| Age | Commit message (Collapse) | Author |
|---|
|
* Specify glsl semantic format - such that conversions are possible from hlsl sematics.
* Comment improvements. Give appropriate type in glsl for sv_tessfactor. Note that sv_tessfactor is not functional though.
* Work in progress for comparison of types.
* * Fix type comparison issues around the hash.
* Fix tests whos output changed with use of isTypeEqual
|
|
Previously, interface types were allowed to be used directly as function parameters, local variables, and global shader parameters.
Using an interface type as a field of a `struct` type or a `cbuffer` declaration was not implemented.
This change adds that support, and fixes several unrelated issues that caused problems in doing so.
* The most important work here was adding a case for `IRStructType` to `maybeSpecializeBindExistentialsType` that creates a specialized variant of a `struct` type on-demand based on specialization operands. This logic loops over the fields of the original struct, and creates new fields by binding the existentials/interfaces in the type of each field. Caching is used to ensure that the same `struct` type specialized to the same operands should yield the same result.
* To allow subsequent specialization to occur when a `struct` with interface-type fields is used, it was also necessary to specialize field-address and field-extract instructions in cases where the value that the field is being extracted from is a `wrapExistential`.
* Similarly, we neede to make sure that the logic for specializing called functions based on the concrete types for interfaces in the argument list would also take into account `struct` types with existential-type fields inside of them.
* Doing the above changes revealed some serious flaws in how the `ir-specialize.cpp` logic was tracking which instructions still needed to be processed. It had previously been assuming that it could assume any relevant instructions were on its work list, and when the work list went empty it could exit. This runs into two problems: (1) sometimes we create new instructions when specializing, and it may be impossible to ensure that all the new instructions (e.g., those created by utility routines in other files) get added to the work list, and (2) sometimes the instruction(s) that need to be re-visited when we specialize something aren't its direct users, but instead somethign that transitively depends on the instruction.
These issues were fixed by two changes to the pass: (1) we now maintain a list of known "clean" instructions instead of implicitly using the work-list as a list of "dirty" instructions (so that implicitly any new instruction is dirty), and periodically iterating over all instructions to add the non-clean ones to the work list for processing, and (2) when an instruction is specialized/replaced we mark everything that transitively depends on it "dirty" (by removing it from the "clean" list).
* Added some logic to "fix up" the type of an IR function after changes that might modify its parameter list. Failing to have this logic meant that certain types were still live (because they were referenced by a function type) that couldn't actually be emitted as legal HLSL/GLSL.
* Added some special cases to IR instruction creation for `wrapExistential` and `BindExistentialsType` so that they act as no-ops when there are no "slots" providing specialization information. This helps avoid some special cases when specializing structure fields (since some fields specialization and others don't, so in general there are zero or more operands specific to each field).
* Added a test case that uses an interface type in a `cbuffer`, as well as an interface type in a `struct` passed as an entry-point `uniform` parameter.
* Fixed up some parts of the `.natvis` files to reflect naming changes from a previous PR and thus restore some of the useful Visual Studio debugging experience for Slang.
|
|
* A few changes required for application adoption of interface-type parameters
There are a few small changes here that are all related in that they arose from trying to integrate support for specialization via global interface-type shader parameters into a real application.
Allow querying the "pending" layout via reflection API
------------------------------------------------------
The naming here isn't ideal, and could probably use a round of "bikeshedding" to arrive at something better, but the basic idea is that when you have a type like:
```
struct MyStuff
{
int a;
IFoo foo;
int b;
}
```
the fields `a` and `b` get allocated space directly in the "primary" layout for `MyStuff` (at offsets 0 and 4, with `sizeof(MyStuff) == 8`), but the `foo` field can't be allocated space until we know what concrete type will get plugged in there.
If we have a concrete type in mind:
```
struct Bar : IFoo { int bar; }
```
then we can know how much space the `foo` field will take up, but we still can't allocate it space directly in `MyStuff`, because we already decided that `sizeof(MyStuff) == 8`.
Now imagine we place some `MyStuff` values into constant buffers:
```
cbuffer X {
MyStuff x;
}
cbuffer Y {
MyStuff y;
float4 z;
}
```
In each case we know that we want to place the `MyStuff::foo` field at the end of the containing constant buffer so that it doesn't disrupt the layout of the existing fields. But that means that the offset of `MyStuff::foo` relative to the start of the `MyStuff` isn't fixed, because of unrelated fields like `z` that need to get in between.
In our layout code, we handle this by having a notion of a "pending" layout. Once we know how `MyStuff::foo` will be specialized, we can compute both a "primary" and a "pending" layout for `MyStuff`, which basically treats it as if it were two distinct types:
```
struct MyStuff_Primary
{
int a;
int b;
}
struct MyStuff_Pending
{
Bar foo;
}
```
Layout for an aggregate type like the `X` or `Y` constant buffer then proceeds by computing an aggregate primary layout and an aggregate pending layout, and then finally a constant buffer or parameter block "flushes" all or part of the pending data by appending it to the primary data to get the final layout.
What all this means is that a type like `MyStuff` will have two different layouts (a default one for the primary data and a "pending" one for any specialized interface-type fields), and a variable like `Y::y` will also have two variable layouts that specify offsets (one set of offsets for its primary part, and one set of offsets for its pending part).
In order to handle interface-type fields with these layout rules, an application needs a way to query the "pending" part of a type or variable layout, which luckily gives it back just another type/variable layout. The API change here is minimal, although actually exploiting the new API correctly in application code could prove challenging.
Allow creating of explicitly specialized types
----------------------------------------------
This feature isn't actually implemented all the way through the compiler (I just needed enough to make the API calls go through), but I've added support for specializing a type that has interface-type fields through the reflection API. This maps to an `ExistentialSpecializedType` in the AST, and I'm lowering it to the IR as a `BindExistentialsType`, although that isn't 100% correct for the future.
