This repository has been archived on 2023-10-09. You can view files and clone it, but cannot push or open issues or pull requests.
Files
blender-archive/source/blender/gpu/intern/gpu_vertex_buffer.cc

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

337 lines
7.9 KiB
C++
Raw Normal View History

/* SPDX-License-Identifier: GPL-2.0-or-later
* Copyright 2016 by Mike Erwin. All rights reserved. */
/** \file
* \ingroup gpu
*
* GPU vertex buffer
*/
#include "MEM_guardedalloc.h"
#include "gpu_backend.hh"
#include "gpu_vertex_format_private.h"
#include "gl_vertex_buffer.hh" /* TODO: remove. */
#include "gpu_context_private.hh" /* TODO: remove. */
#include "gpu_vertex_buffer_private.hh"
#include <cstring>
/* -------------------------------------------------------------------- */
/** \name VertBuf
* \{ */
namespace blender::gpu {
size_t VertBuf::memory_usage = 0;
VertBuf::VertBuf()
{
/* Needed by some code check. */
format.attr_len = 0;
}
VertBuf::~VertBuf()
{
/* Should already have been cleared. */
BLI_assert(flag == GPU_VERTBUF_INVALID);
}
void VertBuf::init(const GPUVertFormat *format, GPUUsageType usage)
{
usage_ = usage;
flag = GPU_VERTBUF_DATA_DIRTY;
GPU_vertformat_copy(&this->format, format);
if (!format->packed) {
VertexFormat_pack(&this->format);
}
flag |= GPU_VERTBUF_INIT;
}
void VertBuf::clear()
{
this->release_data();
flag = GPU_VERTBUF_INVALID;
}
VertBuf *VertBuf::duplicate()
{
VertBuf *dst = GPUBackend::get()->vertbuf_alloc();
/* Full copy. */
*dst = *this;
/* Almost full copy... */
dst->handle_refcount_ = 1;
/* Duplicate all needed implementation specifics data. */
this->duplicate_data(dst);
return dst;
}
void VertBuf::allocate(uint vert_len)
{
BLI_assert(format.packed);
/* Catch any unnecessary usage. */
BLI_assert(vertex_alloc != vert_len || data == nullptr);
vertex_len = vertex_alloc = vert_len;
this->acquire_data();
flag |= GPU_VERTBUF_DATA_DIRTY;
}
void VertBuf::resize(uint vert_len)
{
/* Catch any unnecessary usage. */
BLI_assert(vertex_alloc != vert_len);
vertex_len = vertex_alloc = vert_len;
this->resize_data();
flag |= GPU_VERTBUF_DATA_DIRTY;
}
void VertBuf::upload()
{
this->upload_data();
}
} // namespace blender::gpu
/** \} */
/* -------------------------------------------------------------------- */
/** \name C-API
* \{ */
using namespace blender;
using namespace blender::gpu;
/* -------- Creation & deletion -------- */
GPUVertBuf *GPU_vertbuf_calloc()
{
return wrap(GPUBackend::get()->vertbuf_alloc());
}
2018-07-18 23:09:31 +10:00
GPUVertBuf *GPU_vertbuf_create_with_format_ex(const GPUVertFormat *format, GPUUsageType usage)
{
GPUVertBuf *verts = GPU_vertbuf_calloc();
unwrap(verts)->init(format, usage);
return verts;
}
void GPU_vertbuf_init_with_format_ex(GPUVertBuf *verts_,
const GPUVertFormat *format,
GPUUsageType usage)
{
unwrap(verts_)->init(format, usage);
}
OpenSubDiv: add support for an OpenGL evaluator This evaluator is used in order to evaluate subdivision at render time, allowing for faster renders of meshes with a subdivision surface modifier placed at the last position in the modifier list. When evaluating the subsurf modifier, we detect whether we can delegate evaluation to the draw code. If so, the subdivision is first evaluated on the GPU using our own custom evaluator (only the coarse data needs to be initially sent to the GPU), then, buffers for the final `MeshBufferCache` are filled on the GPU using a set of compute shaders. However, some buffers are still filled on the CPU side, if doing so on the GPU is impractical (e.g. the line adjacency buffer used for x-ray, whose logic is hardly GPU compatible). This is done at the mesh buffer extraction level so that the result can be readily used in the various OpenGL engines, without having to write custom geometry or tesselation shaders. We use our own subdivision evaluation shaders, instead of OpenSubDiv's vanilla one, in order to control the data layout, and interpolation. For example, we store vertex colors as compressed 16-bit integers, while OpenSubDiv's default evaluator only work for float types. In order to still access the modified geometry on the CPU side, for use in modifiers or transform operators, a dedicated wrapper type is added `MESH_WRAPPER_TYPE_SUBD`. Subdivision will be lazily evaluated via `BKE_object_get_evaluated_mesh` which will create such a wrapper if possible. If the final subdivision surface is not needed on the CPU side, `BKE_object_get_evaluated_mesh_no_subsurf` should be used. Enabling or disabling GPU subdivision can be done through the user preferences (under Viewport -> Subdivision). See patch description for benchmarks. Reviewed By: campbellbarton, jbakker, fclem, brecht, #eevee_viewport Differential Revision: https://developer.blender.org/D12406
