[Libclc-dev] [PATCH 1/3] geometric/floatn.inc: Add vec8 and vec16 types

[Aaron Watry]

Do we have a reason to do this expansion to float8/float16 types?

The CL 1.2 and 2.0  specs say that all of the geometric functions operate
on float, float2, float3, and float4 data types (and the double
equivalents), but make no mention of float8/float16 being supported (which
seems a bit silly given that almost everything else supports 8/16 element
vectors if they support them at all).

--Aaron

On Tue, Feb 24, 2015 at 2:15 PM, Tom Stellard <thomas.stellard at amd.com>
wrote:

> ---
>  generic/include/clc/geometric/floatn.inc | 16 ++++++++++++++++
>  1 file changed, 16 insertions(+)
>
> diff --git a/generic/include/clc/geometric/floatn.inc
> b/generic/include/clc/geometric/floatn.inc
> index fb7a9ae..b683b85 100644
> --- a/generic/include/clc/geometric/floatn.inc
> +++ b/generic/include/clc/geometric/floatn.inc
> @@ -16,6 +16,14 @@
>  #include __CLC_BODY
>  #undef __CLC_FLOATN
>
> +#define __CLC_FLOATN float8
> +#include __CLC_BODY
> +#undef __CLC_FLOATN
> +
> +#define __CLC_FLOATN float16
> +#include __CLC_BODY
> +#undef __CLC_FLOATN
> +
>  #undef __CLC_FLOAT
>
>  #ifdef cl_khr_fp64
> @@ -38,6 +46,14 @@
>  #include __CLC_BODY
>  #undef __CLC_FLOATN
>
> +#define __CLC_FLOATN double8
> +#include __CLC_BODY
> +#undef __CLC_FLOATN
> +
> +#define __CLC_FLOATN double16
> +#include __CLC_BODY
> +#undef __CLC_FLOATN
> +
>  #undef __CLC_FLOAT
>
>  #endif
> --
> 2.0.4
>
>
> _______________________________________________
> Libclc-dev mailing list
> Libclc-dev at pcc.me.uk
> http://www.pcc.me.uk/cgi-bin/mailman/listinfo/libclc-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/libclc-dev/attachments/20150225/4144b724/attachment.html>