Add a new JIT compiler for CPU code (#693)
* Start of the ARMeilleure project
* Refactoring around the old IRAdapter, now renamed to PreAllocator
* Optimize the LowestBitSet method
* Add CLZ support and fix CLS implementation
* Add missing Equals and GetHashCode overrides on some structs, misc small tweaks
* Implement the ByteSwap IR instruction, and some refactoring on the assembler
* Implement the DivideUI IR instruction and fix 64-bits IDIV
* Correct constant operand type on CSINC
* Move division instructions implementation to InstEmitDiv
* Fix destination type for the ConditionalSelect IR instruction
* Implement UMULH and SMULH, with new IR instructions
* Fix some issues with shift instructions
* Fix constant types for BFM instructions
* Fix up new tests using the new V128 struct
* Update tests
* Move DIV tests to a separate file
* Add support for calls, and some instructions that depends on them
* Start adding support for SIMD & FP types, along with some of the related ARM instructions
* Fix some typos and the divide instruction with FP operands
* Fix wrong method call on Clz_V
* Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes
* Implement SIMD logical instructions and more misc. fixes
* Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations
* Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes
* Implement SIMD shift instruction and fix Dup_V
* Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table
* Fix check with tolerance on tester
* Implement FP & SIMD comparison instructions, and some fixes
* Update FCVT (Scalar) encoding on the table to support the Half-float variants
* Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes
* Use old memory access methods, made a start on SIMD memory insts support, some fixes
* Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes
* Fix arguments count with struct return values, other fixes
* More instructions
* Misc. fixes and integrate LDj3SNuD fixes
* Update tests
* Add a faster linear scan allocator, unwinding support on windows, and other changes
* Update Ryujinx.HLE
* Update Ryujinx.Graphics
* Fix V128 return pointer passing, RCX is clobbered
* Update Ryujinx.Tests
* Update ITimeZoneService
* Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks
* Use generic GetFunctionPointerForDelegate method and other tweaks
* Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics
* Remove some unused code on the assembler
* Fix REX.W prefix regression on float conversion instructions, add some sort of profiler
* Add hardware capability detection
* Fix regression on Sha1h and revert Fcm** changes
* Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator
* Fix silly mistake introduced on last commit on CpuId
* Generate inline stack probes when the stack allocation is too large
* Initial support for the System-V ABI
* Support multiple destination operands
* Fix SSE2 VectorInsert8 path, and other fixes
* Change placement of XMM callee save and restore code to match other compilers
* Rename Dest to Destination and Inst to Instruction
* Fix a regression related to calls and the V128 type
* Add an extra space on comments to match code style
* Some refactoring
* Fix vector insert FP32 SSE2 path
* Port over the ARM32 instructions
* Avoid memory protection races on JIT Cache
* Another fix on VectorInsert FP32 (thanks to LDj3SNuD
* Float operands don't need to use the same register when VEX is supported
* Add a new register allocator, higher quality code for hot code (tier up), and other tweaks
* Some nits, small improvements on the pre allocator
* CpuThreadState is gone
* Allow changing CPU emulators with a config entry
* Add runtime identifiers on the ARMeilleure project
* Allow switching between CPUs through a config entry (pt. 2)
* Change win10-x64 to win-x64 on projects
* Update the Ryujinx project to use ARMeilleure
* Ensure that the selected register is valid on the hybrid allocator
* Allow exiting on returns to 0 (should fix test regression)
* Remove register assignments for most used variables on the hybrid allocator
* Do not use fixed registers as spill temp
* Add missing namespace and remove unneeded using
* Address PR feedback
* Fix types, etc
* Enable AssumeStrictAbiCompliance by default
* Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 14:56:22 -04:00
|
|
|
using ARMeilleure.Decoders;
|
|
|
|
using ARMeilleure.IntermediateRepresentation;
|
|
|
|
using ARMeilleure.Translation;
|
2019-12-29 20:22:47 -05:00
|
|
|
using System.Diagnostics;
|
Add a new JIT compiler for CPU code (#693)
* Start of the ARMeilleure project
* Refactoring around the old IRAdapter, now renamed to PreAllocator
* Optimize the LowestBitSet method
* Add CLZ support and fix CLS implementation
* Add missing Equals and GetHashCode overrides on some structs, misc small tweaks
* Implement the ByteSwap IR instruction, and some refactoring on the assembler
* Implement the DivideUI IR instruction and fix 64-bits IDIV
* Correct constant operand type on CSINC
* Move division instructions implementation to InstEmitDiv
