Skip to content

Latest commit

 

History

History
70 lines (61 loc) · 3.02 KB

aarch64.md

File metadata and controls

70 lines (61 loc) · 3.02 KB

The A64 instruction set defines the following instruction bit patterns as reserved:

Bit Width Meaning Notes
29 3 op0 Must be 0
25 4 Must be 0
21 4 op1 (hi) Must be 1
16 5 op1 (lo)
0 16

AMX instructions fit into this space like so:

Bit Width Meaning Notes
29 3 op0 Must be 0
25 4 Must be 0
21 4 op1 (hi) Must be 1
16 5 op1 (lo) Must be 0
12 4 Must be 1
10 2 Must be 0
5 5 Instruction 0 through 22 used
0 5 Operand Either 5-bit immediate or 5-bit GPR index, depending on instruction

Inline assembly syntax can be used to emit AMX instructions from C code. The easy case is when the operand is a 5-bit immediate:

#define AMX_OP_IMM5(op, imm5) \
    __asm(".word (0x201000 + (%0 << 5) + %1)" : : "i"(op), "i"(imm5) : "memory")

The harder case is the operand being a 5-bit GPR index. The C compiler will choose the particular GPR however it sees fit, which will be one of x0 through x31, but inline assembly syntax receives this choice as a string rather than an integer. However, prepending the string 0 to the choice gives the strings 0x0 through 0x31, which just happen to be parsable as hexadecimal numbers. They can be used as such, with some math to convert from base 16 back to base 10:

#define AMX_OP_GPR(op, gpr) \
    __asm(".word (0x201000 + (%0 << 5) + 0%1 - ((0%1 >> 4) * 6))" : : "i"(op), "r"((uint64_t)(gpr)) : "memory")

Instruction 17 is used for setup and clearing, and is the sole instruction which uses a 5-bit immediate; all other instructions use a 5-bit GPR index. In Apple's Accelerate, instruction 17 is apparently always prefixed by three nops. The same can be done in inline assembly:

#define AMX_NOP_OP_IMM5(op, imm5) \
    __asm("nop\nnop\nnop\n.word (0x201000 + (%0 << 5) + %1)" : : "i"(op), "i"(imm5) : "memory")

With these macros, the full AMX instruction set can be accessed:

#define AMX_LDX(gpr)    AMX_OP_GPR( 0, gpr)
#define AMX_LDY(gpr)    AMX_OP_GPR( 1, gpr)
#define AMX_STX(gpr)    AMX_OP_GPR( 2, gpr)
#define AMX_STY(gpr)    AMX_OP_GPR( 3, gpr)
#define AMX_LDZ(gpr)    AMX_OP_GPR( 4, gpr)
#define AMX_STZ(gpr)    AMX_OP_GPR( 5, gpr)
#define AMX_LDZI(gpr)   AMX_OP_GPR( 6, gpr)
#define AMX_STZI(gpr)   AMX_OP_GPR( 7, gpr)
#define AMX_EXTRX(gpr)  AMX_OP_GPR( 8, gpr)
#define AMX_EXTRY(gpr)  AMX_OP_GPR( 9, gpr)
#define AMX_FMA64(gpr)  AMX_OP_GPR(10, gpr)
#define AMX_FMS64(gpr)  AMX_OP_GPR(11, gpr)
#define AMX_FMA32(gpr)  AMX_OP_GPR(12, gpr)
#define AMX_FMS32(gpr)  AMX_OP_GPR(13, gpr)
#define AMX_MAC16(gpr)  AMX_OP_GPR(14, gpr)
#define AMX_FMA16(gpr)  AMX_OP_GPR(15, gpr)
#define AMX_FMS16(gpr)  AMX_OP_GPR(16, gpr)
#define AMX_SET()       AMX_NOP_OP_IMM5(17, 0)
#define AMX_CLR()       AMX_NOP_OP_IMM5(17, 1)
#define AMX_VECINT(gpr) AMX_OP_GPR(18, gpr)
#define AMX_VECFP(gpr)  AMX_OP_GPR(19, gpr)
#define AMX_MATINT(gpr) AMX_OP_GPR(20, gpr)
#define AMX_MATFP(gpr)  AMX_OP_GPR(21, gpr)
#define AMX_GENLUT(gpr) AMX_OP_GPR(22, gpr)