[prev in list] [next in list] [prev in thread] [next in thread] 

List:       wine-devel
Subject:    Re: [PATCH vkd3d 2/3] vkd3d-shader: Introduce a compile option to use Unknown format for typed UAV r
From:       Conor McCarthy <conor.mccarthy.444 () gmail ! com>
Date:       2022-07-29 10:58:58
Message-ID: CAOua6-89j=fNgYuQpZP9XJ=AExRJ9nVeF1R_7UCymE1swUWO1A () mail ! gmail ! com
[Download RAW message or body]

On Fri, Jul 22, 2022 at 8:28 PM Henri Verbeet <hverbeet@gmail.com> wrote:

> I'm not entirely sure we're decoding these correctly. The format is
> certainly specified in the HLSL; it would seem surprising (though not
> inconceivable) that the information just gets dropped when generating
> the bytecode.
>

It appears from the spec we are not missing anything. The obvious way to
encode a format component count is in the dword which specifies the type of
each component. Those beyond the last component could be
D3D11_SB_RETURN_TYPE_UNUSED, but instead they all have the same type.
There's no other place to store it. The output from /dumpbin has four
components for the declaration too, and the load instruction specifies a
vec4 load. Looks like the actual format component count is left out.

[Attachment #3 (text/html)]

<div dir="ltr">On Fri, Jul 22, 2022 at 8:28 PM Henri Verbeet &lt;<a \
href="mailto:hverbeet@gmail.com">hverbeet@gmail.com</a>&gt; wrote:<br><div \
class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px \
0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">I&#39;m not entirely \
sure we&#39;re decoding these correctly. The format is<br> certainly specified in the \
HLSL; it would seem surprising (though not<br> inconceivable) that the information \
just gets dropped when generating<br> the \
bytecode.<br></blockquote><div><br></div><div>It appears from the spec we are not \
missing anything. The obvious way to encode a format component count is in the dword \
which specifies the type of each component. Those beyond the last component could be \
D3D11_SB_RETURN_TYPE_UNUSED, but instead they all have the same type. There&#39;s no \
other place to store it. The output from /dumpbin has four components for the \
declaration too, and the load instruction specifies a vec4 load. Looks like the \
actual format component count is left out.</div></div></div>



[prev in list] [next in list] [prev in thread] [next in thread]