Fix Tier-0 fusion failure on graphs with QLinearConcat - #85
Merged
Conversation
Two bugs in the multi-op-translation path blocked QLinearConcat from fusing, forcing per-op fallback: 1. RebuildTensorDescPointers / RebuildSubNodePointers set op_desc.Desc = desc_storage.get(), assuming the operator desc is the first member of the storage object. The concat's primary dequant and Join sub-node point desc_storage at the shared QLinConcatStorage (first member is a vector), so op_desc.Desc was left pointing at the wrong bytes. Restore the correct pointer at the end of both fixups. 2. Sub-node graph_inputs wiring handled constants and partition-boundary inputs but not partition-internal producers, leaving such inputs with no graph edge. Add a value_producer branch in both Compile and TryCompilePartition. Also switch the concat's dequant/quantize decomposition to the modern DML_OPERATOR_DEQUANTIZE / DML_OPERATOR_QUANTIZE, which take per-tensor scalar scale/zero-point natively via QuantizationTensors.
tperry-amd
approved these changes
Aug 13, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Two bugs in the multi-op-translation path blocked QLinearConcat from fusing, forcing per-op fallback:
RebuildTensorDescPointers / RebuildSubNodePointers set op_desc.Desc = desc_storage.get(), assuming the operator desc is the first member of the storage object. The concat's primary dequant and Join sub-node point desc_storage at the shared QLinConcatStorage (first member is a vector), so op_desc.Desc was left pointing at the wrong bytes. Restore the correct pointer at the end of both fixups.
Sub-node graph_inputs wiring handled constants and partition-boundary inputs but not partition-internal producers, leaving such inputs with no graph edge. Add a value_producer branch in both Compile and TryCompilePartition.
Also switch the concat's dequant/quantize decomposition to the modern DML_OPERATOR_DEQUANTIZE / DML_OPERATOR_QUANTIZE, which take per-tensor scalar scale/zero-point natively via QuantizationTensors.