On Thu, Sep 17, 2015 at 05:00:08PM +0100, Neil Roberts wrote:
> When 16x MSAA is used for sampling with texelFetch the compiler needs
> to use a different instruction which passes more arguments for the MCS
> data. Previously on skl+ it was unconditionally using this new
> instruction. However since 16x MSAA is probably going to be pretty
> rare, it is probably worthwhile to avoid using this instruction for
> the other sample counts. In order to do that this patch adds a new
> member to brw_sampler_prog_key_data to track when a sampler refers to
> a buffer with 16 samples.
>
> Note that this isn't done for the vec4 backend because it wouldn't
> change how many registers it uses.
Hmm. I was of the other opinion actually, I'd prefer to always use the
new instruction to limit the variability between tests. I suppose if we're often
passing the LOD parameter that is problematic, but we don't do that either,
right?
> ---
> src/mesa/drivers/dri/i965/brw_fs_visitor.cpp | 2 +-
> src/mesa/drivers/dri/i965/brw_program.h | 7 +++++++
> src/mesa/drivers/dri/i965/brw_wm.c | 8 ++++++++
> 3 files changed, 16 insertions(+), 1 deletion(-)
>
> diff --git a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
> b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
> index ea7b3eb..7937c48 100644
> --- a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
> +++ b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
> @@ -287,7 +287,7 @@ fs_visitor::emit_texture(ir_texture_opcode op,
> opcode = SHADER_OPCODE_TXF_LOGICAL;
> break;
> case ir_txf_ms:
> - if (devinfo->gen >= 9)
> + if ((key_tex->msaa_16 & (1 << sampler)))
> opcode = SHADER_OPCODE_TXF_CMS_W_LOGICAL;
> else
> opcode = SHADER_OPCODE_TXF_CMS_LOGICAL;
> diff --git a/src/mesa/drivers/dri/i965/brw_program.h
> b/src/mesa/drivers/dri/i965/brw_program.h
> index 00e8f3f..b27ab7f 100644
> --- a/src/mesa/drivers/dri/i965/brw_program.h
> +++ b/src/mesa/drivers/dri/i965/brw_program.h
> @@ -72,6 +72,13 @@ struct brw_sampler_prog_key_data {
> uint32_t compressed_multisample_layout_mask;
>
> /**
> + * Whether this sampler is using 16x multisampling. If so fetching from
> + * this sampler will be handled with a different instruction, ld2ms_w
> + * instead of ld2ms.
> + */
> + uint32_t msaa_16;
> +
> + /**
> * For Sandybridge, which shader w/a we need for gather quirks.
> */
> enum gen6_gather_sampler_wa gen6_gather_wa[MAX_SAMPLERS];
> diff --git a/src/mesa/drivers/dri/i965/brw_wm.c
> b/src/mesa/drivers/dri/i965/brw_wm.c
> index 35c0908..2d6f10d 100644
> --- a/src/mesa/drivers/dri/i965/brw_wm.c
> +++ b/src/mesa/drivers/dri/i965/brw_wm.c
> @@ -292,6 +292,9 @@ brw_debug_recompile_sampler_key(struct brw_context *brw,
> found |= key_debug(brw, "compressed multisample layout",
> old_key->compressed_multisample_layout_mask,
> key->compressed_multisample_layout_mask);
> + found |= key_debug(brw, "16x msaa",
> + old_key->msaa_16,
> + key->msaa_16);
>
> for (unsigned int i = 0; i < MAX_SAMPLERS; i++) {
> found |= key_debug(brw, "textureGather workarounds",
> @@ -451,6 +454,11 @@ brw_populate_sampler_prog_key_data(struct gl_context
> *ctx,
> if (brw->gen >= 7 &&
> intel_tex->mt->msaa_layout == INTEL_MSAA_LAYOUT_CMS) {
> key->compressed_multisample_layout_mask |= 1 << s;
> +
> + if (intel_tex->mt->num_samples >= 16) {
> + assert(brw->gen >= 9);
> + key->msaa_16 |= 1 << s;
> + }
> }
> }
> }
> --
> 1.9.3
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/mesa-dev
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
