rocm_jax/jaxlib/rocm/hipblas_kernels.h
Peter Hawkins 5617a02fa4 Remove JAX custom call implementation of batched triangular solve.
XLA supports batched triangular solve on GPU and has since February 2022, which is older than the minimum jaxlib version. We can therefore delete our implementation and just use XLA's implementation.

PiperOrigin-RevId: 482031830
2022-10-18 15:04:14 -07:00

58 lines
1.5 KiB
C++

/* Copyright 2021 The JAX Authors.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
==============================================================================*/
#ifndef JAXLIB_HIPBLAS_KERNELS_H_
#define JAXLIB_HIPBLAS_KERNELS_H_
#include <cstddef>
#include "rocm/include/hip/hip_runtime_api.h"
#include "rocm/include/hipblas.h"
#include "tensorflow/compiler/xla/service/custom_call_status.h"
namespace jax {
// Set of types known to Hipsolver.
enum class HipblasType {
F32,
F64,
C64,
C128,
};
// Batched LU decomposition: getrfbatched
struct GetrfBatchedDescriptor {
HipblasType type;
int batch, n;
};
void GetrfBatched(hipStream_t stream, void** buffers, const char* opaque,
size_t opaque_len, XlaCustomCallStatus* status);
// Batched QR decomposition: geqrfbatched
struct GeqrfBatchedDescriptor {
HipblasType type;
int batch, m, n;
};
void GeqrfBatched(hipStream_t stream, void** buffers, const char* opaque,
size_t opaque_len, XlaCustomCallStatus* status);
} // namespace jax
#endif // JAXLIB_HIPBLAS_KERNELS_H_