69091c5028
With these changes, we can now mark devices which are expected to work as performant as possible, and devices which were not optimized for some reason. For example, because the device was released after the Blender release, making it impossible for developers to optimize for devices in already released unchangeable code. This is primarily relevant for the LTS versions, which are supported for two years and require proper communication about optimization status for the new devices released during this time. This is implemented for oneAPI devices. Other device types currently are marked as optimized for compatibility with old behavior, but may implement the same in the future. Pull Request: https://projects.blender.org/blender/blender/pulls/139751
68 lines
2.6 KiB
C++
68 lines
2.6 KiB
C++
/* SPDX-FileCopyrightText: 2021-2022 Intel Corporation
|
|
*
|
|
* SPDX-License-Identifier: Apache-2.0 */
|
|
|
|
#pragma once
|
|
|
|
#ifdef WITH_ONEAPI
|
|
|
|
# include <stddef.h>
|
|
|
|
/* NOTE(@nsirgien): Should match underlying type in the declaration inside "kernel/types.h"
|
|
* TODO: use kernel/types.h directly. */
|
|
enum DeviceKernel : int;
|
|
|
|
# ifndef CYCLES_KERNEL_ONEAPI_EXPORT
|
|
# ifdef _WIN32
|
|
# if defined(ONEAPI_EXPORT)
|
|
# define CYCLES_KERNEL_ONEAPI_EXPORT extern __declspec(dllexport)
|
|
# else
|
|
# define CYCLES_KERNEL_ONEAPI_EXPORT extern __declspec(dllimport)
|
|
# endif
|
|
# else
|
|
# define CYCLES_KERNEL_ONEAPI_EXPORT extern __attribute__((visibility("default")))
|
|
# endif
|
|
# endif
|
|
|
|
class SyclQueue;
|
|
class SyclDevice;
|
|
|
|
typedef void (*OneAPIErrorCallback)(const char *error, void *user_ptr);
|
|
|
|
struct KernelContext {
|
|
/* Queue, associated with selected device */
|
|
SyclQueue *queue = nullptr;
|
|
/* Pointer to USM device memory with all global/constant allocation on this device */
|
|
void *kernel_globals = nullptr;
|
|
/* We needs this additional data for some kernels. */
|
|
int scene_max_shaders = 0;
|
|
};
|
|
|
|
/* Use extern C linking so that the symbols can be easily load from the dynamic library at runtime.
|
|
*/
|
|
# ifdef __cplusplus
|
|
extern "C" {
|
|
# endif
|
|
|
|
CYCLES_KERNEL_ONEAPI_EXPORT bool oneapi_run_test_kernel(SyclQueue *queue_);
|
|
CYCLES_KERNEL_ONEAPI_EXPORT bool oneapi_zero_memory_on_device(SyclQueue *queue_,
|
|
void *device_pointer,
|
|
size_t num_bytes);
|
|
CYCLES_KERNEL_ONEAPI_EXPORT void oneapi_set_error_cb(OneAPIErrorCallback cb, void *user_ptr);
|
|
CYCLES_KERNEL_ONEAPI_EXPORT size_t oneapi_suggested_gpu_kernel_size(const DeviceKernel kernel);
|
|
CYCLES_KERNEL_ONEAPI_EXPORT bool oneapi_enqueue_kernel(KernelContext *context,
|
|
const int kernel,
|
|
const size_t global_size,
|
|
const size_t local_size,
|
|
const unsigned int kernel_features,
|
|
bool use_hardware_raytracing,
|
|
void **args);
|
|
CYCLES_KERNEL_ONEAPI_EXPORT bool oneapi_load_kernels(SyclQueue *queue,
|
|
const unsigned int kernel_features,
|
|
bool use_hardware_raytracing);
|
|
# ifdef __cplusplus
|
|
}
|
|
|
|
# endif
|
|
#endif /* WITH_ONEAPI */
|