On Mon, 9 Aug 2021 at 12:58, John Naylor wrote:
>
> On Sun, Aug 8, 2021 at 8:31 PM David Rowley wrote:
> >
> > I've attached a v2 patch which I think is more along the lines of what
> > you had in mind.
>
> LGTM
Thanks for the review.
Pushed.
David
On Sun, Aug 8, 2021 at 8:31 PM David Rowley wrote:
>
> I've attached a v2 patch which I think is more along the lines of what
> you had in mind.
LGTM
--
John Naylor
EDB: http://www.enterprisedb.com
On Thu, 5 Aug 2021 at 07:02, John Naylor wrote:
> > #if defined(_MSC_VER) && defined(_WIN64)
> > #define HAVE_X86_64_POPCNTQ
> > #endif
>
> That seems fine. I don't know PG can build with Arm on Windows, but for the
> cpuid to work, it seems safer to also check for __x86_64?
That's a good point.
On Tue, Aug 3, 2021 at 11:36 PM David Rowley wrote:
>
> On Tue, 3 Aug 2021 at 22:43, John Naylor
wrote:
> > 1. the __popcnt64() intrinsic is put inside pg_popcount64_asm(), which
is a bit of a misnomer since it's not assembly. Renaming s/_asm/_fast/
would help it look better. But then looking aro
On Tue, 3 Aug 2021 at 22:43, John Naylor wrote:
> 1. the __popcnt64() intrinsic is put inside pg_popcount64_asm(), which is a
> bit of a misnomer since it's not assembly. Renaming s/_asm/_fast/ would help
> it look better. But then looking around, other platforms have intrinsics
> coded, but fo
On Tue, Aug 3, 2021 at 10:43 PM John Naylor
wrote:
> (Side note, but sort of related to #1 above: non-x86 platforms have to
> indirect through a function pointer even though they have no fast
> implementation to make it worth their while. It would be better for them if
> the "slow" implementati
On Tue, Aug 3, 2021 at 5:03 AM David Rowley wrote:
>
> Going by [1], it looks like we can use the __popcnt and __popcnt64
> intrinsic functions on MSVC if the CPU supports POPCNT. We already
> have code to check for that, we just need to enable it on MSVC.
>
> The attached patch seems to be all t