This feature will require a future PR to actually flesh out the implementation work, but I'll wait until that is the sticking point on the application side before I do that.
Introduce a tiny `Hasher` abstraction
-------------------------------------
While implementing all the boilerplate for a new `Type` subclass (we really need to reduce that work...), I got fed up with how we do hash-code computation and introduced a small utility `Hasher` type that is intended to wrap up the idiom of combining hashes. For now this isn't a major change, but in the future I'd like to expand on the design a bit to clean up some of the warts around how we handle hashing:
* The `Hasher` implementation can and should switch from maintaining a single `HashCode` as its state to something that contains a more complete state (larger than the hash code) and just hashes new bytes into that state as it goes. This should make it possible to implement a `Hasher` for more serious hash functions, whether MD5, CityHash, or whatever we decide is good default.
* Things that are hashable shouldn't have a `getHashCode()` method, but instead should have something like a `hashInto(Hasher&)` method. This change would have the dual benefits that (1) a composite type can easily hash all the fields that contribute to its identity into the hasher with minimal fuss/boilerplate, and (2) the hashes for composite types will be of higher quality because they can exploit all the bits of the hasher's state to combine the fields, instead of restricting each sub-field to just the bits in a hash code.
We should be able to incrementally improve the quality of our design there over future changes, but for now it probably isn't a critical priority.
Fixes for legalization of existential types
-------------------------------------------
There were some missing cases in the handling of type legalization, such that a global interface-type shader parameter that got specialized to a type that contains *only* resource-type fields would cause a crash in the legalization step.
I added a test for this case, and then made `ir-legalize-types.cpp` account for this case (the code to handle it ias a bit of a kludge, and shows that the `declareVars()` routine there is getting to a level of complexity that is worrying.
* fixup: review feedback
|
|
* Attempt to improve the glsl handling of hlsl semantics by taking into account the underlying glsl type.
* Improve comments around 'NV_VIEWPORT_MASK' on glsl.
|
|
trys to cast for glsl targets to avoid glslang producing a type error. (#962)
|
|
* * Fix warning in vk-swap-chain around use of Index. Rename _indexOf to _indexOfFormat.
* Rename IntSet to UIntSet and put in own files slang-uint-set.h.cpp. Use UInt as the held type.
* On UintSet setMax -> resizeAndClear. Doing so revealed bug in add.
* Closer following of conventions - use kPrefix for constants (even though held in 'enum')
* Small fixes/improvements
* * Add some documentation to UIntSet methods
* Use memset to set/reset bits
* Fix some tabbing.
Rename oldBufferSize -> oldCount
* Fix tabs.
|
|
* Convert bitwise Or & And to logical operations on scalar bools
* Test bitwise operations on scalar bools
|
|
* List made members m_
Tweaked types to closer match conventions.
* Use asserts for checking conditions on List.
Other small improvements.
* List<T>.Count() -> getSize()
* List<T>
Add -> add
First -> getFirst
Last -> getLast
RemoveLast -> removeLast
ReleaseBuffer -> detachBuffer
GetArrayView -> getArrayView
* List<T>::
AddRange -> addRange
Capacity -> getCapacity
Insert -> insert
InsertRange -> insertRange
AddRange -> addRange
RemoveRange -> removeRange
RemoveAt -> removeAt
Remove -> remove
Reverse -> reverse
FastRemove -> fastRemove
FastRemoveAt -> fastRemoveAt
Clear -> clear
* List<T>
FreeBuffer -> _deallocateBuffer
Free -> clearAndDeallocate
SwapWith -> swapWith
* List<T>
SetSize -> setSize
Reserve -> reserve
GrowToSize growToSize
* UnsafeShrinkToSize -> unsafeShrinkToSize
Compress -> compress
FindLast -> findLastIndex
FindLast -> findLastIndex
Simplify Contains
* List<T>
Removed m_allocator (wasn't used)
Swap -> swapElements
Sort -> sort
Contains -> contains
ForEach -> forEach
QuickSort -> quickSort
InsertionSort -> insertionSort
BinarySearch -> binarySearch
Max -> calcMax
Min -> calcMin
* Initializer::Initialize -> initialize
List<T>::
Allocate -> _allocate
Init -> _init
IndexOf -> indexOf
* * Put #include <assert.h> in common.h, and remove unneeded inclusions
* Small refactor of ArrayView - remove stride as not used
* getSize -> getCount
setSize -> setCount
unsafeShrinkToSize->unsafeShrinkToCount
growToSize -> growToCount
m_size -> m_count
* Some tidy up around Allocator.
* Use Index type on List.
* Refactor of IntSet.
First tentative look at using Index.
* Made Index an Int
Did preliminary fixes.
Made String use Index.
* Partial refactor of String.
* String::Buffer -> getBuffer
ToWString -> toWString
* Small improvements to String.
String::
Buffer() -> getBuffer()
Equals() -> equals
* Try to use Index where appropriate.
* Fix warnings on windows x86 builds.
|
|
Fixes #858
The `precise` keyword exists in both HLSL and GLSL and when applied to a variable declaration is supposed to indicate that all computations that contribute to the value of that variable should not be altered based on "fast-math" optimizations. The main examples are that separate multiply and add operations should not be turned into fused multiply-add (fma) operations, and that operations cannot ignore the possibility of infinity or not-a-number values (e.g., by assuming that `x * 0.0f` is always `0.0f`).
(Aside: it is possible that my understanding of what the semantics of `precise` are in HLSL and GLSL is imperfect so that either the GLSL variant isn't sufficient to provide the semantics of the HLSL keyword, or that the definition of "all computations that contribute" to a value isn't actually correct. We may need to revise this implementation based on subsequent learnings.)
