These functions all map to the corresponding LLVM intrinsics, but the
vector intrinsics weren't being generated. The intrinsic mapping from
CLC vector function to vector intrinsic was working correctly, but the
mapping from OpenCL builtin to CLC function was suboptimally recursively
splitting vectors in halves.
For example, with this change, `ceil(float16)` calls `llvm.ceil.v16f32`
directly once optimizations are applied.
Now also, instead of generating LLVM intrinsics through `__asm` we now
call clang elementwise builtins for each CLC builtin. This should be a
more standard way of achieving the same result
The CLC versions of each of these builtins are also now built and
enabled for SPIR-V targets. The LLVM -> SPIR-V translator maps the
intrinsics to the appropriate OpExtInst, so there should be no
difference in semantics, despite the newly introduced indirection from
OpenCL builtin through the CLC builtin to the intrinsic.
The AMDGPU targets make use of the same `_CLC_DEFINE_UNARY_BUILTIN`
macro to override `sqrt`, so those functions also appear more optimal
with this change, calling the vector `llvm.sqrt.vXf32` intrinsics
directly.
Using '.hi' on a vector3 is technically allowed by the spec and is
treated as a 4-element vector with an "undefined" w component. However,
it's more undef/poison code for the compiler to process and remove. We
can easily avoid it with a dedicated macro.
This keeps values in vectors, rather than scalarizing them and then
reconstituting the vector. The builtin is identical to performing a
C-style cast on each element, which is what we were doing by recursively
splitting the vector down to calling the "base" conversion function on
each element.
The OpenCL relational functions now call their CLC counterparts, and the
CLC relational functions are defined identically to how the OpenCL
functions were defined.
As usual, clspv and spir-v targets bypass these.
No observable changes to any libclc target (measured with llvm-diff).
These functions are all mapped to LLVM intrinsics.
The clspv and spirv targets don't declare or define any of these CLC
functions, and instead map these to their corresponding OpenCL symbols.
These functions are "shared" between integer and floating-point types,
hence the directory name. They are used in several CLC internal
functions such as __clc_ldexp.
Note that clspv and spirv targets don't want to define these functions,
so pre-processor macros replace calls to __clc_min with regular min, for
example. This means they can use as much of the generic CLC source files
as possible, but where CLC functions would usually call out to an
external __clc_min symbol, they call out to an external min symbol. Then
they opt out of defining __clc_min itself in their CLC builtins library.
Preprocessor definitions for these targets have also been changed
somewhat: what used to be CLC_SPIRV (the 32-bit target) is now
CLC_SPIRV32, and CLC_SPIRV now represents either CLC_SPIRV32 or
CLC_SPIRV64. Same goes for CLC_CLSPV.
There are no differences (measured with llvm-diff) in any of the final
builtins libraries for nvptx, amdgpu, or clspv. Neither are there
differences in the SPIR-V targets' LLVM IR before it's actually lowered
to SPIR-V.
Some libclc builtins currently use internal builtins prefixed with
'__clc_' for various reasons, e.g., to avoid naming clashes.
This commit formalizes this concept by starting to isolate the
definitions of these internal clc builtins into a separate
self-contained bytecode library, which is linked into each target's
libclc OpenCL builtins before optimization takes place.
The goal of this step is to allow additional libraries of builtins
that provide entry points (or bindings) that are not written in OpenCL C
but still wish to expose OpenCL-compatible builtins. By moving the
implementations into a separate self-contained library, entry points can
share as much code as possible without going through OpenCL C.
The overall structure of the internal clc library is similar to the
current OpenCL structure, with SOURCES files and targets being able to
override the definitions of builtins as needed. The idea is that the
OpenCL builtins will begin to need fewer target-specific overrides, as
those will slowly move over to the clc builtins instead.
Another advantage of having a separate bytecode library with the CLC
implementations is that we can internalize the symbols when linking it
(separately), whereas currently the CLC symbols make it into the final
builtins library (and perhaps even the final compiled binary).
This patch starts of with 'dot' as it's relatively self-contained, as
opposed to most of the maths builtins which tend to pull in other
builtins.
We can also start to clang-format the builtins as we go, which should
help to modernize the codebase.
Increase fp16 support to allow clspv to continue to be OpenCL compliant
following the update of the OpenCL-CTS adding more testing on math
functions and conversions with half.
Math functions are implemented by upscaling to fp32 and using the fp32
implementation. It garantees the accuracy required for half-precision
float-point by the CTS.
Instead add a proper attribute in clang, and add convert it to function
metadata to keep the information in the IR. The goal is to remove the
dependency on __attribute__((assume)) that should have not be there in
the first place.
