intel / intel/torch-xpu-ops

[XPU] aten::_dyn_quant_pack_4bit_weight not implemented, causing dyn_quant tests to skip

Open
#5,332 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
113
Forks
128
Avg merge
5d 9h
Merged PRs (30d)
112

Description

Dynamic quantization tests that use 4-bit weight packing are skipped on XPU due to missing kernel implementation. The aten::_dyn_quant_pack_4bit_weight operator is not registered for the XPU device, causing all dependent tests to skip.

Affected Tests (3 total):

test__dyn_quant_pack_4bit_weight[xpu]
test__dyn_quant_matmul_4bit[xpu]
test_compile_dyn_quant_matmul_4bit[xpu]
Root Cause:
The aten::_dyn_quant_pack_4bit_weight operator used for dynamic quantization in PyTorch does not have a kernel implementation for XPU devices. This is a PyTorch quantization utility operator, not a standard matrix operation.

python -m pytest test/test_linalg.py::TestLinalgDeviceXPU::test__dyn_quant_pack_4bit_weight -xvs

branch: gplutop7:gplutopx/xpu-enable-test-cpp-extensions-aot
PR: https://github.com/pytorch/pytorch/pull/196597

### Versions

PyTorch version: 2.15.0a0+git32b89e1
Is debug build: False
CUDA used to build PyTorch: None
ROCm SDK used to build PyTorch: N/A
HIP used to build PyTorch: N/A

OS: Ubuntu 24.04.4 LTS (x86_64)
GCC version: (Ubuntu 13.3.0-6ubuntu2~24.04.1) 13.3.0
Clang version: Could not collect
CMake version: version 3.31.6
Libc version: glibc-2.39

Python version: 3.12.3 (main, Jul 15 2026, 23:46:41) [GCC 13.3.0] (64-bit runtime)
Python platform: Linux-6.17.0-1010-intel-x86_64-with-glibc2.39

Contributor guide

Open the contributing guide

Research direction

Start by running test/test_linalg.py::TestLinalgDeviceXPU::test__dyn_quant_pack_4bit_weight -xvs and trace registration of aten::_dyn_quant_pack_4bit_weight for XPU. Review the linked PR for the work already underway. Done means the three listed XPU dynamic-quantization tests run instead of skipping and pass.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.