For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Mojo struct

FourWaveOps

struct FourWaveOps

Op tags for the 4-wave matmul kernel.

Fields

• ​value (Int): Integer tag identifying the op kind.

Implemented traits

AnyType, Deinitable, Equatable, Movable, ScheduleOps

comptime members

BARRIER

comptime BARRIER = FourWaveOps(Int(128))

Workgroup-wide barrier (e.g. s_barrier on AMD).

COMPUTE

comptime COMPUTE = FourWaveOps(Int(2))

Generic compute op (unused by 4-wave; reserved tag).

LOAD_A

comptime LOAD_A = FourWaveOps(Int(0))

DRAM->LDS prefetch of an A sub-tile.

LOAD_B

comptime LOAD_B = FourWaveOps(Int(1))

DRAM->LDS prefetch of a B sub-tile.

MMA

comptime MMA = FourWaveOps(Int(5))

MFMA quadrant compute op.

MMA_LOAD_A

comptime MMA_LOAD_A = FourWaveOps(Int(3))

LDS->register fragment load of an A quadrant.

MMA_LOAD_B

comptime MMA_LOAD_B = FourWaveOps(Int(4))

LDS->register fragment load of a B quadrant.

NONE

comptime NONE = FourWaveOps(Int(255))

Sentinel value for an absent or unspecified op.

SCHED_GROUP_BARRIER

comptime SCHED_GROUP_BARRIER = FourWaveOps(Int(133))

LLVM scheduling-group barrier hint (s_sched_group_barrier).

SCHEDULE_BARRIER

comptime SCHEDULE_BARRIER = FourWaveOps(Int(132))

LLVM scheduling-barrier hint (schedule_barrier).

SET_PRIO

comptime SET_PRIO = FourWaveOps(Int(131))

Wave priority hint (e.g. s_setprio).

WAIT_LGKM

comptime WAIT_LGKM = FourWaveOps(Int(130))

Wait on outstanding LDS / scalar-memory ops (e.g. s_waitcnt lgkmcnt).

WAIT_VM

comptime WAIT_VM = FourWaveOps(Int(129))

Wait on outstanding global-memory loads (e.g. s_waitcnt vmcnt).