[llvm] [AMDGPU] Set inst_pref_size to maximum (PR #126981)

Stanislav Mekhanoshin via llvm-commits llvm-commits at lists.llvm.org
Wed Feb 19 12:30:58 PST 2025


================
@@ -199,3 +201,28 @@ const MCExpr *SIProgramInfo::getPGMRSrc2(CallingConv::ID CC,
 
   return MCConstantExpr::create(0, Ctx);
 }
+
+uint64_t SIProgramInfo::getFunctionCodeSize(const MachineFunction &MF) {
----------------
rampitec wrote:

Hm... I have the draft, but the output comes out as
```
.amdhsa_inst_pref_size ((((255*((((.Lfunc_end0-large)+127)/128)>255))+((((.Lfunc_end0-large)+127)/128)*((((.Lfunc_end0-large)+127)/128)<=255)))<<4)&4080)>>4
```
I am not sure we really want this.

https://github.com/llvm/llvm-project/pull/126981


More information about the llvm-commits mailing list