AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

hk: wire up indirect tess

Browse Source 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
This commit is contained in:
Alyssa Rosenzweig
2024-10-16 11:14:38 -04:00
parent a4e3ca1fc5
commit 3ee036b2b5
2 changed files with 67 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/asahi/vulkan/hk_cmd_buffer.h
+1
View File
@@ -262,6 +262,7 @@ struct hk_graphics_state {
/* Tessellation state */
uint64_t tess_out_draws;
uint64_t tess_grids;
/* Needed by vk_command_buffer::dynamic_graphics_state */
struct vk_vertex_input_state _dynamic_vi;
src/asahi/vulkan/hk_cmd_draw.c
+66 -50
View File
@@ -1276,17 +1276,22 @@ hk_upload_geometry_params(struct hk_cmd_buffer *cmd, struct hk_draw draw)
* another software stage (geometry shading), or if we need accurate primitive
* IDs in the linked fragment shader (since that would require a prefix sum
* anyway).
*
* The indirect tess path currently only supports indexed, but that part of
* libagx is on life support until I get to a wholesale replacement.
*/
static bool
hk_tess_needs_prefix_sum(struct hk_cmd_buffer *cmd)
hk_tess_needs_prefix_sum(struct hk_cmd_buffer *cmd, struct hk_draw draw)
{
struct hk_graphics_state *gfx = &cmd->state.gfx;
return gfx->shaders[MESA_SHADER_GEOMETRY] || gfx->generate_primitive_id;
return gfx->shaders[MESA_SHADER_GEOMETRY] || gfx->generate_primitive_id ||
draw.b.indirect;
}
static uint64_t
hk_upload_tess_params(struct hk_cmd_buffer *cmd, struct hk_draw draw)
static void
hk_upload_tess_params(struct hk_cmd_buffer *cmd, struct libagx_tess_args *out,
struct hk_draw draw)
{
struct hk_device *dev = hk_cmd_buffer_device(cmd);
struct vk_dynamic_graphics_state *dyn = &cmd->vk.dynamic_graphics_state;
@@ -1307,10 +1312,10 @@ hk_upload_tess_params(struct hk_cmd_buffer *cmd, struct hk_draw draw)
.tcs_per_vertex_outputs = tcs->info.tcs.per_vertex_outputs,
};
bool with_counts = hk_tess_needs_prefix_sum(cmd);
bool with_counts = hk_tess_needs_prefix_sum(cmd, draw);
/* This assumes !with_counts, if we have counts it's only one draw */
uint32_t draw_stride_el = tes->info.ts.point_mode ? 4 : 6;
uint32_t draw_stride_el =
with_counts ? 5 : (tes->info.ts.point_mode ? 4 : 6);
size_t draw_stride_B = draw_stride_el * sizeof(uint32_t);
/* heap is allocated by hk_geometry_state */
@@ -1338,7 +1343,7 @@ hk_upload_tess_params(struct hk_cmd_buffer *cmd, struct hk_draw draw)
if (with_counts) {
/* Single API draw */
alloc += 5 * sizeof(uint32_t);
alloc += draw_stride_B;
} else {
/* Padding added because VDM overreads */
alloc += (draw_stride_B * unrolled_patches) +
@@ -1352,8 +1357,6 @@ hk_upload_tess_params(struct hk_cmd_buffer *cmd, struct hk_draw draw)
args.nr_patches = unrolled_patches;
args.out_draws = blob.gpu + draw_offs;
gfx->tess_out_draws = args.out_draws;
if (with_counts) {
args.counts = blob.gpu + count_offs;
} else {
@@ -1366,38 +1369,40 @@ hk_upload_tess_params(struct hk_cmd_buffer *cmd, struct hk_draw draw)
}
}
} else {
unreachable("todo: indirect with tess");
#if 0
args.tcs_statistic = agx_get_query_address(
batch, ctx->pipeline_statistics[PIPE_STAT_QUERY_HS_INVOCATIONS]);
args.tcs_statistic = hk_pipeline_stat_addr(
cmd,
VK_QUERY_PIPELINE_STATISTIC_TESSELLATION_CONTROL_SHADER_PATCHES_BIT);
args.indirect = agx_indirect_buffer_ptr(batch, indirect);
args.indirect = draw.b.ptr;
/* Allocate 3x indirect global+local grids for VS/TCS/tess */
uint32_t grid_stride = sizeof(uint32_t) * 6;
args.grids = agx_pool_alloc_aligned(&batch->pool, grid_stride * 3, 4).gpu;
args.grids = hk_pool_alloc(cmd, grid_stride * 3, 4).gpu;
gfx->tess_grids = args.grids;
vs_grid = agx_grid_indirect_local(args.grids + 0 * grid_stride);
tcs_grid = agx_grid_indirect_local(args.grids + 1 * grid_stride);
tess_grid = agx_grid_indirect_local(args.grids + 2 * grid_stride);
args.vertex_outputs = ctx->vs->b.info.outputs;
struct hk_shader *vs = hk_bound_sw_vs(gfx);
args.vertex_outputs = vs->b.info.outputs;
args.vertex_output_buffer_ptr =
agx_pool_alloc_aligned(&batch->pool, 8, 8).gpu;
batch->uniforms.vertex_output_buffer_ptr = args.vertex_output_buffer_ptr;
gfx->root +
offsetof(struct hk_root_descriptor_table, draw.vertex_output_buffer);
args.ia = gfx->descriptors.root.draw.input_assembly;