The basic idea here is to turn the AST `precise` keyword into a `[precise]` decoration in the IR and then emit that as a `precise` keyword again in the output.
The main catch is that whereas most of our existing IR decorations apply to things like global shader parameters or `struct` members that usually stick around for the duration of compilation, `[precise]` will get slapped on local variables that will often get optimized away by our SSA pass. There are two ways a variable can get eliminated/replaced during the SSA pass:
1. A use of the variable can be replaced with an ordinary instruction that computes its value.
2. A use of the variable can be replaced with a reference to a "phi node" that will take on the appropriate value based on control flow.
These two cases already had logic to copy a "name hint" decoration from the variable over to an instruction that will replace it, and I simply extended them to also propagate over a `[precise]` decoration.
The test case added with this change intentionally constructs a case where `[precise]` needs to be propagated over to an SSA "phi node" in order to generate correct output code.
The other gotcha is that we can emit variable declarations in various places in `emit.cpp`, and all of these needed to handle `[precise]`. Not only do we have actually local variables (`IRVar`), but we also have SSA phi nodes (`IRParam`), and then there are cases where an intermediate computation (an ordinary instruction) should be `[precise]` and thus we need to emit it as a temporary (not folding it into its use sites) and make sure that the temporary itself gets the `precise` keyword.
I have manually confirmed that in the output SPIR-V, this change results in the `NoContraction` SPIR-V decoration being added to the relevant operations, and the output DXBC contains a multiply and an add in place of a multiply-add. The output DXIL does not show any obvious changes due to `precise`, although the exact order and operands of the math instructions emitted does differ when `precise` is added/removed. In all cases the output is equivalent to hand-written HLSL/GLSL with a `precise`-qualified local variable.
|
|
Fixes #941
The GLSL we were emitting for unbounded-size arrays was the obvious:
```hlsl
// This HLSL:
Texture2D t[];
```
```glsl
// ... becomes this GLSL:
texture2D t[];
```
Unfortunately, the legacy GLSL behavior for an array without a declared size is what is called an "implicitly-sized" array, which means that it is assumed to actually have a fixed size, which is determined by the maximum integer constant value used to index into it (and only integer constants are allowed to be used when indexing into it).
Users hadn't noticed the issue for a while, because most of our users who rely on unbounded-size arrays were also using the HLSL `NonUniformResourceIndex` function:
```hlsl
float4 v = t[NonUniformResourceIndex(idx)].Sample(...);
```
When mapping such code to GLSL we use the `nonuniformEXT` qualifier added by the `GL_EXT_nonuniform_qualifier` extension, and it turns out that a secondary feature of that extension is that it changes the GLSL language semantics for arrays (of resources) with an unspecified size, so that they instead behave like we want. So users were happy and we were blissfully ignorant of the lurking issue.
The problem is that as soon as a user neglects to use `NonUniformResourceIndex` (perhaps because an index really is uniform):
```hlsl
cbuffer C { uint definitelyUniform; }
...
float4 v = t[definitelyUniform].Sample(...);
```
Now the code we emit doesn't need `nonuniformEXT` so it doesn't enable `GL_EXT_nonuniform_qualifier` and the declaration of `t` now falls under the "implicitly-sized" array rules, and thus the code fails because `definitelyUniform` is being used as an index but is *not* an integer constant.
The fix is pretty simple: when emitting a declaration of a global shader parameter to GLSL, we check if it is an unbounded-size array of resources and, if so, enable the `GL_EXT_nonuniform_qualifier` extension.
We don't need any clever handling to deal with resource parameters nested in `struct` types or in entry-point parameter lists, etc., because previous IR passes will have split up complex types and moved everything to the global scope already.
|
|
* * Moved CPU determination macros to slang.h
* Determine SlangUInt/SlangInt from the pointer width (determined from CPU macros)
* Removed the UnambiguousInt and UnambigousUInt types - as a previous fragile work around
* Removed UInt/Int definition from smart-pointer.h as now in common.h
* * Remove ambiguity for PrettyWriter and ints
* Improve comment around SlangInt/UInt
* More fixes around ambiguity with PrettyWriter and integral types.
* Disable VK on OSX.
* Define guids with inner braces.
* For glslang use linux ossource for macosx.
* Pull is ossource for OSX.
* Fix dll loading for OSX.
* Added how to build for OSX to building.md.
* Force CI to rebuild as spurious error.
* Improvements to the building.md documentation.
* Small doc fix.
|
|
* * Moved CPU determination macros to slang.h
* Determine SlangUInt/SlangInt from the pointer width (determined from CPU macros)
* Removed the UnambiguousInt and UnambigousUInt types - as a previous fragile work around
* Removed UInt/Int definition from smart-pointer.h as now in common.h
* * Remove ambiguity for PrettyWriter and ints
* Improve comment around SlangInt/UInt
* More fixes around ambiguity with PrettyWriter and integral types.
* Disable VK on OSX.
* Force CI to rebuild as spurious error.
|
|
* * Make Path:: use lowerCamel method names as per coding standard
* Small improvements to make closer to standard
* GetDirectoryName -> getParentDirectory - previous method name's action was somewhat unclear, hopefully this is better
* * Can build on clang and gcc on CygWin
* Fix problem on cygwin loading shared libraries
* Renamed Path::isRelative to ::hasRelativeElement because isRelative implies the path is 'relative to the current path' and which isn't quite what it does
* Documented how to build for CygWin
* * Fix small bug creating platform shared library name.
* Small typo fixes in building.md
|
|
* Small improvements to make closer to standard
* GetDirectoryName -> getParentDirectory - previous method name's action was somewhat unclear, hopefully this is better
|
|
* * Remove Makefile
* Document how to create build using premake5
* Added support for finding the executable path
* If binDir not set on command line use the executable path
* Fix getting exe path on linux.