2021-12-27 16:34:47 +01:00
void GPU_vertbuf_init_build_on_device(GPUVertBuf *verts, GPUVertFormat *format, uint v_len)
{
GPU_vertbuf_init_with_format_ex(verts, format, GPU_USAGE_DEVICE_ONLY);
GPU_vertbuf_data_alloc(verts, v_len);
}
GPUVertBuf *GPU_vertbuf_duplicate(GPUVertBuf *verts_)
{
return wrap(unwrap(verts_)->duplicate());
}
const void *GPU_vertbuf_read(GPUVertBuf *verts)
{
return unwrap(verts)->read();
}
void *GPU_vertbuf_unmap(const GPUVertBuf *verts, const void *mapped_data)
{
return unwrap(verts)->unmap(mapped_data);
}
void GPU_vertbuf_clear(GPUVertBuf *verts)
{
unwrap(verts)->clear();
}
void GPU_vertbuf_discard(GPUVertBuf *verts)
{
unwrap(verts)->clear();
unwrap(verts)->reference_remove();
}
void GPU_vertbuf_handle_ref_add(GPUVertBuf *verts)
{
unwrap(verts)->reference_add();
}
void GPU_vertbuf_handle_ref_remove(GPUVertBuf *verts)
{
unwrap(verts)->reference_remove();
}
/* -------- Data update -------- */
void GPU_vertbuf_data_alloc(GPUVertBuf *verts, uint v_len)
{
unwrap(verts)->allocate(v_len);
}
void GPU_vertbuf_data_resize(GPUVertBuf *verts, uint v_len)
{
unwrap(verts)->resize(v_len);
}
void GPU_vertbuf_data_len_set(GPUVertBuf *verts_, uint v_len)
{
VertBuf *verts = unwrap(verts_);
BLI_assert(verts->data != nullptr); /* Only for dynamic data. */
BLI_assert(v_len <= verts->vertex_alloc);
verts->vertex_len = v_len;
}
void GPU_vertbuf_attr_set(GPUVertBuf *verts_, uint a_idx, uint v_idx, const void *data)
{
VertBuf *verts = unwrap(verts_);
2018-07-18 23:09:31 +10:00
const GPUVertFormat *format = &verts->format;
const GPUVertAttr *a = &format->attrs[a_idx];
BLI_assert(v_idx < verts->vertex_alloc);
BLI_assert(a_idx < format->attr_len);
BLI_assert(verts->data != nullptr);
verts->flag |= GPU_VERTBUF_DATA_DIRTY;
memcpy(verts->data + a->offset + v_idx * format->stride, data, a->sz);
}
void GPU_vertbuf_attr_fill(GPUVertBuf *verts_, uint a_idx, const void *data)
{
VertBuf *verts = unwrap(verts_);
2018-07-18 23:09:31 +10:00
const GPUVertFormat *format = &verts->format;
BLI_assert(a_idx < format->attr_len);
const GPUVertAttr *a = &format->attrs[a_idx];
const uint stride = a->sz; /* tightly packed input data */
verts->flag |= GPU_VERTBUF_DATA_DIRTY;
GPU_vertbuf_attr_fill_stride(verts_, a_idx, stride, data);
}
void GPU_vertbuf_vert_set(GPUVertBuf *verts_, uint v_idx, const void *data)
Overlay Engine: Refactor & Cleanup This is the unification of all overlays into one overlay engine as described in T65347. I went over all the code making it more future proof with less hacks and removing old / not relevent parts. Goals / Acheivements: - Remove internal shader usage (only drw shaders) - Remove viewportSize and viewportSizeInv and put them in gloabl ubo - Fixed some drawing issues: Missing probe option and Missing Alt+B clipping of some shader - Remove old (legacy) shaders dependancy (not using view UBO). - Less shader variation (less compilation time at first load and less patching needed for vulkan) - removed some geom shaders when I could - Remove static e_data (except shaders storage where it is OK) - Clear the way to fix some anoying limitations (dithered transparency, background image compositing etc...) - Wireframe drawing now uses the same batching capabilities as workbench & eevee (indirect drawing). - Reduced complexity, removed ~3000 Lines of code in draw (also removed a lot of unused shader in GPU). - Post AA to avoid complexity and cost of MSAA. Remaining issues: - ~~Armature edits, overlay toggles, (... others?) are not refreshing viewport after AA is complete~~ - FXAA is not the best for wires, maybe investigate SMAA - Maybe do something more temporally stable for AA. - ~~Paint overlays are not working with AA.~~ - ~~infront objects are difficult to select.~~ - ~~the infront wires sometimes goes through they solid counterpart (missing clear maybe?) (toggle overlays on-off when using infront+wireframe overlay in solid shading)~~ Note: I made some decision to change slightly the appearance of some objects to simplify their drawing. Namely the empty arrows end (which is now hollow/wire) and distance points of the cameras/spots being done by lines. Reviewed By: jbakker Differential Revision: https://developer.blender.org/D6296