* Fix destination type for the ConditionalSelect IR instruction
* Implement UMULH and SMULH, with new IR instructions
* Fix some issues with shift instructions
* Fix constant types for BFM instructions
* Fix up new tests using the new V128 struct
* Update tests
* Move DIV tests to a separate file
* Add support for calls, and some instructions that depends on them
* Start adding support for SIMD & FP types, along with some of the related ARM instructions
* Fix some typos and the divide instruction with FP operands
* Fix wrong method call on Clz_V
* Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes
* Implement SIMD logical instructions and more misc. fixes
* Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations
* Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes
* Implement SIMD shift instruction and fix Dup_V
* Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table
* Fix check with tolerance on tester
* Implement FP & SIMD comparison instructions, and some fixes
* Update FCVT (Scalar) encoding on the table to support the Half-float variants
* Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes
* Use old memory access methods, made a start on SIMD memory insts support, some fixes
* Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes
* Fix arguments count with struct return values, other fixes
* More instructions
* Misc. fixes and integrate LDj3SNuD fixes
* Update tests
* Add a faster linear scan allocator, unwinding support on windows, and other changes
* Update Ryujinx.HLE
* Update Ryujinx.Graphics
* Fix V128 return pointer passing, RCX is clobbered
* Update Ryujinx.Tests
* Update ITimeZoneService
* Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks
* Use generic GetFunctionPointerForDelegate method and other tweaks
* Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics
* Remove some unused code on the assembler
* Fix REX.W prefix regression on float conversion instructions, add some sort of profiler
* Add hardware capability detection
* Fix regression on Sha1h and revert Fcm** changes
* Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator
* Fix silly mistake introduced on last commit on CpuId
* Generate inline stack probes when the stack allocation is too large
* Initial support for the System-V ABI
* Support multiple destination operands
* Fix SSE2 VectorInsert8 path, and other fixes
* Change placement of XMM callee save and restore code to match other compilers
* Rename Dest to Destination and Inst to Instruction
* Fix a regression related to calls and the V128 type
* Add an extra space on comments to match code style
* Some refactoring
* Fix vector insert FP32 SSE2 path
* Port over the ARM32 instructions
* Avoid memory protection races on JIT Cache
* Another fix on VectorInsert FP32 (thanks to LDj3SNuD
* Float operands don't need to use the same register when VEX is supported
* Add a new register allocator, higher quality code for hot code (tier up), and other tweaks
* Some nits, small improvements on the pre allocator
* CpuThreadState is gone
* Allow changing CPU emulators with a config entry
* Add runtime identifiers on the ARMeilleure project
* Allow switching between CPUs through a config entry (pt. 2)
* Change win10-x64 to win-x64 on projects
* Update the Ryujinx project to use ARMeilleure
* Ensure that the selected register is valid on the hybrid allocator
* Allow exiting on returns to 0 (should fix test regression)
* Remove register assignments for most used variables on the hybrid allocator
* Do not use fixed registers as spill temp
* Add missing namespace and remove unneeded using
* Address PR feedback
* Fix types, etc
* Enable AssumeStrictAbiCompliance by default
* Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 14:56:22 -04:00
|
|
|
|
|
|
|
using static ARMeilleure.Instructions.InstEmitHelper;
|
|
|
|
using static ARMeilleure.Instructions.InstEmitSimdHelper;
|
|
|
|
using static ARMeilleure.IntermediateRepresentation.OperandHelper;
|
|
|
|
|
|
|
|
namespace ARMeilleure.Instructions
|
|
|
|
{
|
|
|
|
static partial class InstEmit
|
|
|
|
{
|
|
|
|
public static void And_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
if (Optimizations.UseSse2)
|
|
|
|
{
|
|
|
|
OpCodeSimdReg op = (OpCodeSimdReg)context.CurrOp;
|
|
|
|
|
|
|
|
Operand n = GetVec(op.Rn);
|
|
|
|
Operand m = GetVec(op.Rm);
|
|
|
|
|
|
|
|
Operand res = context.AddIntrinsic(Intrinsic.X86Pand, n, m);
|
|
|
|
|
|
|
|
if (op.RegisterSize == RegisterSize.Simd64)
|
|
|
|
{
|
|
|
|
res = context.VectorZeroUpper64(res);
|
|
|
|
}
|
|
|
|
|
|
|
|
context.Copy(GetVec(op.Rd), res);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
EmitVectorBinaryOpZx(context, (op1, op2) => context.BitwiseAnd(op1, op2));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Bic_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
if (Optimizations.UseSse2)
|
|
|
|
{
|
|
|
|
OpCodeSimdReg op = (OpCodeSimdReg)context.CurrOp;
|
|
|
|
|
|
|
|
Operand n = GetVec(op.Rn);
|
|
|
|
Operand m = GetVec(op.Rm);
|
|
|
|
|
|
|
|
Operand res = context.AddIntrinsic(Intrinsic.X86Pandn, m, n);
|
|
