[llvm] [AMDGPU] Support byte_sel modifier on v_cvt_sr_fp8_f32 and v_cvt_sr_bf8_f32 (PR #90244)

[Stanislav Mekhanoshin via llvm-commits]

================
@@ -97,9 +97,7 @@ define amdgpu_cs void @test_cvt_sr_bf8_f32_byte0(i32 %a, i32 %r, i32 %old, ptr a
 define amdgpu_cs void @test_cvt_sr_fp8_f32_byte1(i32 %a, i32 %r, i32 %old, ptr addrspace(1) %out) {
 ; GFX12-LABEL: test_cvt_sr_fp8_f32_byte1:
 ; GFX12:       ; %bb.0:
-; GFX12-NEXT:    v_mov_b32_dpp v0, v0 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf bound_ctrl:1
-; GFX12-NEXT:    s_delay_alu instid0(VALU_DEP_1)
-; GFX12-NEXT:    v_cvt_sr_fp8_f32 v2, v0, v1 op_sel:[0,0,1,0]
+; GFX12-NEXT:    v_cvt_sr_fp8_f32_e64_dpp v2, v0, v1 byte_sel:1 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf bound_ctrl:1
----------------
rampitec wrote:

Yes, there is no problem to use it with DPP.

https://github.com/llvm/llvm-project/pull/90244