We have been interchanging spv and SPIRV_Cross_ for a while, which
causes weirdness since we don't explicitly ban SPIRV_Cross identifiers,
as these identifiers are generally used for interface variable
workarounds.
(GL_EXT_nonuniform_qualifier/SPV_EXT_descriptor_indexing).
MSLResourceBinding includes array size through API, and substitutes
in that size if the image or sampler array is not explicitly sized.
OpCopyObject supports SPIRCombinedImageSampler type in MSL.
`half` cannot be bitcasted to `float`, because the two types are not the
same size. Use an expanding cast instead.
We were already doing this for stores to the tessellation levels; why I
didn't also do this for loads is beyond me.
In some cases, we need to get a literal value from a spec constant op.
Mostly relevant when emitting buffers, so implement a 32-bit integer
scalar subset of the evaluator. Can be extended as needed to support
evaluating any specialization constant operation.
- Do not silently drop reserved identifiers in the parser. This makes it
possible to reflect identifiers which are reserved by the
cross-compiler module.
- Instead of dropping the name, emit _RESERVED_IDENTIFIER_FIXUP in the
source to make it clear that a name has been rewritten.
- Document what is reserved and not.
To facilitate an improved linking-by-name use case for older GL,
we will be more aggressive about merging struct definitions, even for
rather unrelated cases where we don't strictly need to use type aliases.
On MSL, the compiler refuses to allow access chains into a normal vector type.
What happens in practice instead is a read-modify-write where a vector type is
loaded, modified and written back.
The workaround is to convert a vector into a pointer-to-scalar before
the access chain continues to add the scalar index.
When inside a loop, treat any read of outer expressions to happen
multiple times, forcing a temporary of said outer expressions.
This avoids the problem where we can end up relying on loop-invariant code motion to happen in the
compiler when converting optimized shaders.
When we see a switch block which only contains one default block, emit a
do {} while(false) statement instead, which is far more idiomatic and
readable anyways.
When loading and storing array types which belong to buffer objects, we
need to treat these values as not being value types. Also, need to
handle array load/store from/to more address space combinations.
Metal is picky about interface matching. If the types don't match
exactly, down to the number of vector components, Metal fails pipline
compilation. To support pipelines where the number of components
consumed by the fragment shader is less than that produced by the vertex
shader, we have to fix up the fragment shader to accept all the
components produced.
Drop the nonuniform_qualifier in GLSL downcompile NV_gpu_shader5 supports divergent indexing transparently, but obviously runtime sized descriptor arrays are not supported by anything in OpenGL.
Do not attempt to defer declaration. It would happen to work in most
cases, but the edge case is where the first thing that happens to a
variable is being OpStore'd into.
- Fixes issue with clip_distance flattening in MSL where member to
flatten from would come from to_member_name, where it should have used
the builtin name directly. This member name was modified by this patch
and broke clip distance test shaders.
- Some cleanups with ir.meta, use ir.find_meta instead to not create
unnecessary hashmap nodes.
This is implied in both GL and GLES. Emitting memoryBarrierShared() was
based on earlier confusion in the spec which has since been fixed and
clarified.
We had a case where loops were marked complex in a cascading fashion
where each loop iteration would discover one new complex loop. This was
a problem with three nested loops.
This will be awkward to report in GLSL where we check multiple packing
standards, but for HLSL it should be easy since there's only CBuffer
packing standard to worry about.
There is no direct way to express this, so invert boolean results to
force any NaN -> true. glslang emits Ordered compare instructions
everywhere, and the GLSL spec is not clear on this, so assume this is
fine.
It is possible for a shader to declare two plain struct types which
simply share the same OpName without there being an implicit
value/buffer alias relationship.
For to_member_name(), make sure to use the type alias master when
resolving member names. The member name may be different in a type alias
master if the SPIR-V is being intentionally difficult.
First, when generating from HLSL before invoking the code that comes from the HLSL patch-function a control-flow and full memory-barrier are required to ensure that all the temporary values in thread-local storage for the patch are available.