|
|
|
|
|
|
if (op.RegisterSize == RegisterSize.Simd64)
|
|
|
|
{
|
|
|
|
res = context.VectorZeroUpper64(res);
|
|
|
|
}
|
|
|
|
|
|
|
|
context.Copy(GetVec(op.Rd), res);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
EmitVectorBinaryOpZx(context, (op1, op2) =>
|
|
|
|
{
|
|
|
|
return context.BitwiseAnd(op1, context.BitwiseNot(op2));
|
|
|
|
});
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Bic_Vi(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
EmitVectorImmBinaryOp(context, (op1, op2) =>
|
|
|
|
{
|
|
|
|
return context.BitwiseAnd(op1, context.BitwiseNot(op2));
|
|
|
|
});
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Bif_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
EmitBifBit(context, notRm: true);
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Bit_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
EmitBifBit(context, notRm: false);
|
|
|
|
}
|
|
|
|
|
|
|
|
private static void EmitBifBit(ArmEmitterContext context, bool notRm)
|
|
|
|
{
|
|
|
|
OpCodeSimdReg op = (OpCodeSimdReg)context.CurrOp;
|
|
|
|
|
|
|
|
if (Optimizations.UseSse2)
|
|
|
|
{
|
|
|
|
Operand d = GetVec(op.Rd);
|
|
|
|
Operand n = GetVec(op.Rn);
|
|
|
|
Operand m = GetVec(op.Rm);
|
|
|
|
|
|
|
|
Operand res = context.AddIntrinsic(Intrinsic.X86Pxor, n, d);
|
|
|
|
|
|
|
|
if (notRm)
|
|
|
|
{
|
|
|
|
res = context.AddIntrinsic(Intrinsic.X86Pandn, m, res);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
res = context.AddIntrinsic(Intrinsic.X86Pand, m, res);
|
|
|
|
}
|
|
|
|
|
|
|
|
res = context.AddIntrinsic(Intrinsic.X86Pxor, d, res);
|
|
|
|
|
|
|
|
if (op.RegisterSize == RegisterSize.Simd64)
|
|
|
|
{
|
|
|
|
res = context.VectorZeroUpper64(res);
|
|
|
|
}
|
|
|
|
|
2019-12-29 20:22:47 -05:00
|
|
|
context.Copy(d, res);
|
Add a new JIT compiler for CPU code (#693)
* Start of the ARMeilleure project
* Refactoring around the old IRAdapter, now renamed to PreAllocator
* Optimize the LowestBitSet method
* Add CLZ support and fix CLS implementation
* Add missing Equals and GetHashCode overrides on some structs, misc small tweaks
* Implement the ByteSwap IR instruction, and some refactoring on the assembler
* Implement the DivideUI IR instruction and fix 64-bits IDIV
* Correct constant operand type on CSINC
* Move division instructions implementation to InstEmitDiv
* Fix destination type for the ConditionalSelect IR instruction
* Implement UMULH and SMULH, with new IR instructions
* Fix some issues with shift instructions
* Fix constant types for BFM instructions
* Fix up new tests using the new V128 struct
* Update tests
* Move DIV tests to a separate file
* Add support for calls, and some instructions that depends on them
* Start adding support for SIMD & FP types, along with some of the related ARM instructions
* Fix some typos and the divide instruction with FP operands
* Fix wrong method call on Clz_V
* Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes
* Implement SIMD logical instructions and more misc. fixes
* Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations
* Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes
* Implement SIMD shift instruction and fix Dup_V
* Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table
* Fix check with tolerance on tester
* Implement FP & SIMD comparison instructions, and some fixes
* Update FCVT (Scalar) encoding on the table to support the Half-float variants
* Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes
* Use old memory access methods, made a start on SIMD memory insts support, some fixes
* Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes
* Fix arguments count with struct return values, other fixes
* More instructions
* Misc. fixes and integrate LDj3SNuD fixes
* Update tests
* Add a faster linear scan allocator, unwinding support on windows, and other changes
* Update Ryujinx.HLE
* Update Ryujinx.Graphics
* Fix V128 return pointer passing, RCX is clobbered
* Update Ryujinx.Tests
* Update ITimeZoneService
* Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks
* Use generic GetFunctionPointerForDelegate method and other tweaks
* Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics
* Remove some unused code on the assembler
* Fix REX.W prefix regression on float conversion instructions, add some sort of profiler
* Add hardware capability detection
* Fix regression on Sha1h and revert Fcm** changes
* Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator
* Fix silly mistake introduced on last commit on CpuId
* Generate inline stack probes when the stack allocation is too large
* Initial support for the System-V ABI
* Support multiple destination operands
* Fix SSE2 VectorInsert8 path, and other fixes
* Change placement of XMM callee save and restore code to match other compilers
* Rename Dest to Destination and Inst to Instruction
* Fix a regression related to calls and the V128 type
* Add an extra space on comments to match code style
* Some refactoring
* Fix vector insert FP32 SSE2 path
* Port over the ARM32 instructions
* Avoid memory protection races on JIT Cache
