Question regarding optimization patterns for shaders

--0000000000003f3161063f80d89f
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

A graphics middleware library I'm working on outputs a lot of shaders that
involve writing to some temporary location and then copying it into the
output on exit. This seems to be working fine for standard outputs, but I=
=E2=80=99m
curious if for e.g. mesh shaders, with larger output arrays, this is
handled equally well. In particular, we output shaders with a pattern of
copying into a large array in workgroup memory and then copying each
element on exit.

This is my first email to the mesa list. Thanks for reading!

--0000000000003f3161063f80d89f
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div><div dir=3D"ltr">A graphics middleware library I&#39;m working on outp=
uts a lot of shaders that involve writing to some temporary location and th=
en copying it into the output on exit. This seems to be working fine for st=
andard outputs, but I=E2=80=99m curious if for e.g. mesh shaders, with larg=
er output arrays, this is handled equally well. In particular, we output sh=
aders with a pattern of copying into a large array in workgroup=C2=A0memory=
 and then copying each element on exit.</div><div dir=3D"ltr"><br></div><di=
v dir=3D"ltr">This is my first email to the mesa list. Thanks for reading!<=
/div>
</div>

--0000000000003f3161063f80d89f--