| Commit message (Collapse) | Author | Age |
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Armv8.2-a provides a native half-precision floating point (__fp16 aka.
float16_t). This patch introduces
* __fp16 as underlying type of Eigen::half if this type is available
* the packet types Packet4hf and Packet8hf representing float16x4_t and
float16x8_t respectively
* packet-math for the above packets with corresponding scalar type Eigen::half
The packet-math functionality has been implemented by Ashutosh Sharma
<ashutosh.sharma@amperecomputing.com>.
This closes #1940.
|
| |
|
|
|
|
| |
float on Sigmoid function(https://gitlab.com/libeigen/eigen/-/blob/master/Eigen/src/Core/functors/UnaryFunctors.h#L990). As a result SYCL backend breaks since SYCL backend only supports packet operation for vectorized type float4 and double2. The issue has been fixed by adding scalar type float to packet operation pexp for SYCL backend.
|
|
|
|
| |
pexp_float and pexp<Packet16f>
|
| |
|
|
|
|
|
| |
It was only defined under one `#ifdef` case. This fixes the `packetmath_14`
test for MSVC.
|
|
|
|
| |
not available, and avoid undefined behavior in C++. Also mask off the sign bit when extracting the exponent.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The following commit seems to have introduced regressions in ROCm/HIP support.
https://gitlab.com/libeigen/eigen/-/commit/183a208212353ccf81a664d25dc7660b6269acdd
It causes some unit-tests to fail with the following error
```
...
Eigen/src/Core/GenericPacketMath.h:322:3: error: no member named 'bit_and' in the global namespace; did you mean 'std::bit_and'?
...
Eigen/src/Core/GenericPacketMath.h:329:3: error: no member named 'bit_or' in the global namespace; did you mean 'std::bit_or'?
...
Eigen/src/Core/GenericPacketMath.h:336:3: error: no member named 'bit_xor' in the global namespace; did you mean 'std::bit_xor'?
...
```
The error occurs because, when compiling the device code in HIP/CUDA, the compiler will pick up the some of the std functions (whose calls are prefixed by EIGEN_USING_STD) from the global namespace (i.e. use ::bit_xor instead of std::bit_xor). For this to work, those functions must be declared in the global namespace in the HIP/CUDA header files. The `bit_and`, `bit_or` and `bit_xor` routines are not declared in the HIP header file that contain the decls for the std math functions ( `math_functions.h` ), and this is the cause of the error above.
It seems that the newer HIP compilers do support the calling of `std::` math routines within device code, and the ideal fix here would have been to change all calls to std math functions in EIGEN to use the `std::` namespace (instead of the global namespace ), when compiling with HIP compiler. However it seems there was a recent commit to remove the EIGEN_USING_STD_MATH macro and collapse it uses into the EIGEN_USING_STD macro ( https://gitlab.com/libeigen/eigen/-/commit/4091f6b25c5ad0ca3f7c00bd82bfd7ca1bbedee3 ).
Replacing all std math calls will essentially require re-surrecting the EIGEN_USING_STD_MATH macro, so not choosing that option.
Also HIP compilers only have support std math calls within device code, and not all std functions (specifically not for malloc/free which are prefixed via EIGEN_USING_STD). So modyfing EIGEN_USE_STD implementation to use std:: namspace for HIP will not work either.
Hence going for the ugly solution of special casing the three calls that breaking the HIP compile, to explicitly use the std:: namespace
|
|
|
|
| |
BFloat16.h build with older compilers.
|
|
|
|
| |
plog<Packet16f> op with generic api
|
|
|
|
| |
for SSE/AVX/AVX512.
|
|
|
|
| |
`predux_any` with MSVC.
|
| |
|
|
|
|
| |
`predux_fmax_nan` that implement reductions with `PropagateNaN`, and `PropagateNumbers` semantics. Add (slow) generic implementations for most reductions.
|
| |
|
| |
|
|
|
|
| |
(almost) all packetmath tests with SSE, AVX, and AVX512.
|
| |
|
| |
|
|
|
|
|
|
| |
across platforms.
