36	-0.18%	✅
whisper-tiny-decoder	1	413.11	128.92	220.44%	🔆
llama2_7b	1	10.07	11.70	-13.86%	🔴
qwen1.5-7b	1	23.59	15.16	55.57%	🔆
phi3-3.8b	1	25.64	13.69	87.31%	🔆
llama3-8b	1	21.64	15.19	42.47%	🔆
whisper-large-encoder	1	10.28	5.78	77.83%	🔆
whisper-large-decoder	1	34.57	31.17	10.92%	🔆
mistral-7b	1	19.11	21.52	-11.18%	🔴
FLUX.1-schnell	1	767.34	586.06	30.93%	🔆

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

[ahsan-ca]

Looks good. A unit test might be helpful.