add runtime symbol resolution op

[shivadbhavsar]

Motivation

For dynamic shapes, the runtime needs to be able to evaluate symbols using concrete input dimensions to be able to compute dynamic ops (eg. slice)

Technical Details

Attributes (std::vector<sym::expr>, reflected directly):

• expressions — symbolic expressions to evaluate.
  Inputs: one tensor whose original compile-time shape contains the referenced root symbols as direct dimensions.
  At runtime, each direct symbolic dimension is bound to the corresponding concrete input lens:

compile-time: {N, 3, H, W}
runtime:      {7, 3, 10, 12}
bindings:     N=7, H=10, W=12

Output: one packed int64 vector of shape {expressions.size()}. Element i is eval(expressions[i]), unclamped.
Implementation:

• compute_shape requires one input, verifies every referenced root symbol appears as a direct input dimension, and returns int64_type, {expressions.size()}.
• compute reads the original input shape from dyn_output::input_shapes and the concrete dimensions from the runtime argument shape.
• It maps each direct root symbol to its runtime lens, verifies repeated symbols resolve consistently, then evaluates every expression with eval_uint.
  Restriction: compound input dimensions are not inverted. For example, an input dimension N/2 cannot be used to recover N; N must appear directly on an input axis.
  Example: expressions=[N, H/2, W/2], compile-time input {N,3,H,W}, runtime input {7,3,10,12} → [7,5,6].

Usage

x = @param:x -> float_type, {N[1..16], 3, H[1..32], W[1..32]}, {3*H*W, H*W, W, 1}
@1 = eval_expr[expressions={N, H/2, W/2}](x) -> int64_type, {3}, {1}
@2 = slice[axes={0, 2, 3},starts={0, 0, 0}](x,@1) -> dynamic, float_type, {[0..16], 3, [0..32], [0..32]}
@3 = @return(@2)

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 adds a new resolve_sym_expr runtime operator to evaluate symbolic dimension expressions from concrete runtime dimension values, enabling dynamic-shape lowering patterns (e.g., feeding symbolic slice bounds as runtime inputs).

Changes:

• Added op::resolve_sym_expr (reflected attributes exprs/symbols) that returns a tuple of evaluated int64 scalars.
• Added reference and shape tests for resolve_sym_expr, plus an integration-style ref test demonstrating use with dynamic multi-input slice.
• Registered the new operator in the build system.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
src/include/migraphx/op/resolve_sym_expr.hpp	New operator implementation for runtime symbol expression evaluation.
src/CMakeLists.txt	Registers resolve_sym_expr in register_migraphx_ops().
test/ref/resolve_sym_expr.cpp	New ref tests covering single/multi-symbol evaluation behavior.
test/op_shape_test.cpp	New shape tests (valid shape + bad input arity) for the operator.
test/ref/slice.cpp	New ref test demonstrating resolve_sym_expr feeding a runtime slice bound.

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

Test	Batch	New Rate (4c03db)	Old Rate (5c50d0)*	Diff	Status
torchvision-resnet50	64	3,264.14	3,278.37	-0.43%	✅
torchvision-resnet50_fp16	64	7,544.73	7,611.38	-0.88%	✅
torchvision-densenet121	32	2,482.20	2,501.41	-0.77%	✅
torchvision-densenet121_fp16	32	4,997.39	5,004.64	-0.14%	✅
torchvision-inceptionv3	32	2,058.30	2,059.27	-0.05%	✅
torchvision-inceptionv3_fp16	32	4,415.10	4,417.65	-0.06%	✅
cadene-inceptionv4	16	821.29	819.80	0.18%	✅
cadene-resnext64x4	16	782.34	783.52	-0.15%	✅
slim-mobilenet	64	8,384.84	8,394.63	-0.12%	✅
slim-nasnetalarge	64	228.86	197.60	15.82%	🔆
slim-resnet50v2	64	3,181.56	3,185.87	-0.14%	✅
bert-mrpc-onnx	8	1,168.15	1,161.65	0.56%	✅
bert-mrpc-tf	1	488.68	486.30	0.49%	✅
pytorch-examples-wlang-gru	1	497.24	469.65	5.87%	🔆
pytorch-examples-wlang-lstm	1	405.68	379.72	6.84%	🔆
torchvision-resnet50_1	1	1,027.20	1,023.59	0.35%	✅
cadene-dpn92_1	1	445.07	436.69	1.92%	✅
cadene-resnext101_1	1	363.01	364.25	-0.34%	✅
onnx-taau-downsample	1	846.67	845.97	0.08%	✅
dlrm-criteoterabyte	1	32.38	32.54	-0.50%	✅
dlrm-criteoterabyte_fp16	1	51.80	52.59	-1.50%	✅
agentmodel	1	8,993.37	9,452.06	-4.85%	✅
unet_fp16	2	58.81	59.01	-0.34%	✅
resnet50v1_fp16	1	1,381.64	1,378.39	0.24%	✅
resnet50v1_int8	1	1,744.74	1,626.43	7.27%	🔆
bert_base_cased_fp16	64	1,097.54	1,102.97	-0.49%	✅
bert_large_uncased_fp16	32	345.70	346.63	-0.27%	✅
bert_large_fp16	1	202.31	205.51	-1.56%	✅
distilgpt2_fp16	16	2,091.39	2,098.82	-0.35%	✅
yolov5s	1	560.05	562.00	-0.35%	✅
tinyllama	1	45.79	45.93	-0.29%	✅
vicuna-fastchat	1	44.01	44.06	-0.11%	✅
whisper-tiny-encoder	1	412.61	415.56	-0.71%	✅
whisper-tiny-decoder	1	414.86	410.79	0.99%	✅
llama2_7b	1	20.83	21.01	-0.85%	✅
qwen1.5-7b	1	23.56	23.60	-0.19%	✅
phi3-3.8b	1	26.71	26.80	-0.36%	✅
llama3-8b	1	21.75	21.70	0.22%	✅
whisper-large-encoder	1	10.18	10.22	-0.41%	✅
whisper-large-decoder	1	106.38	104.52	1.78%	✅
mistral-7b	1	23.74	23.74	-0.01%	✅
FLUX.1-schnell	1	765.69	723.54	5.83%	🔆

Check flagged results 🔆

* No develop baseline was found for this PR's branch point; compared against the latest available develop run instead.

[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
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