tcgen05.mma.sp GPU Virtual ISA NVIDIA

tcgen05.mma.sp TensorCore 5th Generation of MMA Instructions

// 1. Floating-point type without block scaling:
tcgen05.mma.sp.cta_group.kind [d-tmem], a-desc, b-desc, [sp-meta-tmem] , idesc,
{ disable-output-lane }, enable-input-d{, scale-input-d};

Instruction tcgen05.mma.sp is an asynchronous instruction which initiates an MxNxK matrix multiply and accumulate operation of the form D = A*B+D where the A matrix is Mx(K/2), the B matrix is KxN, and the D matrix is MxN.

Encoding

PTX is a virtual instruction set. It has no single, stable native binary encoding - the compiler lowers this instruction to different native machine code depending on the selected NVIDIA target architecture (compute capability). This page intentionally shows no bit-diagram; see the target/version requirements below for what governs how this instruction compiles.
PTX ISA Version Introduced PTX ISA 8.6
Minimum Target sm_100a

Syntax Forms

One mnemonic covers many type / state-space / scope / modifier combinations - each row below is an independently valid form.

Syntax Data Types State Space(s) Modifiers Min. Target Description
// 1. Floating-point type without block scaling: tcgen05.mma.sp.cta_group.kind [d-tmem], a-desc, b-desc, [sp-meta-tmem] , idesc, { disable-output-lane }, enable-input-d{, scale-input-d}; sm_100a Instruction tcgen05.mma.sp is an asynchronous instruction which initiates an MxNxK matrix multiply and accumulate operation of the form D = A*B+D where the A matrix is Mx(K/2), the B matrix is KxN, an… (see the official PTX ISA docs for the full description)

Operands

At a Glance

Data Types -

Related

More in TensorCore 5th Generation of MMA Instructions

Reference

NVIDIA PTX ISA

Description

Instruction tcgen05.mma.sp is an asynchronous instruction which initiates an MxNxK matrix multiply and accumulate operation of the form D = A*B+D where the A matrix is Mx(K/2), the B matrix is KxN, and the D matrix is MxN. Sparse Matrices describes the details of the sparsity. The operation of the form D = A*B is issued when the input predicate argument enable-input-d is false. The optional immediate argument scale-input-d can be specified to scale the input matrix D as follows: D = A*B+D * (2 ^ - scale-input-d) The valid range of values for argument scale-input-d is [0, 15]. (see the official PTX ISA docs for the full description)

Examples

tcgen05.mma.sp.cta_group::1.kind::f16      [taddr0],  adesc,  bdesc, [tmem_spmeta0], idesc, p;

tcgen05.mma.sp.cta_group::1.kind::mxf8f6f4.collector::a:fill
                                           [taddr2],  [taddr1],  bdesc, [tmem_spmeta1], idesc,
                                           [tmem_scaleA], [tmem_scaleB], p;
// (truncated - see the official PTX ISA docs for the full example)

Reproduced from NVIDIA's official PTX ISA documentation for technical accuracy.

Sources