Ref https://github.com/llvm/llvm-project/pull/84934
The previous build system was adding custom "OpenCL" and "LLVM IR"
languages in CMake to build the builtin libraries. This was making it
harder to build in-tree because the tool binaries needed to be present
at configure time.
This commit refactors the build system to use custom commands to build
the bytecode files one by one, and link them all together into the final
bytecode library. It also enables in-tree builds by aliasing the
clang/llvm-link/etc. tool targets to internal targets, which are
imported from the LLVM installation directory when building out of tree.
Diffing (with llvm-diff) all of the final bytecode libraries in an
out-of-tree configuration against those built using the current tip
system shows no changes. Note that there are textual changes to metadata
IDs which confuse regular diff, and that llvm-diff 14 and below may show
false-positives.
This commit also removes a file listed in one of the SOURCEs which
didn't exist and which was preventing the use of
ENABLE_RUNTIME_SUBNORMAL when configuring CMake.
Add a clspv switch in gen_convert.cl
This is needed as Vulkan SPIR-V does not respect the assumptions
needed to have the generic convert.cl compliant on many platforms.
It is needed because of the conversion of TYPE_MAX and
TYPE_MIN. Depending on the platform the behaviour can vary, but most
of them just do not convert correctly those 2 values.
Because of that, we also need to avoid having explicit function for
simple conversions because it allows llvm to optimise the code, thus
removing some of the added checks that are in fact needed.
We noticed this same issue in our own implementation of abs_diff, and
the same issue also came up in the abs_diff reference function in the
OpenCL CTS.
Reviewed By: rjodinchr
Differential Revision: https://reviews.llvm.org/D159275
This is an ongoing series of commits that are reformatting our
Python code. This catches the last of the python files to
reformat. Since they where so few I bunched them together.
Reformatting is done with `black`.
If you end up having problems merging this commit because you
have made changes to a python file, the best way to handle that
is to run git checkout --ours <yourfile> and then reformat it
with black.
If you run into any problems, post to discourse about it and
we will try to help.
RFC Thread below:
https://discourse.llvm.org/t/rfc-document-and-standardize-python-code-style
Reviewed By: jhenderson, #libc, Mordante, sivachandra
Differential Revision: https://reviews.llvm.org/D150784
The rounding during type conversion uses multiple conversions, selecting
between them to try to discover if rounding occurred. This appears to
not have been tested, since it would generate code of the form:
float convert_float_rtp(char x)
{
float r = convert_float(x);
char y = convert_char(y);
[...]
}
which will access uninitialised data. The idea appears to have been to
have done a char -> float -> char roundtrip in order to discover the
rounding, so do this.
Discovered by inspection.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed By: jvesely
Differential Revision: https://reviews.llvm.org/D81999
The script's shebang wants Python 3, so we use FindPython3. The
original code didn't work when an unversioned python was not available.
This is explicitly allowed in PEP 394. ("Distributors may choose to set
the behavior of the python command as follows: python2, python3, not
provide python command, allow python to be configurable by an end user
or a system administrator.")
Also I think it's actually required, so let the configuration fail if we
can't find it.
Lastly remove the shebang, since the script is only run via interpreter
and doesn't have the executable bit set anyway.
Reviewed By: jvesely
Differential Revision: https://reviews.llvm.org/D88366
Add targets to emit SPIR-V targeted to Mesa's OpenCL support, using
SPIR-V 1.1.
Substantially based on Dave Airlie's earlier work.
libclc: spirv: remove step/smoothstep apis not defined for SPIR-V
libclc: disable inlines for SPIR-V builds
Reviewed By: jvesely, tstellar, jenatali
Differential Revision: https://reviews.llvm.org/D77589
The SPIR spec states that all OpenCL built-in functions should be
overloadable and mangled, to ensure consistency.
Add the overload attribute to functions which were missing them:
work dimensions, memory barriers and fences, and events.
Reviewed By: tstellar, jenatali
Differential Revision: https://reviews.llvm.org/D82078
Fix FP_ILOGBNAN definition to match the opencl-c-base.h one and
guarantee that FP_ILOGBNAN and FP_ILOGB0 are different. Doing that
implies fixing ilogb() implementation to return the right value.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed By: jvesely
Differential Revision: https://reviews.llvm.org/D83473
Fixes a wimpy-mode CTS failure for asin(float).
Passes non-wimpy for both float/double on RX580.
Signed-off-by: Aaron Watry <awatry@gmail.com>
Tested-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
