Hi all!

I was looking at MachineOperatorReducer in src/compiler and I've noticed 
that, while there are reductions for signed and unsigned 32-bit integer 
divisions (e.g. 
MachineOperatorReducer::ReduceInt32Div), there is no such thing for 64-bit 
divisions.

Is there a reason why 64-bit divisions are not optimized? Maybe the *ctz 
*operation 
is too slow on 64 bits if there is no intrinsic *ctz* instruction? That's 
just a wild (and likely incorrect) guess, I'd just be curious to know more.

To give some context, I am part of the ongoing effort to port V8 to 
RISCV32, and we noticed this possibility for optimization in this issue 
<https://github.com/riscv-collab/v8/issues/574>.

If there is no specific reason, I'd be happy to make this contribution.

Best to all,
- Alvise

-- 
-- 
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
--- 
You received this message because you are subscribed to the Google Groups 
"v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/v8-dev/8f6d3e85-0141-416c-a82a-ed622b2ff30bn%40googlegroups.com.

Reply via email to