Change test to only test for NaN-propagation for pfmin/pfmax.
|
| |
|
| |
|
| |
|
|
|
|
| |
double arguments.
|
| |
|
|
|
|
|
|
|
|
| |
64 bit builds, see:
https://stackoverflow.com/questions/60933486/mmx-intrinsics-like-mm-cvtpd-pi32-not-found-with-msvc-2019-for-64bit-targets-c
Instead use the equivalent SSE2 intrinsics.
|
|
|
|
| |
non-sensical NaN-propagation rules for std::min std::max implemented by pmin and pmax in Eigen is a longstanding source og confusion and bug report. This change is a first step towards addressing it, as discussing in issue #564.
|
|
|
|
| |
definition for SSE. SSE does not support conversion between 64 bit integers and double and the existing implementation of casting between Packet2d and Packer2l results in undefined behavior when casting NaN to int. Since pldexp and pfdexp only manipulate exponent fields that fit in 32 bit, this change provides specializations that use existing instructions _mm_cvtpd_pi32 and _mm_cvtsi32_pd instead.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The current `test/geo_alignedbox` tests fail on 32-bit arm due to small floating-point errors.
In particular, the following is not guaranteed to hold:
```
IsometryTransform identity = IsometryTransform::Identity();
BoxType transformedC;
transformedC.extend(c.transformed(identity));
VERIFY(transformedC.contains(c));
```
since `c.transformed(identity)` is ever-so-slightly different from `c`. Instead, we replace this test with one that checks an identity transform is within floating-point precision of `c`.
Also updated the condition on `AlignedBox::transform(...)` to only accept `Affine`, `AffineCompact`, and `Isometry` modes explicitly. Otherwise, invalid combinations of modes would also incorrectly pass the assertion.
|
|
|
|
| |
Fixes #1995
|
| |
|
| |
|
|
|
|
| |
TypeCasting.h on platforms where uint64_t != unsigned long.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The following commit causes regressions in the ROCm/HIP support for Eigen
https://gitlab.com/libeigen/eigen/-/commit/e55182ac09885d7558adf75e9e230b051a721c18
I suspect the same breakages occur on the CUDA side too.
The above commit puts the EIGEN_CONSTEXPR attribute on `half_base` constructor. `half_base` is derived from `__half_raw`.
When compiling with GPU support, the definition of `__half_raw` gets picked up from the GPU Compiler specific header files (`hip_fp16.h`, `cuda_fp16.h`). Properly supporting the above commit would require adding the `constexpr` attribute to the `__half_raw` constructor (and other `*half*` routines) in those header files. While that is something we can explore in the future, for now we need to undo the above commit when compiling with GPU support, which is what this commit does.
This commit also reverts a small change in the `raw_uint16_to_half` routine made by the above commit. Similar to the case above, that change was leading to compile errors due to the fact that `__half_raw` has a different definition when compiling with DPU support.
|
|
|
|
| |
a: __i28d) ops with MSVC compiler
|
|
|
|
|
|
| |
available on 32 bit x86.
If SSE 4.1 is available use the faster _mm_extract_epi64 intrinsic.
|
| |
|
| |
|
|
|
|
|
|
| |
constants static const or constexpr.
Move macro definition EIGEN_CONSTEXPR to Core and make all methods in NumTraits constexpr when EIGEN_HASH_CONSTEXPR is 1.
|
|
|
| |
'vmvnq_u64' does not exist for some reason.
|
| |
|
| |
|
|
|
|
| |
Packet2d is not supported.
|
|
|
|
|
|
|
|
| |
CastXML simulates the preprocessors of other compilers, but actually
parses the translation unit with an internal Clang compiler.
Use the same `vld1q_u64` workaround that we do for Clang.
Fixes: #1979
|
| |
|
|
|
|
| |
the comments here have long been fixed. The workarounds were now detrimental because (1) they prevented using fused mul-add on Clang/ARM32 and (2) the unnecessary 'volatile' in 'asm volatile' prevented legitimate reordering by the compiler.
|