Open periscop opened 10 years ago
(reported by Tobias Grosser) pluto compiled durbin from polybench 3.2 crashes at execution
(explanation by Uday Bondhugula) the reason is
There is a missing parenthesis in the generated code:
for (t3=t2;t3<=2_t2-1;t3++) { y[-t2+t3][t2]=y[-t2+t3][t2-1]+alpha[t2]_y[t2- -t2+t3-1][t2-1];; ^^^^^ }
since the remappings are:
k = t2 i = t3-t2
It should've been "t2- (-t2+t3)-1]
Can you quickly look at the thread above? It should make sense to you quickly. Given this, why shouldn't the remappings be not generated as:
S8(t2, (t3-t2))
instead of
S8(t2, t3-t2);
?
(reported by Tobias Grosser) pluto compiled durbin from polybench 3.2 crashes at execution
(explanation by Uday Bondhugula) the reason is
There is a missing parenthesis in the generated code:
for (t3=t2;t3<=2_t2-1;t3++) { y[-t2+t3][t2]=y[-t2+t3][t2-1]+alpha[t2]_y[t2- -t2+t3-1][t2-1];; ^^^^^ }
since the remappings are:
k = t2 i = t3-t2
It should've been "t2- (-t2+t3)-1]
Can you quickly look at the thread above? It should make sense to you quickly. Given this, why shouldn't the remappings be not generated as:
S8(t2, (t3-t2))
instead of
S8(t2, t3-t2);
?