Displaying 1 result from an estimated 1 matches for "scalar_loop".
2014 Aug 30
2
[LLVMdev] Loop vectorizer preamble
...Instead, they have a greater-than or less-than condition. If this is the
case, we should be able to elide all of our logic with Count and just count
down until the test is broken. For example:
for (i = 0; i < n; ++i)
...
->
count = 0
loop:
...
count += VF * UF
if count >= n goto scalar_loop else goto loop
This could remove a lot of overflow checks and "urem"s from real code.
Also, we don't currently coalesce overflow checks, vector memchecks and
trip count computation for adjacent and similar loops. For example:
for (i = 0; i < n/2; ++i)
p[i] = q[i+1];
for...