[Mesa-dev] [PATCH 1/5] radv: do not resolve all image layers with compute inside a subpass
Bas Nieuwenhuizen
bas at basnieuwenhuizen.nl
Mon Jun 10 22:43:06 UTC 2019
r-b
On Mon, Jun 10, 2019 at 5:42 PM Samuel Pitoiset
<samuel.pitoiset at gmail.com> wrote:
>
> When resolving inside a subpass, we should rely on the framebuffer
> layer count instead of resolving all images layers. This should
> improve performance of layered resolves a bit.
>
> Signed-off-by: Samuel Pitoiset <samuel.pitoiset at gmail.com>
> ---
> src/amd/vulkan/radv_meta_resolve_cs.c | 12 ++++++++----
> 1 file changed, 8 insertions(+), 4 deletions(-)
>
> diff --git a/src/amd/vulkan/radv_meta_resolve_cs.c b/src/amd/vulkan/radv_meta_resolve_cs.c
> index b6c9e1c2b1f..fc4bcf27bb9 100644
> --- a/src/amd/vulkan/radv_meta_resolve_cs.c
> +++ b/src/amd/vulkan/radv_meta_resolve_cs.c
> @@ -508,6 +508,10 @@ radv_cmd_buffer_resolve_subpass_cs(struct radv_cmd_buffer *cmd_buffer)
> struct radv_framebuffer *fb = cmd_buffer->state.framebuffer;
> const struct radv_subpass *subpass = cmd_buffer->state.subpass;
> struct radv_subpass_barrier barrier;
> + uint32_t layer_count = fb->layers;
> +
> + if (subpass->view_mask)
> + layer_count = util_last_bit(subpass->view_mask);
>
> /* Resolves happen before the end-of-subpass barriers get executed, so
> * we have to make the attachment shader-readable.
> @@ -531,14 +535,14 @@ radv_cmd_buffer_resolve_subpass_cs(struct radv_cmd_buffer *cmd_buffer)
> .srcSubresource = (VkImageSubresourceLayers) {
> .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
> .mipLevel = src_iview->base_mip,
> - .baseArrayLayer = 0,
> - .layerCount = src_iview->image->info.array_size
> + .baseArrayLayer = src_iview->base_layer,
> + .layerCount = layer_count,
> },
> .dstSubresource = (VkImageSubresourceLayers) {
> .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
> .mipLevel = dst_iview->base_mip,
> - .baseArrayLayer = 0,
> - .layerCount = dst_iview->image->info.array_size
> + .baseArrayLayer = dst_iview->base_layer,
> + .layerCount = layer_count,
> },
> .srcOffset = (VkOffset3D){ 0, 0, 0 },
> .dstOffset = (VkOffset3D){ 0, 0, 0 },
> --
> 2.22.0
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
More information about the mesa-dev
mailing list