* Another fix on VectorInsert FP32 (thanks to LDj3SNuD
* Float operands don't need to use the same register when VEX is supported
* Add a new register allocator, higher quality code for hot code (tier up), and other tweaks
* Some nits, small improvements on the pre allocator
* CpuThreadState is gone
* Allow changing CPU emulators with a config entry
* Add runtime identifiers on the ARMeilleure project
* Allow switching between CPUs through a config entry (pt. 2)
* Change win10-x64 to win-x64 on projects
* Update the Ryujinx project to use ARMeilleure
* Ensure that the selected register is valid on the hybrid allocator
* Allow exiting on returns to 0 (should fix test regression)
* Remove register assignments for most used variables on the hybrid allocator
* Do not use fixed registers as spill temp
* Add missing namespace and remove unneeded using
* Address PR feedback
* Fix types, etc
* Enable AssumeStrictAbiCompliance by default
* Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 14:56:22 -04:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
Operand res = context.VectorZero();
|
|
|
|
|
|
|
|
int elems = op.RegisterSize == RegisterSize.Simd128 ? 2 : 1;
|
|
|
|
|
|
|
|
for (int index = 0; index < elems; index++)
|
|
|
|
{
|
|
|
|
Operand d = EmitVectorExtractZx(context, op.Rd, index, 3);
|
|
|
|
Operand n = EmitVectorExtractZx(context, op.Rn, index, 3);
|
|
|
|
Operand m = EmitVectorExtractZx(context, op.Rm, index, 3);
|
|
|
|
|
|
|
|
if (notRm)
|
|
|
|
{
|
|
|
|
m = context.BitwiseNot(m);
|
|
|
|
}
|
|
|
|
|
|
|
|
Operand e = context.BitwiseExclusiveOr(d, n);
|
|
|
|
|
|
|
|
e = context.BitwiseAnd(e, m);
|
|
|
|
e = context.BitwiseExclusiveOr(e, d);
|
|
|
|
|
|
|
|
res = EmitVectorInsert(context, res, e, index, 3);
|
|
|
|
}
|
|
|
|
|
|
|
|
context.Copy(GetVec(op.Rd), res);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Bsl_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
if (Optimizations.UseSse2)
|
|
|
|
{
|
|
|
|
OpCodeSimdReg op = (OpCodeSimdReg)context.CurrOp;
|
|
|
|
|
|
|
|
Operand d = GetVec(op.Rd);
|
|
|
|
Operand n = GetVec(op.Rn);
|
|
|
|
Operand m = GetVec(op.Rm);
|
|
|
|
|
|
|
|
Operand res = context.AddIntrinsic(Intrinsic.X86Pxor, n, m);
|
|
|
|
|
|
|
|
res = context.AddIntrinsic(Intrinsic.X86Pand, res, d);
|
|
|
|
res = context.AddIntrinsic(Intrinsic.X86Pxor, res, m);
|
|
|
|
|
|
|
|
if (op.RegisterSize == RegisterSize.Simd64)
|
|
|
|
{
|
|
|
|
res = context.VectorZeroUpper64(res);
|
|
|
|
}
|
|
|
|
|
2019-12-29 20:22:47 -05:00
|
|
|
context.Copy(d, res);
|
Add a new JIT compiler for CPU code (#693)
* Start of the ARMeilleure project
* Refactoring around the old IRAdapter, now renamed to PreAllocator
* Optimize the LowestBitSet method
* Add CLZ support and fix CLS implementation
* Add missing Equals and GetHashCode overrides on some structs, misc small tweaks
* Implement the ByteSwap IR instruction, and some refactoring on the assembler
* Implement the DivideUI IR instruction and fix 64-bits IDIV
* Correct constant operand type on CSINC
* Move division instructions implementation to InstEmitDiv
* Fix destination type for the ConditionalSelect IR instruction
* Implement UMULH and SMULH, with new IR instructions
* Fix some issues with shift instructions
* Fix constant types for BFM instructions
* Fix up new tests using the new V128 struct
* Update tests
* Move DIV tests to a separate file
* Add support for calls, and some instructions that depends on them
* Start adding support for SIMD & FP types, along with some of the related ARM instructions
* Fix some typos and the divide instruction with FP operands
* Fix wrong method call on Clz_V
* Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes
* Implement SIMD logical instructions and more misc. fixes
* Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations
* Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes
* Implement SIMD shift instruction and fix Dup_V
* Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table
* Fix check with tolerance on tester
* Implement FP & SIMD comparison instructions, and some fixes
* Update FCVT (Scalar) encoding on the table to support the Half-float variants
* Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes
* Use old memory access methods, made a start on SIMD memory insts support, some fixes
* Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes
* Fix arguments count with struct return values, other fixes
* More instructions
* Misc. fixes and integrate LDj3SNuD fixes
* Update tests
* Add a faster linear scan allocator, unwinding support on windows, and other changes
* Update Ryujinx.HLE
* Update Ryujinx.Graphics
* Fix V128 return pointer passing, RCX is clobbered
* Update Ryujinx.Tests
* Update ITimeZoneService
* Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks
* Use generic GetFunctionPointerForDelegate method and other tweaks
* Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics
* Remove some unused code on the assembler
* Fix REX.W prefix regression on float conversion instructions, add some sort of profiler
