Together with:
* CommandBuffer::draw()
* Support for indexed and non-indexed meshes
* Support for setting primitive and stride dynamically
I took one shortcut and vkCmdBindVertexBuffers() is currently called
once for each binding. The interface is ready for this, but I'm not yet
100% sure how to test that it actually does batch the buffers, so it's
left at the lazy implementation for now.
I named it RasterizationPipelineCreateInfo and not
GraphicsPipelineCreateInfo because there's now a
RayTracingPipelineCreateInfo as well, which is *also* graphics, and it
would be confusing for everyone except people already drowned in Vulkan
naming quirks.
Next up is *the unthinkable*, a Vk::Mesh. After that I'll finally have
enough APIs exposed to document everything including command buffer
recording and submission.
These were implemented long before ArrayTuple was a thing and while the
allocation saving optimization makes sense, the implementation is way
too error prone.
This also means Array (and thus <new> and other heavy shit) isn't
included in the headers, which leads to further slimming of compile
times across the library.
Since depth/stencil images can't be linear, I needed buffer/image copies
to test those, and conversely to test buffer/image copies I needed image
clears.
A pretty big chunk of work, and it led to a discovery of a SwiftShader
bug, which I will work around next. First Vulkan driver workaround, so
the whole scaffolding needs to get added as well.
Because (after I read a bunch of articles on this) it can actually be
understood easier that way -- the now-documented snippet shows that in
practice. Also expanded the constructor and usage docs a bit.
Originally I wanted to save people typing and provide "reasonable
defaults" for the image layout, however shortly after I realized I can't
express even the simplest case in the Vulkan Triangle example with
those.
Furthermore the case of having default load/store operations are so
rare that it isn't worth an extra API which tricks the user into
thinking the clear/discard operations are meant to be done somewhere
else. I realized this because it was actually harder to explain these in
a second step than introducing them right from the beginning.
Finally, because there's so many different structures to fill, the
implicit constructors weren't a good idea either, as it again tricked
users to think just PixelFormat or just an index is the parameter, with
nothing else, and then left them wondering where all the other important
params go. Same with {} used instead AttachmentLoadOperation::Load /
AttachmentStoreOperation::Store, unnecessary confusion there so don't
even suggest that.
I was slowly getting cancer from having to write the unreadably awful
VK_FORMAT_R666G666B666A666_SRGB all the time. Besides that:
- All pixel formats are documented to show what's guaranteed for them
by the spec. Pretty useful I'd say.
- The old hasVkFormat() and vkFormat() converters operating on a
VkFormat are deprecated in favor of new hasPixelFormat() and
pixelFormat() that use the PixelFormat enum. Similarly as done in the
GL wrapper.
- All APIs that took a VkFormat before take a PixelFormat now, together
with having conveinience overloads for Magnum::PixelFormat and
Magnum::CompressedPixelFormat. Again similarly as done in the GL
wrapper, also the first step on being able to *directly* use data
imported with the Trade library with Vulkan.
There was a nasty issue when moving the Device -- because the Queue
instances are populated with the Device pointer in the constructor,
moving the Device makes the Queue device pointer invalid and those
instances are useless. Discovered this when using VulkanTester to test
an upcoming Queue::submit() -- until then, the Device pointer in the
Queue wasn't really used for anything and so this wasn't known.
Making the Device immovable is thus the only sane way to keep the
current Queue retrieval workflow, which I'd say is much cleaner than
having to manually query the device for queues later using some
error-prone indices and whatnot. On the other hand, this finally makes
it possible to have a failable device creation, instead of the device
creation failing on an assert (or failing silently when
CORRADE_NO_ASSERT is defined). This is consistent with how the GL
wrapper works.
Also, because all device-dependent objects need to keep a pointer to the
originating Device in order to access Vulkan function pointers, having
it immovable makes it considerably faster. I'll make Instance immovable
with tryCreate() next because it seems like a good thing to do there as
well.
This was way more pain that initially expected, especially in regards to
preserving externally-specified pNext chains without writing to them in
any way.
After I implemented the render pass wrapper, seeing how the
RenderPassCreateInfo structure and its dependencies were HUGE compared
to the actual tiny and lean RenderPass, I felt uneasy dragging their
definition along to every place where a RenderPass gets used. It's not
as bad with the others, but as new extensions are implemented I expect
that to get the same.
This change makes it easier for me to accept that Image.h / Buffer.h
depends on Memory.h. There isn't a real measurable difference when
building Magnum itself (50 ms out of 7 seconds for the Vk library
alone), but that's because most of the code (and tests) needs the
CreateInfo structures anyway.
Although the APIs don't look like that right now, in many cases creation
of a particular Vulkan object isn't all that's there for it. So change
the section names from a generic "Usage" to "Creation".
Quite a big chunk of work, further expanded due to how
VK_KHR_create_renderpass2 is designed -- basically, due to the
tightly-packed nested structures that got replaced with their "version
2", we can no longer just extract the previous structure for backwards
compatibility, but instead have to deep-copy everything to a newly
allocated memory.
Thanks to the the new ArrayTuple structure and a few design iterations I
managed to kick the backwards-compatiblity code into just a single
allocation, while still keeping it possible for the "version 2" code
path to be fully allocation-free (if one passes a completely filled
VkRenderPassCreateInfo2 structure there).
Thus it has no place in the general overview docs of Vk::Instance and
Vk::Device. Better to put it into the VUlkan wrapping docs, where it
also makes sense to have both the device and instance side together.
Makes it possible to write Vk::Instance instance{{argc, argv}} which is
a good tradeoff between passing no arguments at all and doing the fully
verbose thing.
More convenient to use since one doesn't have to explicitly store a
DeviceProperties instance to call pickQueueFamily() on it and then move
it to DeviceCreateInfo to keep it efficient.
This required a slight redesign of how DeviceProperties are stored in
DeviceCreateInfo, now we need the instance to be always valid (but it
can get never used). The wrap() API isn't doing any extra work so this
won't add any inefficiency.
I have more and more cases where I need to query device properties later
down the road (memory capabilities, device name, ...) and leaving all
this up to the user / making this impossible to do in the library
internals is complicating everything too much.
Since there's a shitton of device properties with a new bag of props
coming with every other new extension, I expect the queries to get quite
involved / complicated over time (chaining 100s structs and such), so
let's design this upfront in a way that can avoid reqpeatedly querying
the same thing just because we needlessly discarded a fully populated
instance before.
It also means the users don't need to drag their own DeviceProperties
instance along anymore and can just let the Device take care of that.
Unfortunately the only nice way to make this work with DeviceCreateInfo
method chaining is to add & and && overloads for each. But it's quite
easy to test that all of them work and properly return a r-value
reference so it shouldn't be too much of a maintenance nightmare.
Vladimír Vondruš