Files
test/source/blender/gpu/intern/gpu_state_private.hh
Omar Emara ff3b2226fb GPU: Refactor texture samplers
This patch refactors the texture samples code by mainly splitting the
eGPUSamplerState enum into multiple smaller enums and packing them
inside a GPUSamplerState struct. This was done because many members of
the enum were mutually exclusive, which was worked around during setting
up the samplers in the various backends, and additionally made the API
confusing, like the GPU_texture_wrap_mode function, which had two
mutually exclusive parameters.

The new structure also improved and clarified the backend sampler cache,
reducing the cache size from 514 samplers to just 130 samplers, which
also slightly improved the initialization time. Further, the
GPU_SAMPLER_MAX signal value was naturally incorporated into the
structure using the GPU_SAMPLER_STATE_TYPE_INTERNAL type.

The only expected functional change is in the realtime compositor, which
now supports per-axis repetition control, utilizing new API functions
for that purpose.

This patch is loosely based on an older patch D14366 by Ethan Hall.

Pull Request: https://projects.blender.org/blender/blender/pulls/105642
2023-04-04 15:16:07 +02:00

194 lines
4.4 KiB
C++

/* SPDX-License-Identifier: GPL-2.0-or-later
* Copyright 2020 Blender Foundation. */
/** \file
* \ingroup gpu
*/
#pragma once
#include "BLI_utildefines.h"
#include "GPU_state.h"
#include "gpu_texture_private.hh"
#include <cstring>
namespace blender {
namespace gpu {
/* Encapsulate all pipeline state that we need to track.
* Try to keep small to reduce validation time. */
union GPUState {
struct {
/** eGPUWriteMask */
uint32_t write_mask : 13;
/** eGPUBlend */
uint32_t blend : 4;
/** eGPUFaceCullTest */
uint32_t culling_test : 2;
/** eGPUDepthTest */
uint32_t depth_test : 3;
/** eGPUStencilTest */
uint32_t stencil_test : 3;
/** eGPUStencilOp */
uint32_t stencil_op : 3;
/** eGPUProvokingVertex */
uint32_t provoking_vert : 1;
/** Enable bits. */
uint32_t logic_op_xor : 1;
uint32_t invert_facing : 1;
uint32_t shadow_bias : 1;
/** Number of clip distances enabled. */
/* TODO(fclem): This should be a shader property. */
uint32_t clip_distances : 3;
/* TODO(fclem): remove, old opengl features. */
uint32_t polygon_smooth : 1;
uint32_t line_smooth : 1;
};
/* Here to allow fast bit-wise ops. */
uint64_t data;
};
BLI_STATIC_ASSERT(sizeof(GPUState) == sizeof(uint64_t), "GPUState is too big.");
inline bool operator==(const GPUState &a, const GPUState &b)
{
return a.data == b.data;
}
inline bool operator!=(const GPUState &a, const GPUState &b)
{
return !(a == b);
}
inline GPUState operator^(const GPUState &a, const GPUState &b)
{
GPUState r;
r.data = a.data ^ b.data;
return r;
}
inline GPUState operator~(const GPUState &a)
{
GPUState r;
r.data = ~a.data;
return r;
}
/* Mutable state that does not require pipeline change. */
union GPUStateMutable {
struct {
/* Viewport State */
/** TODO: remove. */
float depth_range[2];
/** Positive if using program point size. */
/* TODO(fclem): should be passed as uniform to all shaders. */
float point_size;
/** Not supported on every platform. Prefer using wide-line shader. */
float line_width;
/** Mutable stencil states. */
uint8_t stencil_write_mask;
uint8_t stencil_compare_mask;
uint8_t stencil_reference;
uint8_t _pad0;
/* IMPORTANT: ensure x64 struct alignment. */
};
/* Here to allow fast bit-wise ops. */
uint64_t data[9];
};
BLI_STATIC_ASSERT(sizeof(GPUStateMutable) == sizeof(GPUStateMutable::data),
"GPUStateMutable is too big.");
inline bool operator==(const GPUStateMutable &a, const GPUStateMutable &b)
{
return memcmp(&a, &b, sizeof(GPUStateMutable)) == 0;
}
inline bool operator!=(const GPUStateMutable &a, const GPUStateMutable &b)
{
return !(a == b);
}
inline GPUStateMutable operator^(const GPUStateMutable &a, const GPUStateMutable &b)
{
GPUStateMutable r;
for (int i = 0; i < ARRAY_SIZE(a.data); i++) {
r.data[i] = a.data[i] ^ b.data[i];
}
return r;
}
inline GPUStateMutable operator~(const GPUStateMutable &a)
{
GPUStateMutable r;
for (int i = 0; i < ARRAY_SIZE(a.data); i++) {
r.data[i] = ~a.data[i];
}
return r;
}
/**
* State manager keeping track of the draw state and applying it before drawing.
* Base class which is then specialized for each implementation (GL, VK, ...).
*/
class StateManager {
public:
GPUState state;
GPUStateMutable mutable_state;
bool use_bgl = false;
public:
StateManager();
virtual ~StateManager(){};
virtual void apply_state() = 0;
virtual void force_state() = 0;
virtual void issue_barrier(eGPUBarrier barrier_bits) = 0;
virtual void texture_bind(Texture *tex, GPUSamplerState sampler, int unit) = 0;
virtual void texture_unbind(Texture *tex) = 0;
virtual void texture_unbind_all() = 0;
virtual void image_bind(Texture *tex, int unit) = 0;
virtual void image_unbind(Texture *tex) = 0;
virtual void image_unbind_all() = 0;
virtual void texture_unpack_row_length_set(uint len) = 0;
};
/**
* GPUFence.
*/
class Fence {
protected:
bool signalled_ = false;
public:
Fence(){};
virtual ~Fence(){};
virtual void signal() = 0;
virtual void wait() = 0;
};
/* Syntactic sugar. */
static inline GPUFence *wrap(Fence *pixbuf)
{
return reinterpret_cast<GPUFence *>(pixbuf);
}
static inline Fence *unwrap(GPUFence *pixbuf)
{
return reinterpret_cast<Fence *>(pixbuf);
}
static inline const Fence *unwrap(const GPUFence *pixbuf)
{
return reinterpret_cast<const Fence *>(pixbuf);
}
} // namespace gpu
} // namespace blender