Use relaxed memory ordering for Triton atomics on AMDGPU.

test-infra branch ref changes for wheels

use 16 for tile padding

rename MXTensor params (pytorch#3831)

rename

refine the device

Merge remote-tracking branch 'origin/main' into enable_a100

[mx] Add missing parameter in mxfp8 kernel

When triton_to_mxfp8_dim1() was called from
_to_mxfp8_dim1_kernel_wrapper(), the scale_calculation_mode
parameter was not passed, resulting in incorrect default
value.

Signed-off-by: Ula Golowicz <urszula.golowicz@intel.com>

Rename tensorwise utils

Test changes to linux_job_v2.yml via regression_test_rocm.yml

Test changes to linux_job_v2.yml via regression_test_rocm.yml