From 4fbea0b196774e8d2babbaa76c60c7bcc40c997d Mon Sep 17 00:00:00 2001 From: Sadeep Madurange Date: Sat, 12 Sep 2026 09:27:59 +0800 Subject: Binaries for pi5. --- bin/deps-pi5/include/ggml-sycl.h | 57 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 57 insertions(+) create mode 100644 bin/deps-pi5/include/ggml-sycl.h (limited to 'bin/deps-pi5/include/ggml-sycl.h') diff --git a/bin/deps-pi5/include/ggml-sycl.h b/bin/deps-pi5/include/ggml-sycl.h new file mode 100644 index 0000000..418a7ba --- /dev/null +++ b/bin/deps-pi5/include/ggml-sycl.h @@ -0,0 +1,57 @@ +// +// MIT license +// Copyright (C) 2024 Intel Corporation +// SPDX-License-Identifier: MIT +// + +#pragma once + +#include "ggml.h" +#include "ggml-backend.h" + +#define GGML_SYCL_NAME "SYCL" +#define GGML_SYCL_MAX_DEVICES 48 + +#ifdef __cplusplus +extern "C" { +#endif + +// backend API +GGML_BACKEND_API ggml_backend_t ggml_backend_sycl_init(int device); + +GGML_BACKEND_API bool ggml_backend_is_sycl(ggml_backend_t backend); + +// devide buffer +GGML_BACKEND_API ggml_backend_buffer_type_t ggml_backend_sycl_buffer_type(int device); + +// split tensor buffer that splits matrices by rows across multiple devices +GGML_BACKEND_API ggml_backend_buffer_type_t ggml_backend_sycl_split_buffer_type(const float * tensor_split); + +// Tensor parallelism (--split-mode tensor): comm_init/free/allreduce_tensor +// trio queried by the meta-backend via ggml_backend_reg_get_proc_address. +// See typedefs in ggml/include/ggml-backend.h. Mirrors the CUDA backend's +// pattern (ggml_backend_cuda_comm_*). +GGML_BACKEND_API void * ggml_backend_sycl_comm_init(ggml_backend_t * backends, size_t n_backends); +GGML_BACKEND_API void ggml_backend_sycl_comm_free(void * comm_ctx); +GGML_BACKEND_API bool ggml_backend_sycl_comm_allreduce_tensor(void * comm_ctx, struct ggml_tensor ** tensors); + +// pinned host buffer for use with the CPU backend for faster copies between CPU and GPU +GGML_BACKEND_API ggml_backend_buffer_type_t ggml_backend_sycl_host_buffer_type(void); + +GGML_BACKEND_API void ggml_backend_sycl_print_sycl_devices(void); +GGML_BACKEND_API void ggml_backend_sycl_get_gpu_list(int *id_list, int max_len); +GGML_BACKEND_API void ggml_backend_sycl_get_device_description(int device, + char *description, + size_t description_size); +GGML_BACKEND_API int ggml_backend_sycl_get_device_count(); +GGML_BACKEND_API void ggml_backend_sycl_get_device_memory(int device, size_t *free, size_t *total); + +// SYCL doesn't support registering host memory, keep here for reference +// GGML_BACKEND_API bool ggml_backend_sycl_register_host_buffer(void * buffer, size_t size); +// GGML_BACKEND_API void ggml_backend_sycl_unregister_host_buffer(void * buffer); + +GGML_BACKEND_API ggml_backend_reg_t ggml_backend_sycl_reg(void); + +#ifdef __cplusplus +} +#endif -- cgit v1.2.3