2019-12-02 01:40:58 +01:00
{
VertBuf *verts = unwrap(verts_);
Overlay Engine: Refactor & Cleanup This is the unification of all overlays into one overlay engine as described in T65347. I went over all the code making it more future proof with less hacks and removing old / not relevent parts. Goals / Acheivements: - Remove internal shader usage (only drw shaders) - Remove viewportSize and viewportSizeInv and put them in gloabl ubo - Fixed some drawing issues: Missing probe option and Missing Alt+B clipping of some shader - Remove old (legacy) shaders dependancy (not using view UBO). - Less shader variation (less compilation time at first load and less patching needed for vulkan) - removed some geom shaders when I could - Remove static e_data (except shaders storage where it is OK) - Clear the way to fix some anoying limitations (dithered transparency, background image compositing etc...) - Wireframe drawing now uses the same batching capabilities as workbench & eevee (indirect drawing). - Reduced complexity, removed ~3000 Lines of code in draw (also removed a lot of unused shader in GPU). - Post AA to avoid complexity and cost of MSAA. Remaining issues: - ~~Armature edits, overlay toggles, (... others?) are not refreshing viewport after AA is complete~~ - FXAA is not the best for wires, maybe investigate SMAA - Maybe do something more temporally stable for AA. - ~~Paint overlays are not working with AA.~~ - ~~infront objects are difficult to select.~~ - ~~the infront wires sometimes goes through they solid counterpart (missing clear maybe?) (toggle overlays on-off when using infront+wireframe overlay in solid shading)~~ Note: I made some decision to change slightly the appearance of some objects to simplify their drawing. Namely the empty arrows end (which is now hollow/wire) and distance points of the cameras/spots being done by lines. Reviewed By: jbakker Differential Revision: https://developer.blender.org/D6296
2019-12-02 01:40:58 +01:00
const GPUVertFormat *format = &verts->format;
BLI_assert(v_idx < verts->vertex_alloc);
BLI_assert(verts->data != nullptr);
verts->flag |= GPU_VERTBUF_DATA_DIRTY;
memcpy(verts->data + v_idx * format->stride, data, format->stride);
Overlay Engine: Refactor & Cleanup This is the unification of all overlays into one overlay engine as described in T65347. I went over all the code making it more future proof with less hacks and removing old / not relevent parts. Goals / Acheivements: - Remove internal shader usage (only drw shaders) - Remove viewportSize and viewportSizeInv and put them in gloabl ubo - Fixed some drawing issues: Missing probe option and Missing Alt+B clipping of some shader - Remove old (legacy) shaders dependancy (not using view UBO). - Less shader variation (less compilation time at first load and less patching needed for vulkan) - removed some geom shaders when I could - Remove static e_data (except shaders storage where it is OK) - Clear the way to fix some anoying limitations (dithered transparency, background image compositing etc...) - Wireframe drawing now uses the same batching capabilities as workbench & eevee (indirect drawing). - Reduced complexity, removed ~3000 Lines of code in draw (also removed a lot of unused shader in GPU). - Post AA to avoid complexity and cost of MSAA. Remaining issues: - ~~Armature edits, overlay toggles, (... others?) are not refreshing viewport after AA is complete~~ - FXAA is not the best for wires, maybe investigate SMAA - Maybe do something more temporally stable for AA. - ~~Paint overlays are not working with AA.~~ - ~~infront objects are difficult to select.~~ - ~~the infront wires sometimes goes through they solid counterpart (missing clear maybe?) (toggle overlays on-off when using infront+wireframe overlay in solid shading)~~ Note: I made some decision to change slightly the appearance of some objects to simplify their drawing. Namely the empty arrows end (which is now hollow/wire) and distance points of the cameras/spots being done by lines. Reviewed By: jbakker Differential Revision: https://developer.blender.org/D6296
