This batches together a couple of micro-optimizations that were discovered in profiling and analyzing disassembly.
Importantly, this reduces the amount of instructions in a heavy loop. This has the biggest impact when there are lots of objects and lights in the scene as the function is called once per objects + once per light that touches the object
Clustered performs the following shadow rendering steps
1. Process objects [0; 10) for cascade 0.
2. Process objects [10; 30) for cascade 1.
3. Process objects [30; 100) for cascade 2.
4. Upload objects [0; 100) to GPU.
5. Draw all cascades.
Mobile was supposed to be doing the same, but instead was doing:
1. Process objects [0; 10) for cascade 0.
2. Upload objects [0; 10) to GPU.
3. Process objects [10; 30) for cascade 1.
4. Upload objects [0; 30) to GPU.
5. Process objects [30; 100) for cascade 2.
6. Upload objects [0; 100) to GPU.
7. Draw all cascades.
That is, always reuploaded everything from scratch.
Therefore it pointlessly (and with geometric growth) wasted BW.
Clustered performs the following shadow rendering steps
1. Process objects [0; 10) for cascade 0.
2. Process objects [10; 30) for cascade 1.
3. Process objects [30; 100) for cascade 2.
4. Upload objects [0; 100) to GPU.
5. Draw all cascades.
Mobile was supposed to be doing the same, but instead was doing:
1. Process objects [0; 10) for cascade 0.
2. Upload objects [0; 10) to GPU.
3. Process objects [10; 30) for cascade 1.
4. Upload objects [0; 30) to GPU.
5. Process objects [30; 100) for cascade 2.
6. Upload objects [0; 100) to GPU.
7. Draw all cascades.
That is, always reuploaded everything from scratch.
Therefore it pointlessly (and with geometric growth) wasted BW.
This is a followup to PR #101344 (commit
0e06eb80bc).
Some of them were not an issue because Godot was initializing all
members, but they were "fixed" just in case since it could become a
problem in the future.
Valgrind was specifically complaining about HashMapData &
GlobalPipelineData.
The work was performed by collaboration of TheForge and Google. I am
merely splitting it up into smaller PRs and cleaning it up.
This is the most "risky" PR so far because the previous ones have been
miscellaneous stuff aimed at either [improve
debugging](https://github.com/godotengine/godot/pull/90993) (e.g. device
lost), [improve Android
experience](https://github.com/godotengine/godot/pull/96439) (add Swappy
for better Frame Pacing + Pre-Transformed Swapchains for slightly better
performance), or harmless [ASTC
improvements](https://github.com/godotengine/godot/pull/96045) (better
performance by simply toggling a feature when available).
However this PR contains larger modifications aimed at improving
performance or reducing memory fragmentation. With greater
modifications, come greater risks of bugs or breakage.
Changes introduced by this PR:
TBDR GPUs (e.g. most of Android + iOS + M1 Apple) support rendering to
Render Targets that are not backed by actual GPU memory (everything
stays in cache). This works as long as load action isn't `LOAD`, and
store action must be `DONT_CARE`. This saves VRAM (it also makes
painfully obvious when a mistake introduces a performance regression).
Of particular usefulness is when doing MSAA and keeping the raw MSAA
content is not necessary.
Some GPUs get faster when the sampler settings are hard-coded into the
GLSL shaders (instead of being dynamically bound at runtime). This
required changes to the GLSL shaders, PSO creation routines, Descriptor
creation routines, and Descriptor binding routines.
- `bool immutable_samplers_enabled = true`
Setting it to false enforces the old behavior. Useful for debugging bugs
and regressions.
Immutable samplers requires that the samplers stay... immutable, hence
this boolean is useful if the promise gets broken. We might want to turn
this into a `GLOBAL_DEF` setting.
Instead of creating dozen/hundreds/thousands of `VkDescriptorSet` every
frame that need to be freed individually when they are no longer needed,
they all get freed at once by resetting the whole pool. Once the whole
pool is no longer in use by the GPU, it gets reset and its memory
recycled. Descriptor sets that are created to be kept around for longer
or forever (i.e. not created and freed within the same frame) **must
not** use linear pools. There may be more than one pool per frame. How
many pools per frame Godot ends up with depends on its capacity, and
that is controlled by
`rendering/rendering_device/vulkan/max_descriptors_per_pool`.
- **Possible improvement for later:** It should be possible for Godot
to adapt to how many descriptors per pool are needed on a per-key basis
(i.e. grow their capacity like `std::vector` does) after rendering a few
frames; which would be better than the current solution of having a
single global value for all pools (`max_descriptors_per_pool`) that the
user needs to tweak.
- `bool linear_descriptor_pools_enabled = true`
Setting it to false enforces the old behavior. Useful for debugging bugs
and regressions.
Setting it to false is required when workarounding driver bugs (e.g.
Adreno 730).
A ridiculous optimization. Ridiculous because the original code
should've done this in the first place. Previously Godot was doing the
following:
1. Create a command buffer **pool**. One per frame.
2. Create multiple command buffers from the pool in point 1.
3. Call `vkBeginCommandBuffer` on the cmd buffer in point 2. This
resets the cmd buffer because Godot requests the
`VK_COMMAND_POOL_CREATE_RESET_COMMAND_BUFFER_BIT` flag.
4. Add commands to the cmd buffers from point 2.
5. Submit those commands.
6. On frame N + 2, recycle the buffer pool and cmd buffers from pt 1 &
2, and repeat from step 3.
