shader/video: Partially implement VMNMX

Implements the common usages for VMNMX. Inputs with a different size
than 32 bits are not supported and sign mismatches aren't supported
either.

VMNMX works as follows:
It grabs Ra and Rb and applies a maximum/minimum on them (this is
defined by .MX), having in mind the input sign. This result can then be
saturated. After the intermediate result is calculated, it applies
another operation on it using Rc. These operations are merges,
accumulations or another min/max pass.

This instruction allows to implement with a more flexible approach GCN's
min3 and max3 instructions (for instance).
This commit is contained in:
ReinUsesLisp 2020-03-28 18:49:50 -03:00
parent 08470d261d
commit 76f178ba6e
3 changed files with 116 additions and 0 deletions

View file

@ -302,6 +302,23 @@ enum class VmadShr : u64 {
Shr15 = 2,
};
enum class VmnmxType : u64 {
Bits8,
Bits16,
Bits32,
};
enum class VmnmxOperation : u64 {
Mrg_16H = 0,
Mrg_16L = 1,
Mrg_8B0 = 2,
Mrg_8B2 = 3,
Acc = 4,
Min = 5,
Max = 6,
Nop = 7,
};
enum class XmadMode : u64 {
None = 0,
CLo = 1,
@ -1662,6 +1679,42 @@ union Instruction {
BitField<47, 1, u64> cc;
} vmad;
union {
BitField<54, 1, u64> is_dest_signed;
BitField<48, 1, u64> is_src_a_signed;
BitField<49, 1, u64> is_src_b_signed;
BitField<37, 2, u64> src_format_a;
BitField<29, 2, u64> src_format_b;
BitField<56, 1, u64> mx;
BitField<55, 1, u64> sat;
BitField<36, 2, u64> selector_a;
BitField<28, 2, u64> selector_b;
BitField<50, 1, u64> is_op_b_register;
BitField<51, 3, VmnmxOperation> operation;
VmnmxType SourceFormatA() const {
switch (src_format_a) {
case 0b11:
return VmnmxType::Bits32;
case 0b10:
return VmnmxType::Bits16;
default:
return VmnmxType::Bits8;
}
}
VmnmxType SourceFormatB() const {
switch (src_format_b) {
case 0b11:
return VmnmxType::Bits32;
case 0b10:
return VmnmxType::Bits16;
default:
return VmnmxType::Bits8;
}
}
} vmnmx;
union {
BitField<20, 16, u64> imm20_16;
BitField<35, 1, u64> high_b_rr; // used on RR
@ -1773,6 +1826,7 @@ public:
MEMBAR,
VMAD,
VSETP,
VMNMX,
FFMA_IMM, // Fused Multiply and Add
FFMA_CR,
FFMA_RC,
@ -2078,6 +2132,7 @@ private:
INST("1110111110011---", Id::MEMBAR, Type::Trivial, "MEMBAR"),
INST("01011111--------", Id::VMAD, Type::Video, "VMAD"),
INST("0101000011110---", Id::VSETP, Type::Video, "VSETP"),
INST("0011101---------", Id::VMNMX, Type::Video, "VMNMX"),
INST("0011001-1-------", Id::FFMA_IMM, Type::Ffma, "FFMA_IMM"),
INST("010010011-------", Id::FFMA_CR, Type::Ffma, "FFMA_CR"),
INST("010100011-------", Id::FFMA_RC, Type::Ffma, "FFMA_RC"),