2019-12-02 01:40:58 +01:00
}
void GPU_vertbuf_attr_fill_stride(GPUVertBuf *verts_, uint a_idx, uint stride, const void *data)
{
VertBuf *verts = unwrap(verts_);
2018-07-18 23:09:31 +10:00
const GPUVertFormat *format = &verts->format;
const GPUVertAttr *a = &format->attrs[a_idx];
BLI_assert(a_idx < format->attr_len);
BLI_assert(verts->data != nullptr);
verts->flag |= GPU_VERTBUF_DATA_DIRTY;
const uint vertex_len = verts->vertex_len;
if (format->attr_len == 1 && stride == format->stride) {
/* we can copy it all at once */
memcpy(verts->data, data, vertex_len * a->sz);
}
else {
/* we must copy it per vertex */
for (uint v = 0; v < vertex_len; v++) {
memcpy(
verts->data + a->offset + v * format->stride, (const uchar *)data + v * stride, a->sz);
}
}
}
void GPU_vertbuf_attr_get_raw_data(GPUVertBuf *verts_, uint a_idx, GPUVertBufRaw *access)
{
VertBuf *verts = unwrap(verts_);
2018-07-18 23:09:31 +10:00
const GPUVertFormat *format = &verts->format;
const GPUVertAttr *a = &format->attrs[a_idx];
BLI_assert(a_idx < format->attr_len);
BLI_assert(verts->data != nullptr);
verts->flag |= GPU_VERTBUF_DATA_DIRTY;
verts->flag &= ~GPU_VERTBUF_DATA_UPLOADED;
access->size = a->sz;
access->stride = format->stride;
access->data = (uchar *)verts->data + a->offset;
access->data_init = access->data;
#ifdef DEBUG
access->_data_end = access->data_init + (size_t)(verts->vertex_alloc * format->stride);
#endif
}
/* -------- Getters -------- */
void *GPU_vertbuf_get_data(const GPUVertBuf *verts)
{
/* TODO: Assert that the format has no padding. */
return unwrap(verts)->data;
}
void *GPU_vertbuf_steal_data(GPUVertBuf *verts_)
{
VertBuf *verts = unwrap(verts_);
/* TODO: Assert that the format has no padding. */
BLI_assert(verts->data);
void *data = verts->data;
verts->data = nullptr;
return data;
}
const GPUVertFormat *GPU_vertbuf_get_format(const GPUVertBuf *verts)
{
return &unwrap(verts)->format;
}
uint GPU_vertbuf_get_vertex_alloc(const GPUVertBuf *verts)
{
return unwrap(verts)->vertex_alloc;
}
uint GPU_vertbuf_get_vertex_len(const GPUVertBuf *verts)
{
return unwrap(verts)->vertex_len;
}
GPUVertBufStatus GPU_vertbuf_get_status(const GPUVertBuf *verts)
{
return unwrap(verts)->flag;
}
OpenSubDiv: add support for an OpenGL evaluator This evaluator is used in order to evaluate subdivision at render time, allowing for faster renders of meshes with a subdivision surface modifier placed at the last position in the modifier list. When evaluating the subsurf modifier, we detect whether we can delegate evaluation to the draw code. If so, the subdivision is first evaluated on the GPU using our own custom evaluator (only the coarse data needs to be initially sent to the GPU), then, buffers for the final `MeshBufferCache` are filled on the GPU using a set of compute shaders. However, some buffers are still filled on the CPU side, if doing so on the GPU is impractical (e.g. the line adjacency buffer used for x-ray, whose logic is hardly GPU compatible). This is done at the mesh buffer extraction level so that the result can be readily used in the various OpenGL engines, without having to write custom geometry or tesselation shaders. We use our own subdivision evaluation shaders, instead of OpenSubDiv's vanilla one, in order to control the data layout, and interpolation. For example, we store vertex colors as compressed 16-bit integers, while OpenSubDiv's default evaluator only work for float types. In order to still access the modified geometry on the CPU side, for use in modifiers or transform operators, a dedicated wrapper type is added `MESH_WRAPPER_TYPE_SUBD`. Subdivision will be lazily evaluated via `BKE_object_get_evaluated_mesh` which will create such a wrapper if possible. If the final subdivision surface is not needed on the CPU side, `BKE_object_get_evaluated_mesh_no_subsurf` should be used. Enabling or disabling GPU subdivision can be done through the user preferences (under Viewport -> Subdivision). See patch description for benchmarks. Reviewed By: campbellbarton, jbakker, fclem, brecht, #eevee_viewport Differential Revision: https://developer.blender.org/D12406
2021-12-27 16:34:47 +01:00
void GPU_vertbuf_tag_dirty(GPUVertBuf *verts)
{
unwrap(verts)->flag |= GPU_VERTBUF_DATA_DIRTY;
}
uint GPU_vertbuf_get_memory_usage()
{
return VertBuf::memory_usage;
}
