Pranav Bhandarkar
Pranav Bhandarkar
@steven-johnson - See #7081 . @rootjalex's testcase passed with LLVM trunk but not LLVM 15. Is moving to trunk an option for you guys?
Perhaps. Checking commit logs in LLVM to see if our guys uploaded anything that sounds like a potential fix.
@steven-johnson & @dsharletg - Due to the winter storm that Texas saw last week, our office where all our hardware for building and testing is saw a massive power outage....
@vladl-innopeaktech - thank you for this PR. @aankit-ca, @suyogsarda - FYI. Can you PTAL?
LGTM. @aankit-ca can help you test this if you need. (sorry, i am on vacation with limited access to my work machine).
The reason this passes for you when running without hvx_v65 is becasue the gather instruction isnt generated if no hvx_v65 is used.
Would it be more helpful to indicate which outputs do not alias inputs? Absent such an indication the compiler always assumes aliasing. The downside though is that we'd then have...
> Maybe the better fix is to just use an actual mutex? On balance I do not see this slowing down code too much. The overhead, IMO would be negligible...
> I think the main reason we've avoided a mutex here is not performance but messiness in initialization. Can we zero initialize a global qurt_mutex_t? Yes, I have always wanted...
> For non-hexagon backends, .stmt files should capture the IR, and assembly output should capture the generated machine code in human readable form. Hexagon is compiled earlier in lowering, so...