Supported GraphBLAS Spec Operations

Supported GraphBLAS Spec operations

Operation

Matrix

Vector

accum

mask

compl. mask

mlir name

comment

mxm

Y

Y

Y

Y

matrix_multiply

vxm

Y

Y

Y

Y

matrix_multiply

mxv

Y

Y

Y

Y

matrix_multiply

eWiseMult

Y

Y

Y

N

N

intersect

eWiseAdd

Y

Y

Y

N

N

union

apply

Y

Y

Y

N

N

matrix_apply

apply_Binop1st

Y

Y

Y

N

N

matrix_apply

apply_Binop2nd

Y

N

Y

N

N

matrix_apply

select (no val)

Y

N

Y

N

N

matrix_select

select (w/ val)

Y

N

Y

N

N

matrix_select

reduce_to_scalar

Y

Y

N

reduce_to_scalar

reduce_to_vector

Y

Y

N

N

reduce_to_vector

transpose

Y

Y

N

N

transpose

kronecker

N

N

N

N

diag

Y

Y

diag

assign

N

N

N

N

N

col/row assign

N

N

N

N

subassign

N

N

N

N

N

GxB

assign scalar many

N

N

N

N

N

subassign scalar many

N

N

N

N

N

GxB

extract

N

N

N

N

N

col extract

N

N

N

N

set element

N

N

extract element

N

N

remove element

N

N

build

N

N

clear

N

N

dup

Y

Y

dup

size/nrows/ncols

Y

Y

size/num_rows/num_cols

nvals

Y

Y

num_vals

resize

N

N

extractTuples

N

N

concat

N

GxB

split

N

GxB

isequal

Y

Y

equal

custom

vxv/inner

Y

N

matrix_multiply

custom

argmin/argmax

Y

Y

Y

N

N

vector_argmin/vector_argmax

custom

select_rows(“random”)

Y

Y

N

matrix_select_random

custom

select(“random_weighted”)

Y

N

Y

N

N

matrix_select_random

custom
Supported GraphBLAS operations for updating

accumulation

mask

replace

is supported

True

True

True

N

True

True

False

N

True

False

Y

False

True

True

N

False

True

False

N

False

False

Y

previous

GraphBLAS Dialect Op Reference

next

GraphBLAS Passes and graphblas-opt