void GPU_vertbuf_use(GPUVertBuf *verts)
{
unwrap(verts)->upload();
}
OpenSubDiv: add support for an OpenGL evaluator This evaluator is used in order to evaluate subdivision at render time, allowing for faster renders of meshes with a subdivision surface modifier placed at the last position in the modifier list. When evaluating the subsurf modifier, we detect whether we can delegate evaluation to the draw code. If so, the subdivision is first evaluated on the GPU using our own custom evaluator (only the coarse data needs to be initially sent to the GPU), then, buffers for the final `MeshBufferCache` are filled on the GPU using a set of compute shaders. However, some buffers are still filled on the CPU side, if doing so on the GPU is impractical (e.g. the line adjacency buffer used for x-ray, whose logic is hardly GPU compatible). This is done at the mesh buffer extraction level so that the result can be readily used in the various OpenGL engines, without having to write custom geometry or tesselation shaders. We use our own subdivision evaluation shaders, instead of OpenSubDiv's vanilla one, in order to control the data layout, and interpolation. For example, we store vertex colors as compressed 16-bit integers, while OpenSubDiv's default evaluator only work for float types. In order to still access the modified geometry on the CPU side, for use in modifiers or transform operators, a dedicated wrapper type is added `MESH_WRAPPER_TYPE_SUBD`. Subdivision will be lazily evaluated via `BKE_object_get_evaluated_mesh` which will create such a wrapper if possible. If the final subdivision surface is not needed on the CPU side, `BKE_object_get_evaluated_mesh_no_subsurf` should be used. Enabling or disabling GPU subdivision can be done through the user preferences (under Viewport -> Subdivision). See patch description for benchmarks. Reviewed By: campbellbarton, jbakker, fclem, brecht, #eevee_viewport Differential Revision: https://developer.blender.org/D12406
2021-12-27 16:34:47 +01:00
void GPU_vertbuf_wrap_handle(GPUVertBuf *verts, uint64_t handle)
{
unwrap(verts)->wrap_handle(handle);
}
void GPU_vertbuf_bind_as_ssbo(struct GPUVertBuf *verts, int binding)
{
unwrap(verts)->bind_as_ssbo(binding);
}
OpenSubDiv: add support for an OpenGL evaluator This evaluator is used in order to evaluate subdivision at render time, allowing for faster renders of meshes with a subdivision surface modifier placed at the last position in the modifier list. When evaluating the subsurf modifier, we detect whether we can delegate evaluation to the draw code. If so, the subdivision is first evaluated on the GPU using our own custom evaluator (only the coarse data needs to be initially sent to the GPU), then, buffers for the final `MeshBufferCache` are filled on the GPU using a set of compute shaders. However, some buffers are still filled on the CPU side, if doing so on the GPU is impractical (e.g. the line adjacency buffer used for x-ray, whose logic is hardly GPU compatible). This is done at the mesh buffer extraction level so that the result can be readily used in the various OpenGL engines, without having to write custom geometry or tesselation shaders. We use our own subdivision evaluation shaders, instead of OpenSubDiv's vanilla one, in order to control the data layout, and interpolation. For example, we store vertex colors as compressed 16-bit integers, while OpenSubDiv's default evaluator only work for float types. In order to still access the modified geometry on the CPU side, for use in modifiers or transform operators, a dedicated wrapper type is added `MESH_WRAPPER_TYPE_SUBD`. Subdivision will be lazily evaluated via `BKE_object_get_evaluated_mesh` which will create such a wrapper if possible. If the final subdivision surface is not needed on the CPU side, `BKE_object_get_evaluated_mesh_no_subsurf` should be used. Enabling or disabling GPU subdivision can be done through the user preferences (under Viewport -> Subdivision). See patch description for benchmarks. Reviewed By: campbellbarton, jbakker, fclem, brecht, #eevee_viewport Differential Revision: https://developer.blender.org/D12406
2021-12-27 16:34:47 +01:00
void GPU_vertbuf_update_sub(GPUVertBuf *verts, uint start, uint len, const void *data)
{
unwrap(verts)->update_sub(start, len, data);
}
2020-10-02 09:48:41 +10:00
/** \} */