global_atomic_fmax GPU Native ISA AMD Vector

Vector/Global Memory

global_atomic_fmax

Select the maximum of two single-precision float inputs, given two values stored in the data register and a location in the global aperture.

Also written as global_atomic_max_f32, global_atomic_max_num_f32. AMD's machine-readable ISA specification lists these names for the same instruction.

Encoding

Verified bit-level encoding data is not yet available for this instruction. The instruction-format classification below (GLOBAL) is well-documented and stable; exact per-target opcode/field bit positions have not yet been imported from a verified source.
Format GLOBAL

Operands

  • VDST
    Written. Operand must be a vector GPR. Uses an 8-bit operand field. Data format: Value can be anything. (OPR_VGPR, FMT_ANY)
  • ADDR
    Read. Operand must be a vector GPR. Uses an 8-bit operand field. Data format: Value can be anything. (OPR_VGPR, FMT_ANY)
  • DATA
    Read. Operand must be a vector GPR. Uses an 8-bit operand field. Data format: Value can be anything. (OPR_VGPR, FMT_ANY)
  • SADDR
    Read. N/A Data format: Value can be anything. (OPR_SREG, FMT_ANY)

In AMD's order, as its machine-readable ISA specification lists them for the ENC_FLAT_GLBL encoding (AMD RDNA 2).

GFX Target Compatibility

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

Related

More in Vector/Global Memory

Reference

AMDGPU / GFX ISA

Description

Select the maximum of two single-precision float inputs, given two values stored in the data register and a location in the global aperture. Update the global aperture with the selected value. Store the original value from global aperture into a vector register iff the GLC bit is set.

Example

global_atomic_fmax v[1:2], v2, off dlc

A real instruction accepted by the LLVM assembler, taken verbatim from LLVM's own AMDGPU MC test suite (gfx10_asm_flat.s). Not from an AMD document, and not authored here.

Sources