* Add hardware capability detection
* Fix regression on Sha1h and revert Fcm** changes
* Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator
* Fix silly mistake introduced on last commit on CpuId
* Generate inline stack probes when the stack allocation is too large
* Initial support for the System-V ABI
* Support multiple destination operands
* Fix SSE2 VectorInsert8 path, and other fixes
* Change placement of XMM callee save and restore code to match other compilers
* Rename Dest to Destination and Inst to Instruction
* Fix a regression related to calls and the V128 type
* Add an extra space on comments to match code style
* Some refactoring
* Fix vector insert FP32 SSE2 path
* Port over the ARM32 instructions
* Avoid memory protection races on JIT Cache
* Another fix on VectorInsert FP32 (thanks to LDj3SNuD
* Float operands don't need to use the same register when VEX is supported
* Add a new register allocator, higher quality code for hot code (tier up), and other tweaks
* Some nits, small improvements on the pre allocator
* CpuThreadState is gone
* Allow changing CPU emulators with a config entry
* Add runtime identifiers on the ARMeilleure project
* Allow switching between CPUs through a config entry (pt. 2)
* Change win10-x64 to win-x64 on projects
* Update the Ryujinx project to use ARMeilleure
* Ensure that the selected register is valid on the hybrid allocator
* Allow exiting on returns to 0 (should fix test regression)
* Remove register assignments for most used variables on the hybrid allocator
* Do not use fixed registers as spill temp
* Add missing namespace and remove unneeded using
* Address PR feedback
* Fix types, etc
* Enable AssumeStrictAbiCompliance by default
* Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 14:56:22 -04:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
EmitVectorTernaryOpZx(context, (op1, op2, op3) =>
|
|
|
|
{
|
|
|
|
return context.BitwiseExclusiveOr(
|
|
|
|
context.BitwiseAnd(op1,
|
|
|
|
context.BitwiseExclusiveOr(op2, op3)), op3);
|
|
|
|
});
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Eor_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
if (Optimizations.UseSse2)
|
|
|
|
{
|
|
|
|
OpCodeSimdReg op = (OpCodeSimdReg)context.CurrOp;
|
|
|
|
|
|
|
|
Operand n = GetVec(op.Rn);
|
|
|
|
Operand m = GetVec(op.Rm);
|
|
|
|
|
|
|
|
Operand res = context.AddIntrinsic(Intrinsic.X86Pxor, n, m);
|
|
|
|
|
|
|
|
if (op.RegisterSize == RegisterSize.Simd64)
|
|
|
|
{
|
|
|
|
res = context.VectorZeroUpper64(res);
|
|
|
|
}
|
|
|
|
|
|
|
|
context.Copy(GetVec(op.Rd), res);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
EmitVectorBinaryOpZx(context, (op1, op2) => context.BitwiseExclusiveOr(op1, op2));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Not_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
if (Optimizations.UseSse2)
|
|
|
|
{
|
|
|
|
OpCodeSimd op = (OpCodeSimd)context.CurrOp;
|
|
|
|
|
|
|
|
Operand n = GetVec(op.Rn);
|
|
|
|
|
|
|
|
Operand mask = X86GetAllElements(context, -1L);
|
|
|
|
|
|
|
|
Operand res = context.AddIntrinsic(Intrinsic.X86Pandn, n, mask);
|
|
|
|
|
|
|
|
if (op.RegisterSize == RegisterSize.Simd64)
|
|
|
|
{
|
|
|
|
res = context.VectorZeroUpper64(res);
|
|
|
|
}
|
|
|
|
|
|
|
|
context.Copy(GetVec(op.Rd), res);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
EmitVectorUnaryOpZx(context, (op1) => context.BitwiseNot(op1));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Orn_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
if (Optimizations.UseSse2)
|
|
|
|
{
|
|
|
|
OpCodeSimdReg op = (OpCodeSimdReg)context.CurrOp;
|
|
|
|
|
|
|
|
Operand n = GetVec(op.Rn);
|
|
|
|
Operand m = GetVec(op.Rm);
|
|
|
|
|
|
|
|
Operand mask = X86GetAllElements(context, -1L);
|
|
|
|
|
|
|
|
Operand res = context.AddIntrinsic(Intrinsic.X86Pandn, m, mask);
|
|
|
|
|
|
|
|
res = context.AddIntrinsic(Intrinsic.X86Por, res, n);
|
|
|
|
|
|
|
|
if (op.RegisterSize == RegisterSize.Simd64)
|
|
|
|
{
|
|
|
|
res = context.VectorZeroUpper64(res);
|
|
|
|
}
|
|
|
|
|
|
|
|
context.Copy(GetVec(op.Rd), res);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
EmitVectorBinaryOpZx(context, (op1, op2) =>
|
|
|
|
{
|
|
|
|
return context.BitwiseOr(op1, context.BitwiseNot(op2));
|
|
|
|
});
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Orr_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
if (Optimizations.UseSse2)
|
|
|
|
{
|
|
|
|
OpCodeSimdReg op = (OpCodeSimdReg)context.CurrOp;
|
|
|
|
|
|
|
|
Operand n = GetVec(op.Rn);
|
|
|
|
Operand m = GetVec(op.Rm);
|
|
|
|
|
|
|
|
Operand res = context.AddIntrinsic(Intrinsic.X86Por, n, m);
|
|
|
|
|
|
|
|
if (op.RegisterSize == RegisterSize.Simd64)
|
|
|
|
{
|
|
|
|
res = context.VectorZeroUpper64(res);
|
|
|
|
}
|
|
|
|
|
|
|
|
context.Copy(GetVec(op.Rd), res);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
EmitVectorBinaryOpZx(context, (op1, op2) => context.BitwiseOr(op1, op2));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Orr_Vi(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
