radv/gfx9: fix 3d image clears on compute queues

This fixes some of the broken:
dEQP-VK.synchronization.op.multi_queue.*64x64x8* tests.

Fixes: e38685cc62 'Revert "radv: disable support for VEGA for now."'
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This commit is contained in:
Dave Airlie 2017-12-29 08:52:07 +10:00
parent d08f267814
commit 09612a62e1
2 changed files with 65 additions and 9 deletions

View file

@ -667,15 +667,16 @@ radv_device_finish_meta_itoi_state(struct radv_device *device)
}
static nir_shader *
build_nir_cleari_compute_shader(struct radv_device *dev)
build_nir_cleari_compute_shader(struct radv_device *dev, bool is_3d)
{
nir_builder b;
const struct glsl_type *img_type = glsl_sampler_type(GLSL_SAMPLER_DIM_2D,
enum glsl_sampler_dim dim = is_3d ? GLSL_SAMPLER_DIM_3D : GLSL_SAMPLER_DIM_2D;
const struct glsl_type *img_type = glsl_sampler_type(dim,
false,
false,
GLSL_TYPE_FLOAT);
nir_builder_init_simple_shader(&b, NULL, MESA_SHADER_COMPUTE, NULL);
b.shader->info.name = ralloc_strdup(b.shader, "meta_cleari_cs");
b.shader->info.name = ralloc_strdup(b.shader, is_3d ? "meta_cleari_cs_3d" : "meta_cleari_cs");
b.shader->info.cs.local_size[0] = 16;
b.shader->info.cs.local_size[1] = 16;
b.shader->info.cs.local_size[2] = 1;
@ -696,12 +697,29 @@ build_nir_cleari_compute_shader(struct radv_device *dev)
nir_intrinsic_instr *clear_val = nir_intrinsic_instr_create(b.shader, nir_intrinsic_load_push_constant);
nir_intrinsic_set_base(clear_val, 0);
nir_intrinsic_set_range(clear_val, 16);
nir_intrinsic_set_range(clear_val, 20);
clear_val->src[0] = nir_src_for_ssa(nir_imm_int(&b, 0));
clear_val->num_components = 4;
nir_ssa_dest_init(&clear_val->instr, &clear_val->dest, 4, 32, "clear_value");
nir_builder_instr_insert(&b, &clear_val->instr);
nir_intrinsic_instr *layer = nir_intrinsic_instr_create(b.shader, nir_intrinsic_load_push_constant);
nir_intrinsic_set_base(layer, 0);
nir_intrinsic_set_range(layer, 20);
layer->src[0] = nir_src_for_ssa(nir_imm_int(&b, 16));
layer->num_components = 1;
nir_ssa_dest_init(&layer->instr, &layer->dest, 1, 32, "layer");
nir_builder_instr_insert(&b, &layer->instr);
nir_ssa_def *global_z = nir_iadd(&b, nir_channel(&b, global_id, 2), &layer->dest.ssa);
nir_ssa_def *comps[4];
comps[0] = nir_channel(&b, global_id, 0);
comps[1] = nir_channel(&b, global_id, 1);
comps[2] = global_z;
comps[3] = nir_imm_int(&b, 0);
global_id = nir_vec(&b, comps, 4);
nir_intrinsic_instr *store = nir_intrinsic_instr_create(b.shader, nir_intrinsic_image_store);
store->src[0] = nir_src_for_ssa(global_id);
store->src[1] = nir_src_for_ssa(nir_ssa_undef(&b, 1, 32));
@ -717,8 +735,10 @@ radv_device_init_meta_cleari_state(struct radv_device *device)
{
VkResult result;
struct radv_shader_module cs = { .nir = NULL };
cs.nir = build_nir_cleari_compute_shader(device);
struct radv_shader_module cs_3d = { .nir = NULL };
cs.nir = build_nir_cleari_compute_shader(device, false);
if (device->physical_device->rad_info.chip_class >= GFX9)
