fix non-standard literals causing failures in mlir modules

[shivadbhavsar]

Motivation

Found some while refactoring torch builders where non-standard literals are causing compile errors

Technical Details

Add similar standardization used in other lowering passes

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.

[Copilot]

Pull request overview

This PR updates the GPU prepare_mlir lowering pass to normalize non-standard-strided literal instructions into standard (contiguous) literals so MLIR compilation doesn’t fail on transposed/strided constants.

Changes:

• Add a matcher in gpu::prepare_mlir to rewrite non-standard, non-broadcast literals into standard literals via a contiguous const-fold.
• Add a new GPU unit test covering normalization behavior and “unchanged” cases for already-standard and broadcast literals.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
src/targets/gpu/prepare_mlir.cpp	Adds a matcher to normalize non-standard literals prior to MLIR lowering.
test/gpu/prepare_mlir.cpp	Adds GPU tests validating literal normalization/unchanged behavior.

[pfultz2]

Can you add a test to test/mlir.cpp as well?

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

Test	Batch	New Rate (b7d572)	Old Rate (1a5bb6)	Diff	Status
torchvision-resnet50	64	3,265.02	3,265.81	-0.02%	✅
torchvision-resnet50_fp16	64	7,547.45	7,548.56	-0.01%	✅
torchvision-densenet121	32	2,483.47	2,481.85	0.07%	✅
torchvision-densenet121_fp16	32	5,002.65	4,975.40	0.55%	✅
torchvision-inceptionv3	32	2,059.41	2,060.05	-0.03%	✅
torchvision-inceptionv3_fp16	32	4,411.64	4,414.40	-0.06%	✅
cadene-inceptionv4	16	820.95	584.00	40.58%	🔆
cadene-resnext64x4	16	615.04	782.84	-21.43%	🔴
slim-mobilenet	64	8,339.29	7,190.62	15.97%	🔆
slim-nasnetalarge	64	226.86	222.84	1.80%	✅
slim-resnet50v2	64	2,290.51	2,017.65	13.52%	🔆
bert-mrpc-onnx	8	1,169.34	1,168.85	0.04%	✅
bert-mrpc-tf	1	487.84	487.56	0.06%	✅
pytorch-examples-wlang-gru	1	282.26	477.33	-40.87%	🔴
pytorch-examples-wlang-lstm	1	388.45	387.83	0.16%	✅
torchvision-resnet50_1	1	584.53	1,038.23	-43.70%	🔴
cadene-dpn92_1	1	444.93	449.51	-1.02%	✅
cadene-resnext101_1	1	366.13	365.36	0.21%	✅
onnx-taau-downsample	1	455.44	844.73	-46.08%	🔴
dlrm-criteoterabyte	1	32.41	17.44	85.83%	🔆
dlrm-criteoterabyte_fp16	1	33.69	31.21	7.95%	🔆
agentmodel	1	6,438.98	5,880.42	9.50%	🔆
unet_fp16	2	15.29	52.15	-70.68%	🔴
resnet50v1_fp16	1	1,451.72	1,385.99	4.74%	✅
resnet50v1_int8	1	1,886.40	1,657.40	13.82%	🔆
bert_base_cased_fp16	64	1,084.74	1,022.83	6.05%	🔆
bert_large_uncased_fp16	32	124.28	339.30	-63.37%	🔴
bert_large_fp16	1	32.09	205.64	-84.40%	🔴
distilgpt2_fp16	16	438.66	2,086.12	-78.97%	🔴
yolov5s	1	76.09	560.85	-86.43%	🔴
tinyllama	1	19.71	45.80	-56.97%	🔴
vicuna-fastchat	1	44.22	44.03	0.45%	✅
whisper-tiny-encoder	1	412.86	328.75	25.58%	🔆
whisper-tiny-decoder	1	408.06	214.20	90.50%	🔆
llama2_7b	1	2.56	14.70	-82.59%	🔴
qwen1.5-7b	1	3.11	23.49	-86.74%	🔴
phi3-3.8b	1	26.66	26.78	-0.41%	✅
llama3-8b	1	21.75	21.70	0.21%	✅
whisper-large-encoder	1	7.75	8.77	-11.60%	🔴
whisper-large-decoder	1	15.49	109.89	-85.90%	🔴
mistral-7b	1	12.69	23.65	-46.36%	🔴
FLUX.1-schnell	1	750.99	172.94	334.26%	🔆

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

[CharlieL7]

LGTM