Second, the inputs to control and evaluation shaders must be properly forwarded from the global variables in SPIRV to the member variables in the relevant input structure.
Finally when arrays of interpolators are used for input or output we need to add an extra level of array indirection because Metal works at a different granularity than SPIRV.
Five parts.
1. Fix tessellation patch function processing.
2. Fix loads from tessellation control inputs not being forwarded to the gl_in structure array.
3. Fix loads from tessellation evaluation inputs not being forwarded to the stage_in structure array.
4. Workaround SPIRV losing an array indirection in tessellation shaders - not the best solution but enough to keep things progressing.
5. Apparently gl_TessLevelInner/Outer is special and needs to not be placed into the input array.
Some fallout where internal functions are using stronger types.
Overkill to move everything over to strong types right now, but perhaps
move over to it slowly over time.
Vulkan has two types of buffer descriptors,
`VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER_DYNAMIC` and
`VK_DESCRIPTOR_TYPE_STORAGE_BUFFER_DYNAMIC`, which allow the client to
offset the buffers by an amount given when the descriptor set is bound
to a pipeline. Metal provides no direct support for this when the buffer
in question is in an argument buffer, so once again we're on our own.
These offsets cannot be stored or associated in any way with the
argument buffer itself, because they are set at bind time. Different
pipelines may have different offsets set. Therefore, we must use a
separate buffer, not in any argument buffer, to hold these offsets. Then
the shader must manually offset the buffer pointer.
This change fully supports arrays, including arrays of arrays, even
though Vulkan forbids them. It does not, however, support runtime
arrays. Perhaps later.
This was straightforward to implement in GLSL. The
`ShadingRateInterlockOrderedEXT` and `ShadingRateInterlockUnorderedEXT`
modes aren't implemented yet, because we don't support
`SPV_NV_shading_rate` or `SPV_EXT_fragment_invocation_density` yet.
HLSL and MSL were more interesting. They don't support this directly,
but they do support marking resources as "rasterizer ordered," which
does roughly the same thing. So this implementation scans all accesses
inside the critical section and marks all storage resources found
therein as rasterizer ordered. They also don't support the fine-grained
controls on pixel- vs. sample-level interlock and disabling ordering
guarantees that GLSL and SPIR-V do, but that's OK. "Unordered" here
merely means the order is undefined; that it just so happens to be the
same as rasterizer order is immaterial. As for pixel- vs. sample-level
interlock, Vulkan explicitly states:
> With sample shading enabled, [the `PixelInterlockOrderedEXT` and
> `PixelInterlockUnorderedEXT`] execution modes are treated like
> `SampleInterlockOrderedEXT` or `SampleInterlockUnorderedEXT`
> respectively.
and:
> If [the `SampleInterlockOrderedEXT` or `SampleInterlockUnorderedEXT`]
> execution modes are used in single-sample mode they are treated like
> `PixelInterlockOrderedEXT` or `PixelInterlockUnorderedEXT`
> respectively.
So this will DTRT for MoltenVK and gfx-rs, at least.
MSL additionally supports multiple raster order groups; resources that
are not accessed together can be placed in different ROGs to allow them
to be synchronized separately. A more sophisticated analysis might be
able to place resources optimally, but that's outside the scope of this
change. For now, we assign all resources to group 0, which should do for
our purposes.
`glslang` doesn't support the `RasterizerOrdered` UAVs this
implementation produces for HLSL, so the test case needs `fxc.exe`.
It also insists on GLSL 4.50 for `GL_ARB_fragment_shader_interlock`,
even though the spec says it needs either 4.20 or
`GL_ARB_shader_image_load_store`; and it doesn't support the
`GL_NV_fragment_shader_interlock` extension at all. So I haven't been
able to test those code paths.
Fixes#1002.
This change introduces functions and in one case, a class, to support
the `VK_KHR_sampler_ycbcr_conversion` extension. Except in the case of
GBGR8 and BGRG8 formats, for which Metal natively supports implicit
chroma reconstruction, we're on our own here. We have to do everything
ourselves. Much of the complexity comes from the need to support
multiple planes, which must now be passed to functions that use the
corresponding combined image-samplers. The rest is from the actual
Y'CbCr conversion itself, which requires additional post-processing of
the sample retrieved from the image.
