v_interp_p2_f16 GPU Native ISA AMD Vector

Vector Arithmetic

v_interp_p2_f16

Given a single-precision float J coordinate in a vector register, an attribute specifier and the result of a prior V_INTERP_P1_F32 in another vector…

Encoding

Binary Layout (AMD CDNA 1)
NEG
63:61
OMOD
60:59
SRC2
58:50
SRC1
49:41
SRC0
40:32
110100
31:26
1001110111
25:16
CLAMP
15
OP_SEL
14:11
ABS
10:8
VDST
7:0
 

The ENC_VOP3 layout from AMD's machine-readable ISA specification (AMD CDNA 1). Opcode 631 in OP rebuilds 0x00000000D2770000, an identifier AMD lists for this encoding. Not the same in every generation: AMD RDNA 2 (opcode 858), AMD RDNA 1 (opcode 858).

Format VOP3
Width 64 bits
Opcode 631
Identifier 0x00000000D2770000

Operands

  • VDST
    Written. N/A Data format: N/A (OPR_VGPR, FMT_NUM_F16)
  • SRC1
    Read. N/A Data format: N/A (OPR_SRC_VGPR, FMT_NUM_F32)
  • SRC0
    Read. N/A Data format: N/A (OPR_ATTR, FMT_NUM_F16)
  • SRC2
    Read. N/A Data format: N/A (OPR_SRC_VGPR, FMT_NUM_F32)

In AMD's order, as its machine-readable ISA specification lists them for the ENC_VOP3 encoding (AMD CDNA 1).

GFX Target Compatibility

Per-target GFX compatibility has not yet been verified for this instruction.

Related

More in Vector Arithmetic

Reference

AMDGPU / GFX ISA

Description

Given a single-precision float J coordinate in a vector register, an attribute specifier and the result of a prior V_INTERP_P1_F32 in another vector register, load a half-precision float parameter value from the local data share, compute the second part of parameter interpolation and store the final result as a half-precision float value into a vector register.

Example

v_interp_p2_f16 v5, v2, attr0.w, v3

A real instruction accepted by the LLVM assembler, taken verbatim from LLVM's own AMDGPU MC test suite (gfx8_asm_vop3_e64.s). Not from an AMD document, and not authored here.

Sources