* Removed CalcExecutablePath from Path:: interface, made implementation internal.
* Documentation improvements.
* Fixes based on review
* Fix some typos
* Removed unused/needed global
|
|
|
|
generation. (#947)
|
|
* Update glslang
This moves to a version of glslang that is hosted with the slang project and that includes a patch for a high-priority fix that hasn't been upstreamed into the main glslang repository yet.
* Change a GLSL extension name
The glslang codebase changed the extension name required to enable certain features from `GL_KHX_shader_explicit_arithmetic_types` to `GL_EXT_shader_explicit_arithmetic_types`.
|
|
* Add better control over image formats for GLSL/SPIR-V targets
Currently Slang emits GLSL code assuming all R/W images need to have explicit formats, and thus we try to infer a format from the element type of the image.
E.g., given a `RWTexture2D<half4>` we might infer that a qualifier of `layout(rgba16f)` should be used.
This strategy has two notable shortcomings:
* Sometimes the user will want a format that doesn't match an existing HLSL type. E.g., if they want the equivalent of `layout(r11f_g11f_b10f)`, then what should they put in their `RWTexture2D<...>` to make the inference do what they need?
* Sometimes the user knows that they don't need to specify a format *at all*, because using the `GL_EXT_shader_image_load_formatted` extension, they can still perform non-atomic load/store on images with no format specified in the SPIR-V.
This change adds two features directed at these challenges.
First, we add an explicit `[format(...)]` attribute that can be used to specify an explicit image format, including ones that don't match any HLSL type.
An example of using this new attribute is:
```hlsl
[format("r11f_g11f_b10f")]
RWTexture2D<float3> myImage;
```
For simplicity in initial bring-up, the new formats all use the same naming as formats in GLSL (this should make it easy for a programmer who knows what they expect to get in the GLSL output). We can change the naming convention for formats at a later time, so long as we keep these existing names in as a compatibility feature.
Note that this is *not* given a `vk::` prefix since the attribute should signal the programmer's intent to provide an image with that format on *all* targets (although only some targets might act on that information).
Also note that the attribute takes a string (`[format("rgba8")`) instead of a bare identifier (`[format(rgba8)]`) because this is consistent with the existing convention for attributes in HLSL.
When `[format(...)]` is left off, the default compiler behavior will still be to infer a format, but this behavior can be overidden for a single image using an explicit format of `"unknown"`:
```hlsl
[format("unknown")]
RWTexture2D<float4> mysteryMachine;
```
The second new feature is that if a user knows they are coding for a GPU that supports the `"unknown"` format in all non-atomic cases, then they can opt into making that the default for images without an explicit `[format(...)]`, using the new `-default-image-format-unknown` command-line option for `slangc`.
The new test case included with this change confirms that we correctly see the explicit formats in the output GLSL and *no* formats for images without explicit `[format(...)]` when using the new command-line option. The test stresses images declared at global scope, in parameter blocks, and in entry-point parameter lists, to try and make sure that all the relevant IR passes in the compiler preserve the format information.
* fixup: missing file
|
|
buffer (#936)
|
|
Fixes #782
There is logic in the compiler to confirm that the argument expression for an `out` or `inout` parameter is an l-value. That logic was producing an internal compiler error if it ran out of arguments while processing the parameter list, on the assumption that this would mean an `out` parameter had a default argument expression (which isn't something we want to support).
The problem was that the checking for call expressions will diagnose a call with too few arguments, and then leave the call in the AST to support subequent checking. This meant that any call where the user didn't supply enough arguments *and* one of the trailing argument is `out` or `inout` would produce the error for the original problem (not enough arguments), but then *also* produce the internal error because there is seemingly no argument to match with the `out` or `inout` parameter.
The right fix is to not take responsibility for diagnosing this problem at the call site, and instead to rule out default value expressions for `out` and `inout` parameters at the declaration site instead.
|
|
* * Added $c macro - that will do casting to target type. Used here to cast texture reads back to half. Works in tandem with $z which will close parens.
* half-texture.slang test
* Make binding failing if TextureView fails
* Simplify logic around parens.
* Improve comment around $c macro.
* Test against hlsl output to avoid error on CI.
|
|
* Overhaul the core routines for implicit conversion
The main user-visible change is that we have fixed the bug where conversions that should only be allowed explicitly were being allowed implicitly. This might be seen as a regression by users, so we'll have to be careful when rolling out the fix.
The core of that fix involves checking whether an `init` declaration that will be invoked as an implicit conversion actually supports implicit conversions.
The main visible change in the code is some renamings to try and help make the core type-coercion routines better fit our naming conventions.
The main cleanup is to enforce the invariant that any of the implicit-conversion core routines will always emit a diagnostic (or have a subroutine it calls do so) when conversion fails and the `outToExpr` parameter is non-null. This is a small change, but should improve the user experience if an implicit conversion fails in the context of a single element of an initializer list (the error should point at the line in question, and not at the whole list).
The big thing that is impacted by removing the ability to use explicit conversions implicitly is conversion of `enum` types to integers. This was intended to be explicit (a la `enum class` in C++), but the bug made it so that implicit conversion was allowed.
Closing up that gap meant that some of the checking around user-defined attributes got wonky, because we attempt to check that the attribute argument is an integer constant expression, but an `enum` case can't possible be an integer constant - it is a value of the `enum` type. I added code to work around that issue by having a parallel path for checking compile-time-constant expressions of `enum` type, but it is clear that a more general solution is needed eventually.
* fixup: test case needs explicit cast
|
|
* Allow plugging in types with resources for interface parameters
The key feature enabled by this change is that you can take a shader declared with interface-type parameters:
```hlsl
ConstantBuffer<ILight> gLight;
float4 myShader(IMaterial material, ...)
