Remove -cl-poison-unsupported-fp64-kernels flag

[chunhuanMeng]

The -cl-poison-unsupported-fp64-kernels option is no longer necessary. The compiler now has the capability to detect and filter out all FP64 computation kernels automatically. Consequently, the IGC does not need to be invoked for any FP64 computation kernels.

[Copilot]

Pull Request Overview

This PR removes the now-unnecessary -cl-poison-unsupported-fp64-kernels flag from the offline compiler options, since FP64 kernels are auto-detected and filtered.

• Deleted the obsolete -cl-poison-unsupported-fp64-kernels setting.
• No other behavior changes to compiler flags.

[pytorchxpubot]

@sys_pytorchxpubot triage result for run 15893471765

Triage bot UT analaysis result for reference only, please note unique error message only report once:

1. third_party.torch-xpu-ops.test.xpu.test_linalg_xpu.TestLinalgXPU test_det_xpu_complex128 got failed with error message

 AssertionError: Scalars are not close! 

Triage bot response:

{
  "similar_issue_id": 1214,
  "similar_issue_state": "open",
  "issue_owner": "daisyden",
  "issue_description": "In the test case test_det_xpu_complex128, an AssertionError occurs with the message 'Scalars are not close!', indicating a discrepancy in scalar values during the test. This issue is likely related to tensor operations on XPU, particularly involving complex128 dtype, similar to previous issues where tensor comparisons failed due to precision or computation errors.",
  "root_causes": [
    "Potential issues in tensor computation or comparison logic specific to XPU operations.",
    "Numerical precision discrepancies affecting scalar comparisons in complex128 dtype operations."
  ],
  "suggested_solutions": [
    "Investigate the computation and comparison logic for determinant operations on XPU tensors.",
    "Review and adjust numerical precision handling in tensor operations to ensure accurate scalar comparisons."
  ]
}

2. third_party.torch-xpu-ops.test.xpu.test_linalg_xpu.TestLinalgXPU test_tensorsolve_xpu_complex128 got failed with error message

 AssertionError: Tensor-likes are not close! 

Triage bot response:

{
  "similar_issue_id": 1214,
  "similar_issue_state": "open",
  "issue_owner": "daisyden",
  "issue_description": "The test test_tensorsolve_xpu_complex128 failed with an AssertionError: Tensor-likes are not close! The error suggests a discrepancy in tensor values between XPU and CPU computations, specifically for complex128 tensors. The failure is not random and points to a potential issue in the tensor solve operation's implementation on XPU.",
  "root_causes": [
    "Discrepancies in tensor computations between CPU and XPU, possibly due to differences in numerical precision or kernel implementations.",
    "Potential issues in the XPU-specific implementation of the tensor solve operation."
  ],
  "suggested_solutions": [
    "Review the implementation of the tensor solve operation on XPU to ensure numerical consistency with CPU computations.",
    "Investigate any recent changes or updates to the XPU kernels that might have introduced this discrepancy.",
    "Compare the results of the tensor solve operation on both CPU and XPU to identify and correct any inconsistencies."
  ]
}

3. third_party.torch-xpu-ops.test.xpu.test_ops_xpu.TestCommonXPU test_numpy_ref_linalg_tensorsolve_xpu_complex128 got failed with error message

 AssertionError: Tensor-likes are not close! ; Exception: Caused by reference input at index 0: SampleInput(input=Tensor[size=(2, 3, 6), device="xpu:0", dtype=torch.complex128], args=TensorList[Tensor[size=(2, 3), device="xpu:0", dtype=torch.complex128]], kwargs={'dims': 'None'}, broadcasts_input=False, name='') 

Triage bot response:

{
  "similar_issue_id": 1214,
  "similar_issue_state": "open",
  "issue_owner": "daisyden",
  "issue_description": "The test test_numpy_ref_linalg_tensorsolve_xpu_complex128 is failing with an AssertionError indicating that the tensors from numpy and XPU are not matching within the allowed tolerance. The error occurs during the comparison of tensor operations involving complex128 dtype. The failure is consistent and not random, suggesting a deterministic issue in the implementation.",
  "root_causes": [
    "Discrepancies in the handling of complex128 dtype operations between numpy and XPU implementations.",
    "Potential differences in precision or kernel behavior affecting the results of linalg_tensorsolve operations."
  ],
  "suggested_solutions": [
    "Review and align the implementation of linalg_tensorsolve for complex128 dtype across numpy and XPU to ensure consistent results.",
    "Investigate and correct any precision-related issues in the XPU kernel implementations that handle complex128 tensors."
  ]
}