EmitVectorImmBinaryOp(context, (op1, op2) => context.BitwiseOr(op1, op2));
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Rbit_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
OpCodeSimd op = (OpCodeSimd)context.CurrOp;
|
|
|
|
|
|
|
|
Operand res = context.VectorZero();
|
|
|
|
|
|
|
|
int elems = op.RegisterSize == RegisterSize.Simd128 ? 16 : 8;
|
|
|
|
|
|
|
|
for (int index = 0; index < elems; index++)
|
|
|
|
{
|
|
|
|
Operand ne = EmitVectorExtractZx(context, op.Rn, index, 0);
|
|
|
|
|
2019-12-29 20:22:47 -05:00
|
|
|
Operand de = EmitReverseBits8Op(context, ne);
|
Add a new JIT compiler for CPU code (#693)
* Start of the ARMeilleure project
* Refactoring around the old IRAdapter, now renamed to PreAllocator
* Optimize the LowestBitSet method
* Add CLZ support and fix CLS implementation
* Add missing Equals and GetHashCode overrides on some structs, misc small tweaks
* Implement the ByteSwap IR instruction, and some refactoring on the assembler
* Implement the DivideUI IR instruction and fix 64-bits IDIV
* Correct constant operand type on CSINC
* Move division instructions implementation to InstEmitDiv
* Fix destination type for the ConditionalSelect IR instruction
* Implement UMULH and SMULH, with new IR instructions
* Fix some issues with shift instructions
* Fix constant types for BFM instructions
* Fix up new tests using the new V128 struct
* Update tests
* Move DIV tests to a separate file
* Add support for calls, and some instructions that depends on them
* Start adding support for SIMD & FP types, along with some of the related ARM instructions
* Fix some typos and the divide instruction with FP operands
* Fix wrong method call on Clz_V
* Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes
* Implement SIMD logical instructions and more misc. fixes
* Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations
* Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes
* Implement SIMD shift instruction and fix Dup_V
* Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table
* Fix check with tolerance on tester
* Implement FP & SIMD comparison instructions, and some fixes
* Update FCVT (Scalar) encoding on the table to support the Half-float variants
* Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes
* Use old memory access methods, made a start on SIMD memory insts support, some fixes
* Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes
* Fix arguments count with struct return values, other fixes
* More instructions
* Misc. fixes and integrate LDj3SNuD fixes
* Update tests
* Add a faster linear scan allocator, unwinding support on windows, and other changes
* Update Ryujinx.HLE
* Update Ryujinx.Graphics
* Fix V128 return pointer passing, RCX is clobbered
* Update Ryujinx.Tests
* Update ITimeZoneService
* Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks
* Use generic GetFunctionPointerForDelegate method and other tweaks
* Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics
* Remove some unused code on the assembler
* Fix REX.W prefix regression on float conversion instructions, add some sort of profiler
* Add hardware capability detection
* Fix regression on Sha1h and revert Fcm** changes
* Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator
* Fix silly mistake introduced on last commit on CpuId
* Generate inline stack probes when the stack allocation is too large
* Initial support for the System-V ABI
* Support multiple destination operands
* Fix SSE2 VectorInsert8 path, and other fixes
* Change placement of XMM callee save and restore code to match other compilers
* Rename Dest to Destination and Inst to Instruction
* Fix a regression related to calls and the V128 type
* Add an extra space on comments to match code style
* Some refactoring
* Fix vector insert FP32 SSE2 path
* Port over the ARM32 instructions
* Avoid memory protection races on JIT Cache
* Another fix on VectorInsert FP32 (thanks to LDj3SNuD
* Float operands don't need to use the same register when VEX is supported
* Add a new register allocator, higher quality code for hot code (tier up), and other tweaks
* Some nits, small improvements on the pre allocator
* CpuThreadState is gone
* Allow changing CPU emulators with a config entry
* Add runtime identifiers on the ARMeilleure project
* Allow switching between CPUs through a config entry (pt. 2)
* Change win10-x64 to win-x64 on projects
* Update the Ryujinx project to use ARMeilleure
* Ensure that the selected register is valid on the hybrid allocator
* Allow exiting on returns to 0 (should fix test regression)
* Remove register assignments for most used variables on the hybrid allocator
* Do not use fixed registers as spill temp
* Add missing namespace and remove unneeded using
* Address PR feedback
* Fix types, etc
* Enable AssumeStrictAbiCompliance by default
* Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 14:56:22 -04:00
|
|
|
|
|
|
|
res = EmitVectorInsert(context, res, de, index, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
context.Copy(GetVec(op.Rd), res);
|
|
|
|
}
|
|
|
|
|
2019-12-29 20:22:47 -05:00
|
|
|