{ ... }
```
and specialize its interface-type parameters to concrete type that can contain resources like textures, samplers, etc.
The hard part of doing this layout is that we need to support signatures that include a mix of interface and non-interface types. Imagine this contrived example:
```hlsl
float4 myShader(
Texture2D diffuseMap,
ILight light,
Texture2D specularMap)
{ ... }
```
We end up wanting `diffuseMap` to get `register(t0)` and `specularMap` to get `register(t1)`, so that they have the same location no matter what we plug in for `light`.
But if we plug in a concrete type for `light` that needs a texture register, we need to allocate it *somewhere*.
We handle this by having the `TypeLayout` for `light` come back with a "primary" type layout that doesn't have any texture registers, but with a "pending" type layout that includes the texture register requirements of whatever concrete type we plug in.
This split between "primary" and "pending" layout then needs to work its way up the hierarchy, so that an aggregate `struct` type with a mix of interface and non-interface fields (recursively), needs to compute an aggregate "primary type layout" and an aggregate "pending type layout," and then each field needs to be able to compute its offset in the primary/pending layout of the aggregate.
A large chunk of the work in this PR is then just implementing the split between primary and pending data, and ensuring that layouts are computed appropriately.
The next catch is that when a "parameter group" (either a parameter block or constant buffer) contains one or more values of interface type, then we can allow the parameter group to "mask" some of the resource usage of the concrete types we plug in, but others "bleed through."
For example, if we have:
```hlsl
struct MyStuff { float3 color; ILight light; }
ConstantBuffer<MyStuff> myStuff;
struct SpotLight { float3 position; Texture2D shadowMap; }
``
If we plug in the `SpotLight` type for `myStuff.light`, then the `float3` data for the light can be "masked" by the fact that we have a constant buffer (we can just allocate the `float3` `position` right after `color`), but the `Texture2D` needed for `shadowMap` needs to "bleed through" and become "pending" data for the `myStuff` shader parameter.
Adding support for that detail more or less required a full rewrite of the logic for allocating parameter group type layouts.
The next detail is that when we go to legalize a declaration like the `myStuff` buffer, we will end up with something like:
```hlsl
struct MyStuff_stripped { float3 color; }
struct Wrapped
{
MyStuff_stripped primary;
SpotLight pending;
}
ConstantBuffer<Wrapped> myStuff;
```
This "wrapped" version of the buffer type more accurately reflects the layout we need/want for the uniform/ordinary data, but in order to further legalize it and pull out the resource-type fields like `shadowMap` we need to have accurate layout information, and the problem is that layout information for the original buffer can't apply to this new "wrapped" buffer.
The last major piece of this change is logic that runs during existential type legalization to compute new layouts for "wrapped" buffers like these that embeds correct offset/binding/register information for any resources nested inside them. A key challenge in that code is that existential legalization needs to erase any "pending" data from the program entirely, so that offset information that used to be relatie to the "pending" part of a surrounding type now needs to be relative to the primary part.
The work here may not be 100% complete for all scenarios, but it does well enough on the new and existing tests that I want to checkpoint it. Note that a few other tests have had their output changed, but in all cases I've reviewed the diffs and determined that the change in observable behavior is consistent with what we intened Slang's behavior to be.
Note that there is still one major piece of support for interface-type parameters that is missing here, and which might force us to revisit some of the decisions in this code: we don't properly support user-defined `struct` types with interface-type fields.
* fixup: typos
|
|
An `enum` case currently lowers to the IR as the value of its tag expression, so if we have:
```hlsl
enum E
{
X = 99,
}
```
then `X` will lower as IR for the expression `99` which is just a literal.
If instead we have:
```hlsl
enum E
{
X = 99u,
}
```
then after type checking the expression is `int(99u)`, which will get emitted to the global (module) scope as a cast/conversion instruction, that then gets referenced at the use sites of `E.X`.
The emit logic wasn't set up to handle the case of referencing something directly from the global scope, so this change makes it so that side-effect-free global instructions are treated just like literal constants. This simple handling is fine for now since it will only apply to `enum` tag values (true global constants declared with `static const` have different handling).
|
|
If the user declares global shader parameters for D3D SM5.1+ or Vulkan, then they need to go into an appropriate `space` or `set`:
```hlsl
Texture2D t; // should go in space/set 0
SamplerState s; // same here...
```
This also applies to allocation of spaces/sets to parameter blocks:
```hlsl
ParameterBlock<X> x; // should get space/set 0
ParameterBlock<Y> y; // should get space/set 1
```
In cases where there are a combination of explicitly and implicitly bound parameters, anything left implicitly bound goes into a "default" space/set:
```
ParameterBlock<X> x : register(space0); // this has claimed space/set 0
Texture2D t; // this needs a space, so a "default" space/set of 1 will be claimed
SamplerState s; // this also needs a space/set, and will use the default
```
The logic for deciding when a default space/set was needing was, more or less, looking at all the global shader parameters and seeing if any of them needed a `register`/`binding`, and if so determining that a default space /set would be needed.
There was a bug in that logic, though, because of cases like the following:
```hlsl
ParameterBlock<X> x;
Texture2D t : register(t0, space99);
```
In this case, the parameter `t` already has an explicit binding, so it doesn't actually need a default space to be allocated. If we allocate a default space/set of 0 on the basis of `t`, then `x` will end up being shifted to space/set 1.
The fix is to only consider global parameters that need `register`s/`binding`s *if* they don't have an explicit binding already (which is luckily something we are tracking during parameter binding).
