Ryujinx/ChocolArm64/Translation/ILOpCodeLoadState.cs
gdkchan e21ebbf666 Misc. CPU optimizations (#575)
* Add optimizations related to caller/callee saved registers, thread synchronization and disable tier 0

* Refactoring

* Add a config entry to enable or disable the reg load/store opt.

* Remove unnecessary register state stores for calls when the callee is know

* Rename IoType to VarType

* Enable tier 0 while fixing some perf issues related to tier 0

* Small tweak -- Compile before adding to the cache, to avoid lags

* Add required config entry
2019-02-28 13:03:31 +11:00

51 lines
1.6 KiB
C#

using ChocolArm64.State;
using System.Reflection.Emit;
namespace ChocolArm64.Translation
{
struct ILOpCodeLoadState : IILEmit
{
private ILBlock _block;
private bool _isSubEntry;
public ILOpCodeLoadState(ILBlock block, bool isSubEntry = false)
{
_block = block;
_isSubEntry = isSubEntry;
}
public void Emit(ILMethodBuilder context)
{
long intInputs = context.RegUsage.GetIntInputs(_block);
long vecInputs = context.RegUsage.GetVecInputs(_block);
if (Optimizations.AssumeStrictAbiCompliance && context.IsSubComplete)
{
intInputs = RegisterUsage.ClearCallerSavedIntRegs(intInputs, context.IsAarch64);
vecInputs = RegisterUsage.ClearCallerSavedVecRegs(vecInputs, context.IsAarch64);
}
LoadLocals(context, intInputs, RegisterType.Int);
LoadLocals(context, vecInputs, RegisterType.Vector);
}
private void LoadLocals(ILMethodBuilder context, long inputs, RegisterType baseType)
{
for (int bit = 0; bit < 64; bit++)
{
long mask = 1L << bit;
if ((inputs & mask) != 0)
{
Register reg = ILMethodBuilder.GetRegFromBit(bit, baseType);
context.Generator.EmitLdarg(TranslatedSub.StateArgIdx);
context.Generator.Emit(OpCodes.Ldfld, reg.GetField());
context.Generator.EmitStloc(context.GetLocalIndex(reg));
}
}
}
}
}