Passing sampled images to a function was a particular problem. To
support this, I've added a new class which is emitted to MSL shaders
that pass sampled images with Y'CbCr conversions attached around. It
can handle sampled images with or without Y'CbCr conversion. This is an
awful abomination that should not exist, but I'm worried that there's
some shader out there which does this. This support requires Metal 2.0
to work properly, because it uses default-constructed texture objects,
which were only added in MSL 2. I'm not even going to get into arrays of
combined image-samplers--that's a whole other can of worms. They are
deliberately unsupported in this change.
I've taken the liberty of refactoring the support for texture swizzling
while I'm at it. It's now treated as a post-processing step similar to
Y'CbCr conversion. I'd like to think this is cleaner than having
everything in `to_function_name()`/`to_function_args()`. It still looks
really hairy, though. I did, however, get rid of the explicit type
arguments to `spvGatherSwizzle()`/`spvGatherCompareSwizzle()`.
Update the C API. In addition to supporting this new functionality, add
some compiler options that I added in previous changes, but for which I
neglected to update the C API.
ESSL does not support `GL_ARB_post_depth_coverage`. There, we must use
`GL_EXT_post_depth_coverage`. I've added this as a fallback for desktop
as well.
Note that `GL_EXT_post_depth_coverage` also requires the fragment shader
to set `early_fragment_tests` explicitly, while
`GL_ARB_post_depth_coverage` does not. It doesn't really matter either
way, since `SPV_KHR_post_depth_coverage` *also* requires both execution
modes to be explicitly set.
We were going down a tree of expressions multiple times and this caused
an exponential explosion in time, which was not caught until recently.
Fix this by blocking any traversal going through an ID more than one
time.
This fix overall improves performance by almost an order of magnitude on a
particular test shader rather than slowing it down by ~75x.
This is not necessary, as we must emit an invalidating store before we
potentially consume an invalid expression. In fact, we're a bit
conservative here in this case for example:
int tmp = variable;
if (...)
{
variable = 10;
}
else
{
// Consuming tmp here is fine, but it was
// invalidated while emitting other branch.
// Technically, we need to study if there is an invalidating store
// in the CFG between the loading block and this block, and the other
// branch will not be a part of that analysis.
int tmp2 = tmp * tmp;
}
Fixing this case means complex CFG traversal *everywhere*, and it feels like overkill.
Fixing this exposed a bug with access chains, so fix a bug where expression dependencies were not
inherited properly in access chains. Access chains are now considered forwarded if there
is at least one dependency which is also forwarded.
This subtle bug removed any expression validation for trivially swizzled
variables. Make usage suppression a more explicit concept rather than
just hacking off forwarded_temporaries.
There is some fallout here with loop generation since our expression
invalidation is currently a bit too naive to handle loops properly.
The forwarding bug masked this problem until now.
If part of the loop condition is also used in the body, we end up
reading an invalid expression, which in turn forces a temporary to be
generated in the condition block, not good. We'll need to be smarter
here ...
If this is computed *before* a `demote`, but used *after*, forwarding it
will produce the wrong value. This does make for uglier shaders, but
it's necessary right now to ensure correctness.
I needed to use an assembly shader to produce the test for this.
`spirv-opt` is not smart enough (or too smart?) to eliminate the
variable that would be used in GLSL to express this.
This extension provides a new operation which causes a fragment to be
discarded without terminating the fragment shader invocation. The
invocation for the discarded fragment becomes a helper invocation, so
that derivatives will remain defined. The old `HelperInvocation` builtin
becomes undefined when this occurs, so a second new instruction queries
the current helper invocation status.
This is only fully supported for GLSL. HLSL doesn't support the
`IsHelperInvocation` operation and MSL doesn't support the
`DemoteToHelperInvocation` op.
