On 2015/03/03 05:54:34, Benedikt Meurer wrote:
A few years ago I did a similar patch for OCaml
(http://caml.inria.fr/mantis/view.php?id=5180), but using movapd, which
gave a
nice speed up then. I remember that I tried to change this in V8 once,
but it
didn't make any difference on our Xeon's.
I'm fine with landing this CL, just two questions:
(a) Should we use movapd instead of movaps?
(b) Should we do this all over the place on x64, not just in TurboFan?
The patch is just to align with Crankshaft, which has already use movaps in
x64
backend(src/x64/lithium-gap-resolver-x64.cc).
It also restores the regression caused by r26849 on HSW when AVX is
disabled.
Thanks
-Weiliang
https://codereview.chromium.org/972053002/
--
--
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].
For more options, visit https://groups.google.com/d/optout.