The problem here is that step 3 resets each command buffer individually.
Initially Godot used to have 1 cmd buffer per pool, thus the impact is
very low.
But not anymore (specially with Adreno workarounds to force splitting
compute dispatches into a new cmd buffer, more on this later). However
Godot keeps around a very low amount of command buffers per frame.
The recommended method is to reset the whole pool, to reset all cmd
buffers at once. Hence the new steps would be:
1. Create a command buffer **pool**. One per frame.
2. Create multiple command buffers from the pool in point 1.
3. Call `vkBeginCommandBuffer` on the cmd buffer in point 2, which is
already reset/empty (see step 6).
4. Add commands to the cmd buffers from point 2.
5. Submit those commands.
6. On frame N + 2, recycle the buffer pool and cmd buffers from pt 1 &
2, call `vkResetCommandPool` and repeat from step 3.
**Possible issues:** @dariosamo added `transfer_worker` which creates a
command buffer pool:
```cpp
transfer_worker->command_pool =
driver->command_pool_create(transfer_queue_family,
RDD::COMMAND_BUFFER_TYPE_PRIMARY);
```
As expected, validation was complaining that command buffers were being
reused without being reset (that's good, we now know Validation Layers
will warn us of wrong use).
I fixed it by adding:
```cpp
void RenderingDevice::_wait_for_transfer_worker(TransferWorker
*p_transfer_worker) {
driver->fence_wait(p_transfer_worker->command_fence);
driver->command_pool_reset(p_transfer_worker->command_pool); //
! New line !
```
**Secondary cmd buffers are subject to the same issue but I didn't alter
them. I talked this with Dario and he is aware of this.**
Secondary cmd buffers are currently disabled due to other issues (it's
disabled on master).
- `bool RenderingDeviceCommons::command_pool_reset_enabled`
Setting it to false enforces the old behavior. Useful for debugging bugs
and regressions.
There's no other reason for this boolean. Possibly once it becomes well
tested, the boolean could be removed entirely.
Adds `command_bind_render_uniform_sets` and
`add_draw_list_bind_uniform_sets` (+ compute variants).
It performs the same as `add_draw_list_bind_uniform_set` (notice
singular vs plural), but on multiple consecutive uniform sets, thus
reducing graph and draw call overhead.
- `bool descriptor_set_batching = true;`
Setting it to false enforces the old behavior. Useful for debugging bugs
and regressions.
There's no other reason for this boolean. Possibly once it becomes well
tested, the boolean could be removed entirely.
Godot currently does the following:
1. Fill the entire cmd buffer with commands.
2. `submit()`
- Wait with a semaphore for the swapchain.
- Trigger a semaphore to indicate when we're done (so the swapchain
can submit).
3. `present()`
The optimization opportunity here is that 95% of Godot's rendering is
done offscreen.
Then a fullscreen pass copies everything to the swapchain. Godot doesn't
practically render directly to the swapchain.
The problem with this is that the GPU has to wait for the swapchain to
be released **to start anything**, when we could start *much earlier*.
Only the final blit pass must wait for the swapchain.
TheForge changed it to the following (more complicated, I'm simplifying
the idea):
1. Fill the entire cmd buffer with commands.
2. In `screen_prepare_for_drawing` do `submit()`
- There are no semaphore waits for the swapchain.
- Trigger a semaphore to indicate when we're done.
3. Fill a new cmd buffer that only does the final blit to the
swapchain.
4. `submit()`
- Wait with a semaphore for the submit() from step 2.
- Wait with a semaphore for the swapchain (so the swapchain can
submit).
- Trigger a semaphore to indicate when we're done (so the swapchain
can submit).
5. `present()`
Dario discovered this problem independently while working on a different
platform.
**However TheForge's solution had to be rewritten from scratch:** The
complexity to achieve the solution was high and quite difficult to
maintain with the way Godot works now (after Übershaders PR).
But on the other hand, re-implementing the solution became much simpler
because Dario already had to do something similar: To fix an Adreno 730
driver bug, he had to implement splitting command buffers. **This is
exactly what we need!**. Thus it was re-written using this existing
functionality for a new purpose.
To achieve this, I added a new argument, `bool p_split_cmd_buffer`, to
`RenderingDeviceGraph::add_draw_list_begin`, which is only set to true
by `RenderingDevice::draw_list_begin_for_screen`.
The graph will split the draw list into its own command buffer.
- `bool split_swapchain_into_its_own_cmd_buffer = true;`
Setting it to false enforces the old behavior. This might be necessary
for consoles which follow an alternate solution to the same problem.
If not, then we should consider removing it.
PR #90993 added `shader_destroy_modules()` but it was not actually in
use.
This PR adds several places where `shader_destroy_modules()` is called
after initialization to free up memory of SPIR-V structures that are no
longer needed.
- Implements asynchronous transfer queues from PR #87590.
- Adds ubershaders that can run with specialization constants specified as push constants.
- Pipelines with specialization constants can compile in the background.
- Added monitoring for pipeline compilations.
- Materials and shaders can now be created asynchronously on background threads.
- Meshes that are loaded on background threads can also compile pipelines as part of the loading process.
This adds support in all backends, but the Compatibility renderer works the best.
Mobile and Forward+ can only support one directional light shader (the first in the tree)
While the Compatibility renderer supports any number of shadows.
Co-authored-by: Clay John <claynjohn@gmail.com>
