wine/vkd3d
0
0
Fork 0
mirror of https://gitlab.winehq.org/wine/vkd3d.git synced 2024-11-21 16:46:41 -08:00
Code Issues Packages Projects Releases Wiki Activity

vkd3d: Limit the workgroup X count for buffer UAV clears to the supported max.

Browse Source
This commit is contained in:
Conor McCarthy 2024-07-05 15:29:11 +10:00 committed by Henri Verbeet
parent 3e25356abb
commit ae0faf4500
Notes: Henri Verbeet 2024-07-17 15:59:23 +02:00 
Approved-by: Giovanni Mascellani (@giomasce)
Approved-by: Henri Verbeet (@hverbeet)
Merge-Request: https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/935
1 changed files with 24 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
libs/vkd3d/command.c
View File

@ -5289,11 +5289,13 @@ static void d3d12_command_list_clear_uav(struct d3d12_command_list *list,
struct d3d12_resource *resource, struct vkd3d_view *descriptor, const VkClearColorValue *clear_colour, struct d3d12_resource *resource, struct vkd3d_view *descriptor, const VkClearColorValue *clear_colour,
unsigned int rect_count, const D3D12_RECT *rects) unsigned int rect_count, const D3D12_RECT *rects)
{ {
const VkPhysicalDeviceLimits *device_limits = &list->device->vk_info.device_limits;
const struct vkd3d_vk_device_procs *vk_procs = &list->device->vk_procs; const struct vkd3d_vk_device_procs *vk_procs = &list->device->vk_procs;
unsigned int i, miplevel_idx, layer_count; unsigned int i, miplevel_idx, layer_count;
struct vkd3d_uav_clear_pipeline pipeline; struct vkd3d_uav_clear_pipeline pipeline;
struct vkd3d_uav_clear_args clear_args; struct vkd3d_uav_clear_args clear_args;
const struct vkd3d_resource_view *view; const struct vkd3d_resource_view *view;
uint32_t count_x, count_y, count_z;
VkDescriptorImageInfo image_info; VkDescriptorImageInfo image_info;
D3D12_RECT full_rect, curr_rect; D3D12_RECT full_rect, curr_rect;
VkWriteDescriptorSet write_set; VkWriteDescriptorSet write_set;
@ -5384,18 +5386,32 @@ static void d3d12_command_list_clear_uav(struct d3d12_command_list *list,
if (curr_rect.left >= curr_rect.right || curr_rect.top >= curr_rect.bottom) if (curr_rect.left >= curr_rect.right || curr_rect.top >= curr_rect.bottom)
continue; continue;
clear_args.offset.x = curr_rect.left;
clear_args.offset.y = curr_rect.top; clear_args.offset.y = curr_rect.top;
clear_args.extent.width = curr_rect.right - curr_rect.left;
clear_args.extent.height = curr_rect.bottom - curr_rect.top; clear_args.extent.height = curr_rect.bottom - curr_rect.top;
count_y = vkd3d_compute_workgroup_count(clear_args.extent.height, pipeline.group_size.height);
count_z = vkd3d_compute_workgroup_count(layer_count, pipeline.group_size.depth);
if (count_y > device_limits->maxComputeWorkGroupCount[1])
FIXME("Group Y count %u exceeds max %u.\n", count_y, device_limits->maxComputeWorkGroupCount[1]);
if (count_z > device_limits->maxComputeWorkGroupCount[2])
FIXME("Group Z count %u exceeds max %u.\n", count_z, device_limits->maxComputeWorkGroupCount[2]);
do
{
clear_args.offset.x = curr_rect.left;
clear_args.extent.width = curr_rect.right - curr_rect.left;
count_x = vkd3d_compute_workgroup_count(clear_args.extent.width, pipeline.group_size.width);
count_x = min(count_x, device_limits->maxComputeWorkGroupCount[0]);
VK_CALL(vkCmdPushConstants(list->vk_command_buffer, pipeline.vk_pipeline_layout, VK_CALL(vkCmdPushConstants(list->vk_command_buffer, pipeline.vk_pipeline_layout,
VK_SHADER_STAGE_COMPUTE_BIT, 0, sizeof(clear_args), &clear_args)); VK_SHADER_STAGE_COMPUTE_BIT, 0, sizeof(clear_args), &clear_args));
VK_CALL(vkCmdDispatch(list->vk_command_buffer, VK_CALL(vkCmdDispatch(list->vk_command_buffer, count_x, count_y, count_z));
vkd3d_compute_workgroup_count(clear_args.extent.width, pipeline.group_size.width),
vkd3d_compute_workgroup_count(clear_args.extent.height, pipeline.group_size.height), curr_rect.left += count_x * pipeline.group_size.width;
vkd3d_compute_workgroup_count(layer_count, pipeline.group_size.depth))); }
while (curr_rect.right > curr_rect.left);
} }
} }