Remove unused setDeviceIndex in FusionKernelRuntime::prepareInputs

[zasdfgbnm]

No description provided.

[zasdfgbnm]

!test

[github-actions]

Description

• Remove redundant device index setting in fusion kernel runtime

• Eliminates unnecessary setDeviceIndex call in prepareInputs function

• Streamlines input preparation logic for segmented kernels

Changes walkthrough

	Relevant files
Bug fix	fusion_kernel_runtime.cpp Remove redundant device index setting                                        csrc/runtime/fusion_kernel_runtime.cpp Removed line setting device index on group runtime inputs Simplified prepareInputs function by eliminating redundant call Maintains existing cache ID functionality +0/-1

PR Reviewer Guide

Here are some key observations to aid the review process:

🧪 No relevant tests

🔒 No security concerns identified

⚡ Recommended focus areas for review

Missing Context The PR removes a call to setDeviceIndex() without providing explanation of why this change is needed, what problem it solves, or what the expected impact is. This lack of context makes it difficult to assess the correctness and potential side effects of the change. group_runtime_inputs.setCacheId(group_cache_id.value()); Potential Functionality Impact Removing the device index setting could potentially affect device selection and memory allocation behavior. The change should be validated to ensure it doesn't break device-specific operations or cause runtime errors. group_runtime_inputs.setCacheId(group_cache_id.value());

Test failures

• (High, 25) CUDA illegal memory access in TestNvFuserFrontend on A100

  Test Name	A100	Source
  tests.python.test_python_frontend.TestNvFuserFrontend.test_signbit	❌
  tests.python.test_python_frontend.TestNvFuserFrontend.test_slice_api	❌
  tests.python.test_python_frontend.TestNvFuserFrontend.test_slice_error_checks	❌
  tests.python.test_python_frontend.TestNvFuserFrontend.test_squeeze	❌
  tests.python.test_python_frontend.TestNvFuserFrontend.test_static_tensor_sizes	❌
  tests.python.test_python_frontend.TestNvFuserFrontend.test_stride_order_with_explicit_broadcast	❌
  tests.python.test_python_frontend.TestNvFuserFrontend.test_sum_sliced_reshape_to_broadcast	❌
  tests.python.test_python_frontend.TestNvFuserFrontend.test_take_along_axis	❌
  tests.python.test_python_frontend.TestNvFuserFrontend.test_tensor_ndim	❌
  tests.python.test_python_frontend.TestNvFuserFrontend.test_tensor_shape	❌
  ... with 15 more test failures omitted. Check internal logs.

• (High, 1) CUDA illegal memory access in nvFuser test_selected_device on A100

  Test Name	A100	Source
  tests.python.test_python_frontend.TestNvFuserFrontend.test_selected_device	❌

• (Medium, 1) Scalar numerical mismatch in thunder.tests.test_networks (nanoGPT autograd, CUDA, A100)

  Test Name	A100	Source
  thunder.tests.test_networks.test_nanogpt_complete_autograd_nvfuser_cuda_thunder.dtypes.float32	❌