private static Operand EmitReverseBits8Op(ArmEmitterContext context, Operand op)
|
|
|
|
{
|
|
|
|
Debug.Assert(op.Type == OperandType.I64);
|
|
|
|
|
|
|
|
Operand val = context.BitwiseOr(context.ShiftRightUI(context.BitwiseAnd(op, Const(0xaaul)), Const(1)),
|
|
|
|
context.ShiftLeft (context.BitwiseAnd(op, Const(0x55ul)), Const(1)));
|
|
|
|
|
|
|
|
val = context.BitwiseOr(context.ShiftRightUI(context.BitwiseAnd(val, Const(0xccul)), Const(2)),
|
|
|
|
context.ShiftLeft (context.BitwiseAnd(val, Const(0x33ul)), Const(2)));
|
|
|
|
|
|
|
|
return context.BitwiseOr(context.ShiftRightUI(val, Const(4)),
|
|
|
|
context.ShiftLeft (context.BitwiseAnd(val, Const(0x0ful)), Const(4)));
|
|
|
|
}
|
|
|
|
|
Add a new JIT compiler for CPU code (#693)
* Start of the ARMeilleure project
* Refactoring around the old IRAdapter, now renamed to PreAllocator
* Optimize the LowestBitSet method
* Add CLZ support and fix CLS implementation
* Add missing Equals and GetHashCode overrides on some structs, misc small tweaks
* Implement the ByteSwap IR instruction, and some refactoring on the assembler
* Implement the DivideUI IR instruction and fix 64-bits IDIV
* Correct constant operand type on CSINC
* Move division instructions implementation to InstEmitDiv
* Fix destination type for the ConditionalSelect IR instruction
* Implement UMULH and SMULH, with new IR instructions
* Fix some issues with shift instructions
* Fix constant types for BFM instructions
* Fix up new tests using the new V128 struct
* Update tests
* Move DIV tests to a separate file
* Add support for calls, and some instructions that depends on them
* Start adding support for SIMD & FP types, along with some of the related ARM instructions
* Fix some typos and the divide instruction with FP operands
* Fix wrong method call on Clz_V
* Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes
* Implement SIMD logical instructions and more misc. fixes
* Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations
* Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes
* Implement SIMD shift instruction and fix Dup_V
* Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table
* Fix check with tolerance on tester
* Implement FP & SIMD comparison instructions, and some fixes
* Update FCVT (Scalar) encoding on the table to support the Half-float variants
* Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes
* Use old memory access methods, made a start on SIMD memory insts support, some fixes
* Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes
* Fix arguments count with struct return values, other fixes
* More instructions
* Misc. fixes and integrate LDj3SNuD fixes
* Update tests
* Add a faster linear scan allocator, unwinding support on windows, and other changes
* Update Ryujinx.HLE
* Update Ryujinx.Graphics
* Fix V128 return pointer passing, RCX is clobbered
* Update Ryujinx.Tests
* Update ITimeZoneService
* Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks
* Use generic GetFunctionPointerForDelegate method and other tweaks
* Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics
* Remove some unused code on the assembler
* Fix REX.W prefix regression on float conversion instructions, add some sort of profiler
* Add hardware capability detection
* Fix regression on Sha1h and revert Fcm** changes
* Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator
* Fix silly mistake introduced on last commit on CpuId
* Generate inline stack probes when the stack allocation is too large
* Initial support for the System-V ABI
* Support multiple destination operands
* Fix SSE2 VectorInsert8 path, and other fixes
* Change placement of XMM callee save and restore code to match other compilers
* Rename Dest to Destination and Inst to Instruction
* Fix a regression related to calls and the V128 type
* Add an extra space on comments to match code style
* Some refactoring
* Fix vector insert FP32 SSE2 path
* Port over the ARM32 instructions
* Avoid memory protection races on JIT Cache
* Another fix on VectorInsert FP32 (thanks to LDj3SNuD
* Float operands don't need to use the same register when VEX is supported
* Add a new register allocator, higher quality code for hot code (tier up), and other tweaks
* Some nits, small improvements on the pre allocator
* CpuThreadState is gone
* Allow changing CPU emulators with a config entry
* Add runtime identifiers on the ARMeilleure project
* Allow switching between CPUs through a config entry (pt. 2)
* Change win10-x64 to win-x64 on projects
* Update the Ryujinx project to use ARMeilleure
* Ensure that the selected register is valid on the hybrid allocator
* Allow exiting on returns to 0 (should fix test regression)
* Remove register assignments for most used variables on the hybrid allocator
* Do not use fixed registers as spill temp
* Add missing namespace and remove unneeded using
* Address PR feedback
* Fix types, etc
* Enable AssumeStrictAbiCompliance by default
* Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 14:56:22 -04:00
|
|
|
public static void Rev16_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
