bc37068f90
Copy src1 into dest before the masked vrangeps so that unmasked lanes (normal/inf/NaN values) are preserved via merge-masking. Previously, when the register allocator assigned different physical registers for dest and src1, unmasked lanes would retain stale values from dest. Tested with Intel SDE and AVX-512 cpu to verify. Also shift unused FpRangeSign constants into bits [3:2] of the vrangeps immediate, matching the Intel encoding. Only one used in the code was 0 so it was already correct by accident but using the others in the future would produce incorrect results.
62 lines
2.0 KiB
C++
62 lines
2.0 KiB
C++
/**
|
|
******************************************************************************
|
|
* Xenia : Xbox 360 Emulator Research Project *
|
|
******************************************************************************
|
|
* Copyright 2022 Ben Vanik. All rights reserved. *
|
|
* Released under the BSD license - see LICENSE in the root for more details. *
|
|
******************************************************************************
|
|
*/
|
|
|
|
#ifndef XENIA_CPU_BACKEND_X64_X64_UTIL_H_
|
|
#define XENIA_CPU_BACKEND_X64_X64_UTIL_H_
|
|
|
|
#include "xenia/base/vec128.h"
|
|
#include "xenia/cpu/backend/x64/x64_backend.h"
|
|
#include "xenia/cpu/backend/x64/x64_emitter.h"
|
|
|
|
namespace xe {
|
|
namespace cpu {
|
|
namespace backend {
|
|
namespace x64 {
|
|
|
|
// Used to generate ternary logic truth-tables for vpternlog
|
|
// Use these to directly refer to terms and perform binary operations upon them
|
|
// and the resulting value will be the ternary lookup table
|
|
// ex:
|
|
// (TernaryOperand::a | ~TernaryOperand::b) & TernaryOperand::c
|
|
// = 0b10100010
|
|
// = 0xa2
|
|
// vpternlog a, b, c, 0xa2
|
|
//
|
|
// ~(TernaryOperand::a ^ TernaryOperand::b) & TernaryOperand::c
|
|
// = 0b10000010
|
|
// = 0x82
|
|
// vpternlog a, b, c, 0x82
|
|
namespace TernaryOperand {
|
|
constexpr uint8_t a = 0b11110000;
|
|
constexpr uint8_t b = 0b11001100;
|
|
constexpr uint8_t c = 0b10101010;
|
|
} // namespace TernaryOperand
|
|
|
|
// Opcodes for use with vrange* instructions
|
|
namespace FpRangeSelect {
|
|
constexpr uint8_t Min = 0b00;
|
|
constexpr uint8_t Max = 0b01;
|
|
constexpr uint8_t AbsMin = 0b10; // Smaller absolute value
|
|
constexpr uint8_t AbsMax = 0b11; // Larger absolute value
|
|
}; // namespace FpRangeSelect
|
|
|
|
namespace FpRangeSign {
|
|
constexpr uint8_t OperandA = 0b0000; // Copy sign of operand A
|
|
constexpr uint8_t Preserve = 0b0100; // Leave sign as is
|
|
constexpr uint8_t Positive = 0b1000; // Set Positive
|
|
constexpr uint8_t Negative = 0b1100; // Set Negative
|
|
}; // namespace FpRangeSign
|
|
|
|
} // namespace x64
|
|
} // namespace backend
|
|
} // namespace cpu
|
|
} // namespace xe
|
|
|
|
#endif // XENIA_CPU_BACKEND_X64_X64_UTIL_H_
|