#matrix multiply large array size for intel machine
source: gemm.f90
procedure: gemm
format: rose
loop: 0

TI = 128
#TI = 4
TJ = 8
#TK = 3
TK = 512
UI = 2
UJ = 2

permute([3,1,2])
tile(0,2,TJ)
#print space
tile(0,2,TI)
#print space
tile(0,5,TK)
#print space


datacopy(0,3,A,false,-1)
#print space

datacopy(0,4,B)
unroll(0,4,UI)                                                            
unroll(0,5,UJ)