if (Optimizations.UseSsse3)
|
|
|
|
{
|
|
|
|
OpCodeSimd op = (OpCodeSimd)context.CurrOp;
|
|
|
|
|
|
|
|
Operand n = GetVec(op.Rn);
|
|
|
|
|
|
|
|
const long maskE0 = 06L << 56 | 07L << 48 | 04L << 40 | 05L << 32 | 02L << 24 | 03L << 16 | 00L << 8 | 01L << 0;
|
|
|
|
const long maskE1 = 14L << 56 | 15L << 48 | 12L << 40 | 13L << 32 | 10L << 24 | 11L << 16 | 08L << 8 | 09L << 0;
|
|
|
|
|
|
|
|
Operand mask = X86GetScalar(context, maskE0);
|
|
|
|
|
|
|
|
mask = EmitVectorInsert(context, mask, Const(maskE1), 1, 3);
|
|
|
|
|
|
|
|
Operand res = context.AddIntrinsic(Intrinsic.X86Pshufb, n, mask);
|
|
|
|
|
|
|
|
if (op.RegisterSize == RegisterSize.Simd64)
|
|
|
|
{
|
|
|
|
res = context.VectorZeroUpper64(res);
|
|
|
|
}
|
|
|
|
|
|
|
|
context.Copy(GetVec(op.Rd), res);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
EmitRev_V(context, containerSize: 1);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Rev32_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
if (Optimizations.UseSsse3)
|
|
|
|
{
|
|
|
|
OpCodeSimd op = (OpCodeSimd)context.CurrOp;
|
|
|
|
|
|
|
|
Operand n = GetVec(op.Rn);
|
|
|
|
|
|
|
|
Operand mask;
|
|
|
|
|
|
|
|
if (op.Size == 0)
|
|
|
|
{
|
|
|
|
const long maskE0 = 04L << 56 | 05L << 48 | 06L << 40 | 07L << 32 | 00L << 24 | 01L << 16 | 02L << 8 | 03L << 0;
|
|
|
|
const long maskE1 = 12L << 56 | 13L << 48 | 14L << 40 | 15L << 32 | 08L << 24 | 09L << 16 | 10L << 8 | 11L << 0;
|
|
|
|
|
|
|
|
mask = X86GetScalar(context, maskE0);
|
|
|
|
|
|
|
|
mask = EmitVectorInsert(context, mask, Const(maskE1), 1, 3);
|
|
|
|
}
|
|
|
|
else /* if (op.Size == 1) */
|
|
|
|
{
|
|
|
|
const long maskE0 = 05L << 56 | 04L << 48 | 07L << 40 | 06L << 32 | 01L << 24 | 00L << 16 | 03L << 8 | 02L << 0;
|
|
|
|
const long maskE1 = 13L << 56 | 12L << 48 | 15L << 40 | 14L << 32 | 09L << 24 | 08L << 16 | 11L << 8 | 10L << 0;
|
|
|
|
|
|
|
|
mask = X86GetScalar(context, maskE0);
|
|
|
|
|
|
|
|
mask = EmitVectorInsert(context, mask, Const(maskE1), 1, 3);
|
|
|
|
}
|
|
|
|
|
|
|
|
Operand res = context.AddIntrinsic(Intrinsic.X86Pshufb, n, mask);
|
|
|
|
|
|
|
|
if (op.RegisterSize == RegisterSize.Simd64)
|
|
|
|
{
|
|
|
|
res = context.VectorZeroUpper64(res);
|
|
|
|
}
|
|
|
|
|
|
|
|
context.Copy(GetVec(op.Rd), res);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
EmitRev_V(context, containerSize: 2);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void Rev64_V(ArmEmitterContext context)
|
|
|
|
{
|
|
|
|
if (Optimizations.UseSsse3)
|
|
|
|
{
|
|
|
|
OpCodeSimd op = (OpCodeSimd)context.CurrOp;
|
|
|
|
|
|
|
|
Operand n = GetVec(op.Rn);
|
|
|
|
|
|
|
|
Operand mask;
|
|
|
|
|
|
|
|
if (op.Size == 0)
|
|
|
|
{
|
|
|
|
const long maskE0 = 00L << 56 | 01L << 48 | 02L << 40 | 03L << 32 | 04L << 24 | 05L << 16 | 06L << 8 | 07L << 0;
|
|
|
|
const long maskE1 = 08L << 56 | 09L << 48 | 10L << 40 | 11L << 32 | 12L << 24 | 13L << 16 | 14L << 8 | 15L << 0;
|
|
|
|
|
|
|
|
mask = X86GetScalar(context, maskE0);
|
|
|
|
|
|
|
|
mask = EmitVectorInsert(context, mask, Const(maskE1), 1, 3);
|
|
|
|
}
|
|
|
|
else if (op.Size == 1)
|
|
|
|
{
|
|
|
|
const long maskE0 = 01L << 56 | 00L << 48 | 03L << 40 | 02L << 32 | 05L << 24 | 04L << 16 | 07L << 8 | 06L << 0;
|
|
|
|
const long maskE1 = 09L << 56 | 08L << 48 | 11L << 40 | 10L << 32 | 13L << 24 | 12L << 16 | 15L << 8 | 14L << 0;
|
|
|
|
|
|
|
|
mask = X86GetScalar(context, maskE0);
|
|
|
|
|
|
|
|
mask = EmitVectorInsert(context, mask, Const(maskE1), 1, 3);
|
|
|
|
}
|
|
|
|
else /* if (op.Size == 2) */
|
|
|
|
{
|
|
|
|
const long maskE0 = 03L << 56 | 02L << 48 | 01L << 40 | 00L << 32 | 07L << 24 | 06L << 16 | 05L << 8 | 04L << 0;
|
|
|
|
const long maskE1 = 11L << 56 | 10L << 48 | 09L << 40 | 08L << 32 | 15L << 24 | 14L << 16 | 13L << 8 | 12L << 0;
|
|
|
|
|
|
|
|
mask = X86GetScalar(context, maskE0);
|
|
|
|
|
|
|
|
mask = EmitVectorInsert(context, mask, Const(maskE1), 1, 3);
|
|
|
|
}
|
|
|
|
|
|
|
|
Operand res = context.AddIntrinsic(Intrinsic.X86Pshufb, n, mask);
|
|
|
|
|
|
|
|
if (op.RegisterSize == RegisterSize.Simd64)
|
|
|
|
{
|
|
|
|
res = context.VectorZeroUpper64(res);
|
|
|
|
}
|
|
|
|
|
|
|
|
context.Copy(GetVec(op.Rd), res);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
EmitRev_V(context, containerSize: 3);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
private static void EmitRev_V(ArmEmitterContext context, int containerSize)
|
|
|
|
{
|
|
|
|
OpCodeSimd op = (OpCodeSimd)context.CurrOp;
|
|
|
|
|
|
|
|
Operand res = context.VectorZero();
|
|
|
|
|
|
|
|
int elems = op.GetBytesCount() >> op.Size;
|
|
|
|
|
|
|
|
int containerMask = (1 << (containerSize - op.Size)) - 1;
|
|
|
|
|
|
|
|
for (int index = 0; index < elems; index++)
|
|
|
|
{
|
|
|
|
int revIndex = index ^ containerMask;
|
|
|
|
|
|
|
|
Operand ne = EmitVectorExtractZx(context, op.Rn, revIndex, op.Size);
|
|
|
|
|
|
|
|
res = EmitVectorInsert(context, res, ne, index, op.Size);
|
|
|
|
}
|
|
|
|
|
|
|
|
context.Copy(GetVec(op.Rd), res);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|