[pragyam32]

Summary

• Reverts fix(compiler): refine incomplete shapes, unused Split outputs, and unit-slice copies #1306 on main.
• Shape rules again run only for unranked results, so a stored dimension is not replaced by a less specific rule result.
• Unused Split outputs and unit-extent copies return to the previous lowering.

Test plan

• hip-infer-loop-body-shapes LIT
• test_split LIT
• test_memref_copy LIT

Cursor assisted with the revert and this pull request.

[github-actions]

Thanks for opening a PR!

This project follows LLVM's incremental-development and AI-tool-use
guidance. See CONTRIBUTING.md
for the project workflow.

Before requesting review, please check that:

1. The change is focused. Substantial work links the relevant issue
   or design discussion.
2. The PR documents relevant test results and updates affected
   documentation.
3. If AI tools provided substantial assistance, the description
   explains what was assisted and how it was validated, and commit
   trailers identify the tool. The contributor has reviewed and
   understands the result.

Reviewers are assigned through
CODEOWNERS where ownership
is configured.

[Copilot]

🟡 Changes recommended

The revert reintroduces compiler failures, incorrect Split boundaries, and unsafe host access to device memory.

7 open issues

• Allow singleton strides in contiguous view copies · New
• Use ceiling-based lengths for static-axis Split · New
• Validate Split outputs before creating IR · New
• Preserve LeakyRelu shape inference for ranked inputs · New
• Infer unknown Upsample dimensions from known shapes · New
• Prevent compiler abort for unranked Concat operands · New
• Restore ConvTranspose handling for unranked results · New

What changed in this PR

This PR reverts #1306 in hip-ep’s compiler, aiming to preserve stored dimensions by restricting shape inference to unranked results. It also restores earlier Split and memory-copy lowering behavior.

Changes:

• Removes ranked-result refinement and the LeakyRelu and ConvTranspose inference additions.
• Reverts omitted Split output handling and extent-one-aware copy detection.
• Removes corresponding regression tests and restores earlier pitched-copy fixtures.

File	Description
test/​lit/​Dialect/​hip-infer-loop-body-shapes.mlir	Removes Concat, ConvTranspose, and ranked Upsample regression cases.
test/​lit/​Conversion/​onnx-to-hip/​test_split.mlir	Removes omitted-output Split coverage.
test/​lit/​Conversion/​hip-to-llvm/​test_memref_copy.mlir	Removes singleton-copy cases and restores batch-one fixtures.
lib/​Dialect/​Transforms/​InferLoopBodyShapesPass.cpp	Restores unranked-only inference and removes added rules.
lib/​Conversion/​OnnxToHip/​ReshapeConversion.cpp	Restores earlier Split sizing and result handling.
lib/​Conversion/​HipToLLVM/​MemoryLowering.cpp	Restores exact-stride checks for flat device copies.

_{Review effort: Balanced}

[pragyam32]

Not needed, fix for failing case present in PR-1336

[github-actions]

L2 Accuracy Results (EP vs CPU)

Model	Combined L2	Total Elems	Skipped NaN/Inf
conv_test_hybrid	1.12256E-07	64	0
GroupQueryAttention_seq256	59.2561	2621440	0
MatMulNBits_o_seq128	259.905	368640	0
QMoE_seq128	24.2583	368640	0

Threshold: 0.01 | Run: 7973 - Commit: 8a563ce

[github-actions]

Binary sizes

Baseline: run 40457802257 - ae1dc4f (dll-sizes)

File	This (MB)	Main (MB)	Delta (MB)
hipgpu.dll	59.31	59.31	0.00
custom_kernels_gfx1170.dll	5.55	5.55	0.00
custom_kernels_gfx11-generic.dll	5.79	5.79	0.00

MorphiZen EP Performance Results

Model	QPS	Session (s)	1st Infer (ms)	CPU%	Mem (MB)
full_model_seq128	8.75	7.41	434	3	1158
GroupQueryAttention_seq128	10936.70	2.56091	1	5	230
matmul_down_seq128	513.10	3.35	126	3	274

EPContext Export Performance

Model	QPS	Session (s)	1st Infer (ms)	CPU%	Mem (MB)
full_model_seq128	8.82	50.07	247	3	15506

EPContext Import Performance

Model	QPS	Session (s)	1st Infer (ms)	CPU%	Mem (MB)
full_model_seq128	8.81	10.07	247	3	15639

OGA Benchmark Results

Model	Warmup	Reps	Prompt Len	Gen Tokens	TTFT (ms)	TPS	Peak Mem (GB)	GPU Mem (GB)
gpt-oss-20b-webgpu-int4-rtn-block-32	1	5	128	128	90.9	74.3	1.25	12.24
Llama-3.1-8B-awq-g128-int4-asym-fp16-onnx-dml	1	5	128	128	119.2	47.6	1.15	5.23

OGA Wheel Smoke (Python benchmark_e2e.py)

Model	TTFT (ms)	TPS
Llama-3.1-8B-awq-g128-int4-asym-fp16-onnx-dml	118	47.9

VLM Benchmark Results

Model	TTFT (ms)	TPS	Output
gemma-4-E4B-it-fp16-int4-kquant-prune-logits	4001.2	7.9	This is a high-quality, outdoor photograph featuring a beautiful, fluffy white dog, likely a West Highland White Terrier or a si
Qwen3.8-27B-fp16-ve-fp16-int4-k_quant-gs128-text-dml-onnx	4895.7	14.0	The user wants a detailed description of the image. Let me carefully observe all elements before writing. Subject: A white do

TTFT / TPS vs main

Baseline: run 40457802257 - ae1dc4f | Gate: TTFT +20% / TPS -20%

Suite	Model	TTFT (ms)	Main	Delta	TPS	Main	Delta
oga	gpt-oss-20b-webgpu-int4-rtn-block-32	90.9	91.5	-0.7%	74.3	74.2	+0.2%
oga	Llama-3.1-8B-awq-g128-int4-asym-fp16-onnx-dml	119.2	116.9	+2.0%	47.6	47.6	+0.1%
vlm	gemma-4-E4B-it-fp16-int4-kquant-prune-logits	4001.2	4141.2	-3.4%	7.9	7.2	+9.5%
vlm	Qwen3.8-27B-fp16-ve-fp16-int4-k_quant-gs128-text-dml-onnx	4895.7	14900.2	-67.1%	14.0	14.0	+0.1%

Run: 7973 - Commit: 8a563ce