if (with_counts) {
args.out_draws = agx_pool_alloc_aligned_with_bo(
&batch->pool, draw_stride, 4, &draw_bo)
.gpu;
args.out_draws = hk_pool_alloc(cmd, draw_stride_B, 4).gpu;
} else {
unreachable("need an extra indirection...");
}
#endif
if (draw.indexed) {
args.in_index_buffer = draw.index.addr;
args.in_index_size_B = agx_index_size_to_B(draw.index_size);
args.in_index_buffer_range_el =
draw.index.range / args.in_index_size_B;
}
}
return hk_pool_upload(cmd, &args, sizeof(args), 8);
gfx->tess_out_draws = args.out_draws;
memcpy(out, &args, sizeof(args));
}
static struct hk_api_shader *
@@ -1650,6 +1655,8 @@ hk_launch_tess(struct hk_cmd_buffer *cmd, struct hk_cs *cs, struct hk_draw draw)
struct vk_dynamic_graphics_state *dyn = &cmd->vk.dynamic_graphics_state;
uint32_t input_patch_size = dyn->ts.patch_control_points;
bool with_counts = hk_tess_needs_prefix_sum(cmd, draw);
uint64_t state = gfx->descriptors.root.draw.tess_params;
hk_ensure_cs_has_space(cmd, cs, 0x2000 /*XXX*/);
@@ -1661,26 +1668,26 @@ hk_launch_tess(struct hk_cmd_buffer *cmd, struct hk_cs *cs, struct hk_draw draw)
/* Setup grids */
if (draw.b.indirect) {
perf_debug(dev, "Indirect tessellation");
unreachable("todo: indirect tess");
#if 0
struct agx_gs_setup_indirect_key key = {.prim = mode};
struct hk_shader *gsi =
hk_meta_kernel(dev, agx_nir_gs_setup_indirect, &key, sizeof(key));
struct agx_tess_setup_indirect_key key = {
.point_mode = tes->info.ts.point_mode,
.with_counts = with_counts,
};
uint64_t push = hk_upload_gsi_params(cmd, draw);
uint32_t usc = hk_upload_usc_words_kernel(cmd, gsi, &push, sizeof(push));
struct hk_shader *tsi =
hk_meta_kernel(dev, agx_nir_tess_setup_indirect, &key, sizeof(key));
hk_dispatch_with_usc(dev, cs, gsi, usc, hk_grid(1, 1, 1),
/* TODO */
uint32_t usc =
hk_upload_usc_words_kernel(cmd, tsi, &state, sizeof(state));
hk_dispatch_with_usc(dev, cs, tsi, usc, hk_grid(1, 1, 1),
hk_grid(1, 1, 1));
uint64_t geometry_params = desc->root.draw.geometry_params;
grid_vs = hk_grid_indirect(geometry_params +
offsetof(struct agx_geometry_params, vs_grid));
grid_gs = hk_grid_indirect(geometry_params +
offsetof(struct agx_geometry_params, gs_grid));
#endif
uint32_t grid_stride = sizeof(uint32_t) * 6;
grid_vs = hk_grid_indirect_local(gfx->tess_grids + 0 * grid_stride);
grid_tcs = hk_grid_indirect_local(gfx->tess_grids + 1 * grid_stride);
grid_tess = hk_grid_indirect_local(gfx->tess_grids + 2 * grid_stride);
} else {
uint32_t patches = draw.b.count[0] / input_patch_size;
grid_vs = grid_tcs = draw.b;
@@ -1721,11 +1728,6 @@ hk_launch_tess(struct hk_cmd_buffer *cmd, struct hk_cs *cs, struct hk_draw draw)
dev, cs, tcs, hk_upload_usc_words(cmd, tcs, tcs->only_linked), grid_tcs,
hk_grid(tcs->info.tcs.output_patch_size, 1, 1));
/* TODO indirect */
bool with_counts = hk_tess_needs_prefix_sum(cmd);
uint64_t state = gfx->descriptors.root.draw.tess_params;
/* If the domain is flipped, we need to flip the winding order */
bool ccw = tes->info.ts.ccw;
ccw ^= dyn->ts.domain_origin == VK_TESSELLATION_DOMAIN_ORIGIN_LOWER_LEFT;
@@ -3105,8 +3107,10 @@ hk_flush_dynamic_state(struct hk_cmd_buffer *cmd, struct hk_cs *cs,
}
}
struct agx_ptr tess_args = {0};
if (gfx->shaders[MESA_SHADER_TESS_EVAL]) {
gfx->descriptors.root.draw.tess_params = hk_upload_tess_params(cmd, draw);
tess_args = hk_pool_alloc(cmd, sizeof(struct libagx_tess_args), 4);
gfx->descriptors.root.draw.tess_params = tess_args.gpu;
gfx->descriptors.root_dirty = true;
}
@@ -3124,6 +3128,18 @@ hk_flush_dynamic_state(struct hk_cmd_buffer *cmd, struct hk_cs *cs,
if (gfx->descriptors.root_dirty) {
gfx->root =
hk_cmd_buffer_upload_root(cmd, VK_PIPELINE_BIND_POINT_GRAPHICS);
/* Tess parameters depend on the root address, so we defer the upload
* until after uploading root. But the root depends on the tess address,
* so we allocate tess parameters before uploading root.
*
* This whole mechanism is a mess ported over from the GL driver. I'm
* planning to do a massive rework of indirect geom/tess so I'm trying not
* to perfectionism it in the mean time.
*/
if (tess_args.cpu) {
hk_upload_tess_params(cmd, tess_args.cpu, draw);
}
}
/* Hardware dynamic state must be deferred until after the root and fast