Fixes#1052.
The only piece added by this extension is the `DeviceIndex` builtin,
which tells the shader which device in a grouped logical device it is
running on.
Metal's pipeline state objects are owned by the `MTLDevice` that created
them. Since Metal doesn't support logical grouping of devices the way
Vulkan does, we'll thus have to create a pipeline state for each device
in a grouped logical device. The upcoming peer group support in Metal 3
will not change this. For this reason, for Metal, the device index is
supplied as a constant at pipeline compile time.
There's an interaction between `VK_KHR_device_group` and
`VK_KHR_multiview` in the
`VK_PIPELINE_CREATE_VIEW_INDEX_FROM_DEVICE_INDEX_BIT`, which defines the
view index to be the same as the device index. The new
`view_index_from_device_index` MSL option supports this functionality.
Using the `PostDepthCoverage` mode specifies that the `gl_SampleMaskIn`
variable is to contain the computed coverage mask following the early
fragment tests, which this mode requires and implicitly enables.
Note that unlike Vulkan and OpenGL, Metal places this on the sample mask
input itself, and furthermore does *not* implicitly enable early
fragment testing. If it isn't enabled explicitly with an
`[[early_fragment_tests]]` attribute, the compiler will error out. So we
have to enable that mode explicitly if `PostDepthCoverage` is enabled
but `EarlyFragmentTests` isn't.
For Metal, only iOS supports this; for some reason, Apple has yet to
implement it on macOS, even though many desktop cards support it.
There is a case where we can deduce a for/while loop, but the continue
block is actually very painful to deal with, so handle that case as
well. Removes an exceptional case.
This decoration might only be present for the very last ID which is
consumed by a sampling or Load/Store instruction. To make sure our
access chains are emitted correctly, we have to back-propagate this
decoration.
This is quite complex since we cannot flush Phi inside the case labels,
we have to do it outside by emitting a lot of manual branches ourselves.
This should be extremely rare, but we need to handle this case.
If we compile multiple times due to forced_recompile, we had
deferred_declaration = true while emitting function prototypes which
broke an assumption. Fix this by clearing out stale state before leaving
a function.
We seen to have to deal with a case where a block is used multiple times
without any "proper" structured control flow, so we risk losing deferred
declaration state.
This gets rather complicated because MSL does not support OpArrayLength
natively. We need to pass down a buffer which contains buffer sizes, and
we compute the array length on-demand.
Support both discrete descriptors as well as argument buffers.
MSL generally emits the aliases, which means we cannot always place the
master type first, unlike GLSL and HLSL. The logic fix is just to
reorder after we have tagged types with packing information, rather than
doing it in the parser fixup.
MSL does not seem to have a qualifier for this, but HLSL SM 5.1 does.
glslangValidator for HLSL does not support this, so skip any validation,
but it passes in FXC.
Buffer objects can contain arbitrary pointers to blocks.
We can also implement ConvertPtrToU and ConvertUToPtr.
The latter can cast a uint64_t to any type as it pleases,
so we will need to generate fake buffer reference blocks to be able to
cast the type.
We made the mistake of registering a dependency on the atomic variable
even if the atomic result was forced to a temporary. There is no need to
register reads from atomic variables like this as we always force atomic
results to a temporary and argument read/writes do not need to be
tracked.
If we generate an access chain in a loop body, and it is consumed in the
loop continue block, we have a problem because we cannot emit a
temporary here holding the access chain reference. Force a complex loop
body to workaround this exceptionally rare case.
- Replace ostringstream with custom implementation.
~30% performance uplift on vector-shuffle-oom test.
Allocations are measurably reduced in Valgrind.
- Replace std::vector with SmallVector.
Classic malloc optimization, small vectors are backed by inline data.
~ 7-8% gain on vector-shuffle-oom on GCC 8 on Linux.
- Use an object pool for IVariant type.
We generally allocate a lot of SPIR* objects. We can amortize these
allocations neatly by pooling them.
- ~15% overall uplift on ./test_shaders.py --iterations 10000 shaders/.
