Compare commits

...

6 Commits

Author SHA1 Message Date
gdkchan
cb70e7bb30 Fix DrawArrays vertex buffer size (#4141) 2022-12-21 19:08:12 +01:00
riperiperi
c200a7b7c6 ARMeilleure: Hash _data pointer instead of value for Operand (#4156)
I noticed a weirdly high cost for dictionary accesses from MarkLabel etc. Turns out that the hash code was always the same for labels, so the whole point of having a dictionary was missed and it was putting everything in the same bucket. I made it always hash the _data pointer as that's a good source of identifiable and "random" data.
2022-12-21 02:28:18 +01:00
Emmanuel Hansen
6268170a10 fix sw kbd row collision (#4144) 2022-12-19 15:09:36 -03:00
gdkchan
ee0f9b03a4 Eliminate zero-extension moves in more cases on 32-bit games (#4140)
* Eliminate zero-extension moves in more cases on 32-bit games

* PPTC version bump

* Revert X86Optimizer changes
2022-12-19 14:45:58 -03:00
gdkchan
f93c5f006a Revert "ARMeilleure: Add initial support for AVX512(EVEX encoding) (#3663)" (#4145)
This reverts commit 295fbd0542.
2022-12-18 20:21:10 -03:00
Wunk
295fbd0542 ARMeilleure: Add initial support for AVX512(EVEX encoding) (#3663)
* ARMeilleure: Add AVX512{F,VL,DQ,BW} detection

Add `UseAvx512Ortho` and `UseAvx512OrthoFloat` optimization flags as
short-hands for `F+VL` and `F+VL+DQ`.

* ARMeilleure: Add initial support for EVEX instruction encoding

Does not implement rounding, or exception controls.

* ARMeilleure: Add `X86Vpternlogd`

Accelerates the vector-`Not` instruction.

* ARMeilleure: Add check for `OSXSAVE` for AVX{2,512}

* ARMeilleure: Add check for `XCR0` flags

Add XCR0 register checks for AVX and AVX512F, following the guidelines
from section 14.3 and 15.2 from the Intel Architecture Software
Developer's Manual.

* ARMeilleure: Increment InternalVersion

* ARMeilleure: Remove redundant `ReProtect` and `Dispose`, formatting

* ARMeilleure: Move XCR0 procedure to GetXcr0Eax

* ARMeilleure: Add `XCR0` to `FeatureInfo` structure

* ARMeilleure: Utilize `ReadOnlySpan` for Xcr0 assembly

Avoids an additional allocation

* ARMeilleure: Formatting fixes
2022-12-18 16:46:13 -03:00
7 changed files with 29 additions and 19 deletions

View File

@@ -1587,6 +1587,12 @@ namespace ARMeilleure.CodeGen.X86
Debug.Assert(dest.Type.IsInteger() && source.Type.IsInteger());
// We can eliminate the move if source is already 32-bit and the registers are the same.
if (dest.Value == source.Value && source.Type == OperandType.I32)
{
return;
}
context.Assembler.Mov(dest, source, OperandType.I32);
}

View File

@@ -378,14 +378,7 @@ namespace ARMeilleure.IntermediateRepresentation
public override int GetHashCode()
{
if (Kind == OperandKind.LocalVariable)
{
return base.GetHashCode();
}
else
{
return (int)Value ^ ((int)Kind << 16) ^ ((int)Type << 20);
}
return ((ulong)_data).GetHashCode();
}
public bool Equals(Operand operand)

View File

@@ -27,7 +27,7 @@ namespace ARMeilleure.Translation.PTC
private const string OuterHeaderMagicString = "PTCohd\0\0";
private const string InnerHeaderMagicString = "PTCihd\0\0";
private const uint InternalVersion = 3713; //! To be incremented manually for each change to the ARMeilleure project.
private const uint InternalVersion = 4140; //! To be incremented manually for each change to the ARMeilleure project.
private const string ActualDir = "0";
private const string BackupDir = "1";

View File

@@ -45,7 +45,7 @@
TextWrapping="Wrap" />
<TextBox
Name="Input"
Grid.Row="2"
Grid.Row="3"
Grid.Column="1"
HorizontalAlignment="Stretch"
VerticalAlignment="Center"

View File

@@ -142,6 +142,8 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
_drawState.FirstIndex = firstIndex;
_drawState.IndexCount = indexCount;
_drawState.DrawFirstVertex = drawFirstVertex;
_drawState.DrawVertexCount = drawVertexCount;
_currentSpecState.SetHasConstantBufferDrawParameters(false);
engine.UpdateState();
@@ -163,10 +165,8 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
_instancedIndexCount = ibCount != 0 ? ibCount : indexCount;
var drawState = _state.State.VertexBufferDrawState;
_instancedDrawStateFirst = drawState.First;
_instancedDrawStateCount = drawState.Count;
_instancedDrawStateFirst = drawFirstVertex;
_instancedDrawStateCount = drawVertexCount;
_drawState.DrawIndexed = false;
@@ -415,6 +415,7 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
bool oldDrawIndexed = _drawState.DrawIndexed;
_drawState.DrawIndexed = false;
engine.ForceStateDirty(VertexBufferFirstMethodOffset * 4);
DrawEnd(engine, 0, 0, firstVertex, vertexCount);
@@ -526,8 +527,8 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
}
else
{
_state.State.VertexBufferDrawState.First = firstVertex;
_state.State.VertexBufferDrawState.Count = count;
_drawState.DrawFirstVertex = firstVertex;
_drawState.DrawVertexCount = count;
engine.ForceStateDirty(VertexBufferFirstMethodOffset * 4);
}

View File

@@ -17,6 +17,16 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
/// </summary>
public int IndexCount;
/// <summary>
/// First vertex used on non-indexed draws. This value is stored somewhere else on indexed draws.
/// </summary>
public int DrawFirstVertex;
/// <summary>
/// Vertex count used on non-indexed draws. Indexed draws have a index count instead.
/// </summary>
public int DrawVertexCount;
/// <summary>
/// Indicates if the next draw will be a indexed draw.
/// </summary>

View File

@@ -989,6 +989,8 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
bool drawIndexed = _drawState.DrawIndexed;
bool drawIndirect = _drawState.DrawIndirect;
int drawFirstVertex = _drawState.DrawFirstVertex;
int drawVertexCount = _drawState.DrawVertexCount;
uint vbEnableMask = 0;
for (int index = 0; index < Constants.TotalVertexBuffers; index++)
@@ -1050,9 +1052,7 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
int firstInstance = (int)_state.State.FirstInstance;
var drawState = _state.State.VertexBufferDrawState;
size = Math.Min(vbSize, (ulong)((firstInstance + drawState.First + drawState.Count) * stride));
size = Math.Min(vbSize, (ulong)((firstInstance + drawFirstVertex + drawVertexCount) * stride));
}
_pipeline.VertexBuffers[index] = new BufferPipelineDescriptor(_channel.MemoryManager.IsMapped(address), stride, divisor);