Note: just to clarify the behavior here, the "do we need a default space/set?" logic is done *before* automatic binding of parameters, so in a shader with any global texture/buffer/sampler parameters, those will all end up in space/set zero (in the absence of explicit bindings), and explicit blocks will start at space/set one, independent of the order of declaration. This behavior is maybe too subtle, and we might decide we need to change it, but it will have to do for now.
|
|
* Disable Dx12 half tests. The half-calc test runs, but is not actually doing any half maths. If the code is changed such that it is, the device fails when the shader is used. This can be seen by looking at dxil-asm.
* Fix using software driver for dx12 even when hardware is requested.
* * Refactor Dx12 _createAdapter such that it doesn't have side effects and stores desc information
* Disable half on dx12 software renderer because it crashes
* * Disable erroneous warnings from dx12
* Test for adapter creation
* Identify warp specifically
* Structured buffer test now works on dx12.
* Fix intemittent crash on dx12.
Due to if a resource was initialized with data, the actual resource constructed might be larger, for alignment issues. This led to a memcpy potentially copying from after the allocated source data and therefore a crash. Now only copies the non aligned amount of data.
* * Rename the test to use - style
* Disable TextureCube lookup in tests, as does not produce the correct result in dx12 (will fix in future PR)
* Updated hlsl.meta.slang.h that has rcp for glsl.
|
|
|
|
* Look at getting half to work on vk.
* Alter half test so can always produce consistent test results.
* First pass working half on vk.
* Improve comments for vulkan extensions around half.
* Upgraded vulkan headers to v1.1.103
https://github.com/KhronosGroup/Vulkan-Headers
* * Add getFeatures on Render interface
* Vulkan renderer determines at startup if it can support half
* Parse render-features on render-test
* Small changes to half-calc.slang test.
* Structured buffer half access works as expected for Vk, but isn't for dx12, so disable for now.
* Require the half feature for renderers for the half-structured-buffer.slang test.
* * Added ToolReturnCode to be more rigerous about how a return code is passed back from a tool
* Added support for a tool being able to pass back an 'ignored' result.
* Used enum codes to indicate meanings
* Made spawnAndWait return a ToolReturnCode
* Ignore tests that don't have required render-feature
* Fix macro line continuation usage.
* Check dx12 has half support.
* Checking for half on dx12 - if CheckFeatureSupport fails, don't fail renderer initialization.
* Fix typo.
|
|
back from a tool (#911)
* Added support for a tool being able to pass back an 'ignored' result.
* Used enum codes to indicate meanings
* Made spawnAndWait return a ToolReturnCode
|
|
* * Handle ! for bool vector in glsl
* Handle operators that have a boolean return value
* || or && take bool
* * Add comment in bool-op.slang test about doing || or && on vector types not supported for GLSL targets
|
|
* * leftSide and rightSide set op to nullptr, before was just uninitialized
* Added support for GLSL for vector/scalar comparisons
* Added test
* * Remove unneeded precedence code.
* Simplify function to _maybeEmitGLSLCast
* * Take into account precedence & closing of brackets in same way as function call, if function call used for vector comparison (as on GLSL)
|
|
if smaller type is used conversion is performed (#902)
|
|
* * On GLSL targets, output texture layouts that are 3 component as 4 component, because 4 component style is not supported by glslang currently
* * Improve comment around SPIR-V and 3 component layouts.
|
|
RaytracingAccelerationStructureType (#901)
* Add SLANG_ACCELERATION_STRUCTURE resource shape for RaytracingAccelerationStructureType
* Change order of resource shape cases
I've changed the order of the `UNKNOWN` and `ACCELERATION_STRUCTURE` cases so that the binary value of the `UNKNOWN` case isn't changed by the new feature.
|
|
The `tuple` case in `getPointedToType` was failing to add the elements it computed to the output tuple type.
This isn't triggered in any of our test cases, but was caught by some work I was doing in another branch.
|
|
The short version for command-line users is:
* Use `-g` to get debug info in the output, where supported
* Use `-O0` to disable optimizations, in case that improves debugability
* Use `-O2` for optimized/release builds where you can spend the extra compile time
The command-line options are matched with new API functions `spSetDebugInfoLevel()` and `spSetOptimizationLevel()` that set the equivalent information.
Right now these settings only affect how we invoke fxc and dxc. In the longer run I expect we will want to use them to control other things:
* Once we are emitting our own SPIR-V, the `-g` option should control what source-level name information we include in it.
* Whether or not `-g` is used could be used to decide whether we preserve the "name hints" in the IR, which in turn decide whether we output GLSL/HLSL source that uses names based on the original program.
* We will eventually need/want to include some amount of optimization passes on the Slang IR, and the `-O` options should control which of those passes are enabled on a particular invocation.
In this change I decided to expose the options at the level of the entire compile request for API users, and to store the actual information on the Linkage. We might want to revisit this decision and instead allow for the level of optimization to be chosen per-target as part of back-end state. Similarly, we might want to have more fine-grained control over the level of debug output per-target (although we'd still need a front-end setting to determine what debug info is generated into the Slang IR).
|
|
Before type legalization we might have code like: (using pseudo-Slang-IR):
struct P { ... Texture2D<float>[] t; }
global_param p : ParameterBlock<P>;
...
// p.t[someIndex].Load(...);
//
let ptrToArrayOfTextures = getFieldPtr(p, "t") : Ptr<Texture2D<float>[]>;
let ptrToTexture = getElementPtr(ptrToArrayOfTextures, someIndex) : Ptr<Texture2D<float>>;
let texture = load(ptrToTexture) : Texture2D<float>;
let result = call(loadFunc, texture, ...) : float;
Legalization needs to move the `t` array there out of the `p` parameter block, so the global declarations become something like:
struct P_Ordinary { ... }; // no more "t" field
global_param p_ordinary : ParameterBlock<p_ordinary);
global_param p_t : Texture2D<float>[];
In terms of the code to access `p.t[someIndex]` the problem is that `p_t` has one less level of indirection than `p.t` had. We solve this in the type legalization pass using "pseudo-types" and "pseudo-values," where one of the cases is `implicitIndirect` which holds a value of type `T`, but indicates that it should act like a value of type `T*`.
