I have failed to produce the reproducer for RISCV - the vectorizer does
not kick in for the provided sample:

test.c:13:6: note: vectorized 0 loops in function.
test.c:16:18: note: ***** Analysis failed with vector mode DI
test.c:20:6: note: vectorized 0 loops in function.
test.c:23:25: note: ***** Analysis failed with vector mode DI

But reading blogs/documentation[1] points to RISCV having a first-class
support for accessing elements of the vector, where as ARM SVE computes
offsets array (and that's where the bug is).

Conclusion: it is not possible to reproduce the issue with RISCV vector
addressing, but for consistency with earlier GCC SRU it would make sense
to upload gcc-XX-cross-ports so that the compilers behave consistently.

[1] 
https://fprox.substack.com/p/riscv-vector-extension-part5-1-load-store-memory
[2] 
https://lists.riscv.org/g/tech-vector-ext/attachment/691/0/riscv-v-spec-1.0.pdf

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2101084

Title:
  GCC produces wrong code for arm64+sve in some cases

To manage notifications about this bug go to:
https://bugs.launchpad.net/gcc/+bug/2101084/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to