cs_3d.nir = build_nir_cleari_compute_shader(device, true);
/*
* two descriptors one for the image being sampled
@ -752,7 +772,7 @@ radv_device_init_meta_cleari_state(struct radv_device *device)
.setLayoutCount = 1,
.pSetLayouts = &device->meta_state.cleari.img_ds_layout,
.pushConstantRangeCount = 1,
.pPushConstantRanges = &(VkPushConstantRange){VK_SHADER_STAGE_COMPUTE_BIT, 0, 16},
.pPushConstantRanges = &(VkPushConstantRange){VK_SHADER_STAGE_COMPUTE_BIT, 0, 20},
};
result = radv_CreatePipelineLayout(radv_device_to_handle(device),
@ -786,10 +806,38 @@ radv_device_init_meta_cleari_state(struct radv_device *device)
if (result != VK_SUCCESS)
goto fail;
if (device->physical_device->rad_info.chip_class >= GFX9) {
/* compute shader */
VkPipelineShaderStageCreateInfo pipeline_shader_stage_3d = {
.sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO,
.stage = VK_SHADER_STAGE_COMPUTE_BIT,
.module = radv_shader_module_to_handle(&cs_3d),
.pName = "main",
.pSpecializationInfo = NULL,
};
VkComputePipelineCreateInfo vk_pipeline_info_3d = {
.sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO,
.stage = pipeline_shader_stage_3d,
.flags = 0,
.layout = device->meta_state.cleari.img_p_layout,
};
result = radv_CreateComputePipelines(radv_device_to_handle(device),
radv_pipeline_cache_to_handle(&device->meta_state.cache),
1, &vk_pipeline_info_3d, NULL,
&device->meta_state.cleari.pipeline_3d);
if (result != VK_SUCCESS)
goto fail;
ralloc_free(cs_3d.nir);
}
ralloc_free(cs.nir);
return VK_SUCCESS;
fail:
ralloc_free(cs.nir);
ralloc_free(cs_3d.nir);
return result;
}
@ -805,6 +853,8 @@ radv_device_finish_meta_cleari_state(struct radv_device *device)
&state->alloc);
radv_DestroyPipeline(radv_device_to_handle(device),
state->cleari.pipeline, &state->alloc);
radv_DestroyPipeline(radv_device_to_handle(device),
state->cleari.pipeline_3d, &state->alloc);
}
void
@ -1163,19 +1213,24 @@ radv_meta_clear_image_cs(struct radv_cmd_buffer *cmd_buffer,
create_iview(cmd_buffer, dst, &dst_iview);
cleari_bind_descriptors(cmd_buffer, &dst_iview);
if (device->physical_device->rad_info.chip_class >= GFX9 &&
dst->image->type == VK_IMAGE_TYPE_3D)
pipeline = cmd_buffer->device->meta_state.cleari.pipeline_3d;
radv_CmdBindPipeline(radv_cmd_buffer_to_handle(cmd_buffer),
VK_PIPELINE_BIND_POINT_COMPUTE, pipeline);
unsigned push_constants[4] = {
unsigned push_constants[5] = {
clear_color->uint32[0],
clear_color->uint32[1],
clear_color->uint32[2],
clear_color->uint32[3],
dst->layer,
};
radv_CmdPushConstants(radv_cmd_buffer_to_handle(cmd_buffer),
device->meta_state.cleari.img_p_layout,
VK_SHADER_STAGE_COMPUTE_BIT, 0, 16,
VK_SHADER_STAGE_COMPUTE_BIT, 0, 20,
push_constants);
radv_unaligned_dispatch(cmd_buffer, dst->image->info.width, dst->image->info.height, 1);

View file

@ -449,6 +449,7 @@ struct radv_meta_state {
VkPipelineLayout img_p_layout;
VkDescriptorSetLayout img_ds_layout;
VkPipeline pipeline;
VkPipeline pipeline_3d;
} cleari;
struct {