We then use some basic rules for dealing with `implicitIndirect` values, such as:
load(implicitDeref(x)) : T => x : T
getFieldPtr(implicitDeref(s), f) => implicitDeref(getField(s, f))
getElementPtr(implicitDeref(a), i) => implicitDeref(getElement(a, i))
The bug here was that for the `getFieldPtr` and `getElementPtr` cases, we weren't computing the type of the `getField` or `getElement` instruction correctly. We were copying the type from the `getFieldPtr` or `getElementPtr` operation over directly, but those will be *pointer* types and we need the type of whatever they point to.
Once the types are fixed, we can properly generate legalized IR for `p.t[someIndex].Load(...) that looks like:
let arrayOfTextures = p_t : Texture2D<float>[];
let texture = getElement(arrayOfTextures, someIndex) : Texture2D<float>;
let result = call(loadFunc, texture, ...) : float;
The old was giving the `texture` intermediate a type of `Ptr<Texure2D<float>>`. That didn't actually trip up too many things, because we mostly just went on to emit code from something with slightly incorrect types for intermediates that never show up in the generated HLSL/GLSL.
Where this caused a problem is for some of the intrinsic function definitions for the GLSL/Vulkan back-end, because those do things that inspect operand types. In particular the `$z` opcode in our intrinsic function strings triggers logic that looks at a texture operand, and uses its type to try to find the appropriate swizzle to get from a 4-component vector to the appropriate type for the operation (e.g., for a load from a `Texture2D<float>` we need to swizzle with `.x` to get a single scalar out of the matching GLSL texture fetch operation).
The main fix in this change is thus to make `getElementPtr` and `getFieldPtr` legalization properly account for the fact that when switching to `getElement` or `getField` we need a result type that is the "pointee" of the original result.
There was already logic to extract the pointed-to type from a pointer in `ir-specialize.cpp`, so I extracted that to a re-usable function in the IR as `tryGetPointedToType` (returns null if the type isn't actually a pointer).
This logic needed to be extended for type legalization, to deal with the various "pseudo-type" cases.
There is another fix in this change which is marking the `NonUniformResourceIndex` function as `[__readNone]`, which enables it to be more aggressively folded into use sites. Without that fix, we risk emitting code like:
```glsl
int tmp = nonUniformEXT(someIndex);
vec4 result = texelFetch(arrayOfTextures[tmp], ...);
```
The problem with that code is that (at least by my reading of the spec), assigning to the variable `tmp` that isn't declared with the `nonUniformEXT` qualifier effectively loses that qualifier, and drivers are free to assume that `tmp` is uniform when used to index into `arrayOfTextures`.
Marking the `NonUniformResourceIndex` function as `[__readNone]` indicates that it has no side effects, which should mean that our emit logic no longer needs to emit it was its own line of code to be safe.
The effects of this change are confirmed by both the new test case added, and the existing `non-uniform-indexing` test.
|
|
|
|
|
|
* Improve support for interfaces as shader parameters
This change adds two main things over the existing support:
1. It is now possible to plug in concrete types that actually contain (uniform/ordinary) fields for the existential type parameters introduced by interface-type shader parameters. The `interface-shader-param2.slang` test shows that this works.
2. There is a limited amount of support for doing correct layout computation and generating output code that matches that layout, so that interface and ordinary-type fields can be interleaved to a limited extent. The `interface-shader-param3.slang` test confirms this behavior.
There are several moving pieces in the change.
* When it comes to terminology, we try to draw a more clear distinction between existial type parameters/arguments and existential/object value parametes/arguments. A simple way to look at it is that an `IFoo[3]` shader parameter introduces a single existential type parameter (so that a concrete type argument like `SomeThing` can be plugged in for the `IFoo`) but introduces three existential object/value parameters (to represent the concrete values for the array elements).
* At the IR level, we support a few new operations. A `BindExistentialsType` can take a type that is not itself an interface/existential type but which depends on interfaces/existentials (e.g., `ConstantBuffer<IFoo>`) and plug in the concrete types to be used for its existential type slots.
* Then a `wrapExistentials` instruction can take a type with all the existentials plugged in (possibly by `BindExistentialsType`) and wrap it into a value of the existential-using type (e.g., turn `ConstantBuffer<SomeThing>` into a `ConstantBuffer<IFoo>`).
* The IR passes for doing generic/existential specialization have been updated to be able to desugar uses of these new operations just enough so that a `ConstantBuffer<IFoo>` can be used.
* When we specialize an IR parameter of an interface type like `IFoo` based on a concrete type `SomeThing`, we turn the parameter into an `ExistentialBox<SomeThing>` to reflect the fact that we are conceptually referring to `SomeThing` indirectly (it shouldn't be factored into the layout of its surrounding type).
* Parameter binding was updated so that it passes along the bound existential type arguments in a `Program` or `EntryPoint` to type layout, so that we can take them into account. The type layout code needs to do a little work to pass the appropriate range of arguments along to sub-fields when computing layout for aggregate types.
* Type layout was updated to have a notion of "pending" items, which represent the concrete types of data that are logically being referenced by existential value slots. The basic idea is that these values aren't included in the layout of a type by default, but then they get "flushed" to come after all the non-existential-related data in a constant buffer, parameter block, etc.
* The logic for computing a parameter group (`ConstantBuffer` or `ParameterBlock`) layout was updated to always "flush" the pending items on the element type of the group, so that the resource usage of specialized existential slots would be taken into account.
* The type legalization pass has been adapted so that we can derive two different passes from it. One does resource-type legalization (which is all that the original pass did). The new pass uses the same basic machinery to legalize `ExistentialBox<T>` types by moving them out of their containing type(s), and then turning them into ordinary variables/parameters of type `T`.
Big things missing from this change include:
- Nothing is making sure that "pending" items at the global or entry-point level will get proper registers/bindings allocated to them. For the uniform case, all that matters in the current compiler is that we declare them in the right order in the output HLSL/GLSL, but for resources to be supported we will need to compute this layout information and start associating it with the existential/interface-type fields.
- Nothing is being done to support `BindExistentials<S, ...>` where `S` is a `struct` type that might have existential-type fields (or nested fields...). Eventually we need to desugar a type like this into a fresh `struct` type that has the same field keys as `S`, but with fields replaced by suitable `BindExistentials` as needed. (The hard part of this would seem to be computing which slots go to which fields). As a practial matter, this missing feature means that interface-type members of `cbuffer` declarations won't work.
The current tests carefully avoid both of these problems. They don't declare any buffer/texture fields in the concrete types, and they don't make use of `cbuffer` declarations or `ConstantBuffer`s over structure types with interface-type fields.
* fixup: add override to methods
* fixup: typos
|
|
* Fix up handling of NumElements for D3D buffer views
For everything but structured buffers we'd been setting this to the size in bytes, but that isn't really valid at all.
The `NumElements` member in the view descs is supposed to be the number of buffer elements, so it would be capped at the byte size divided by the element size.
This change fixes the computation of `NumElements` to take the size of the format into account for non-structured views.
For the "raw" case, we use a size of 4 bytes since that matches the `DXGI_FORMAT_R32_TYPELESS` format we use (which seems to be required for raw buffers).
I also added support for the raw case for SRVs where it didn't seem to be supported before (not that any of our tests cover it).
* Fix handling of size padding for D3D11 buffers
The existing code was enforcing a 256-byte-aligned size for all buffers, but this can cause problems for a structured buffer.
A structured buffer must have a size that is a multiple of the stride, so a structured buffer with a 48-byte stride and a 96-byte size would get rounded up to 256 bytes, which is not an integer multiple of 48.
This change makes it so that we only apply the padding to constant buffers.
According to MSDN, constant buffers only require padding to a 16-byte aligned size, and no other restrictions are listed for D3D11, but it is difficult to know whether those constrains are exhaustive.
I've left in the 256-byte padding for now (rather than switch to 16-byte), even though I suspect that was only needed as a band-aid for the `NumElements` issue fixed by another commit.
* Fix an IR generation bug when indxing into a strutured buffer element
The problem here arises when we have a structured buffer of matrices (an array type would likely trigger it too):
```hlsl
RWStructuredBuffer<float3x4> gMatrices;
```
and then we index into it directly, rather than copying to a temporary:
```hlsl
// CRASH:
float v = gMatrices[i][j][k];
// OKAY:
float3x4 m = gMatrices[i];
float v = m[j][k];
```
The underlying issue is that our IR lowering pass tries to defer the decision about whether to use a `get` vs. `set` vs. `ref` accessor for a subscript until as late as possible (this is to deal with the fact that sometimes D3D can provide a `ref` accessor where GLSL can only provide a `get` or `set`).
We probably need to overhaul that aspect of IR codegen sooner or later, but this change uses some of the existing machinery to try to force the `gMatrices[i]` subexpression to take the form of a pointer when doing sub-indexing like this. This fixes the present case, and hopefully shouldn't break anything else that used to work (because the subroutines I'm using to coerce the `gMatrices[i]` expression should be idempotent on the cases that were already implemented).
* Add a test case to confirm fxc/dxc layout for structured buffers of matrices
Even when row-major layout is requested globally, fxc and dxc seem to lay out a `StructuredBuffer<float3x4>` with column-major layout on the elements.
This commit adds a test that confirms that behavior.
This commit does not try to implement a fix for the issue (either fixing Slang's layout reflection information to be correct for what fxc/dxc do in practice, or fixing Slang's HLSL output to work around the fxc/dxc behavior), but just documents the status quo.
If/when we decide how we'd like to handle the issue long-term, this test can/should be updated to match the decision we make.
* fixup: build breakage on clang/gcc
This is one of those cases where the Microsoft compiler is letting through some stuff that isn't technically valid C++ ("delayed template parsing").
Fixed by just moving some declarations to earlier in the file.
|
|
|
|
lines (#885)
* * Check for inconsistent command line options for renderer
* Moved RenderApiUtil into core so can be used in slang-test
* Make it use the ShaderdLibrary for API testsing
* Added some simplifying functions to StringUtil for spliting/comparisons
* Refactored the synthesis of rendering tests so that inconsistent combinations are not produced
* Add missing slang-render-api-util.cpp & .h
* Stop warning on linux about _canLoadSharedLibrary not being used.
|
|
step. (#884)
|
|
* Add support for glsl inversesqrt intrinsic
* fixup for test failure
|
|
* Added test for scope operator
|
|
|
|
* Add diagnostic for vk::binding failure.
* Add test for vk::binding failure.
* Add the expected output for glsl-layout-define.hlsl
* * Copy over initialize expr if available when validating unchecked
* Fix unloop - because now it always has one parameter (when before it could have none)
* Split vk::binding and layout tests with invalid parameters
* Removed the diagnostic for 2 ints expected
* Added vk::binding that doesn't specify set in vk-bindings.slang
* * Fix typo
* Improve comments.
|
|
* First pass test to see if GatherRed works.
* Add support for generating R_Float32 textures.
* Set default texture format.
* * Alter the texture2d-gather to work with a R_Float32 texture
* Add support for scalar Texture2d types with GatherXXX in stdlib
* Remove some left over commented out test code from texture2d-gather.hlsl
|
