diff --git a/ggml/src/ggml-rpc.cpp b/ggml/src/ggml-rpc.cpp index 7757615f5..6a9f51f51 100644 --- a/ggml/src/ggml-rpc.cpp +++ b/ggml/src/ggml-rpc.cpp @@ -58,7 +58,7 @@ struct socket_t { }; // ggml_tensor is serialized into rpc_tensor -#pragma pack(push, 1) +#pragma pack(1) struct rpc_tensor { uint64_t id; uint32_t type; @@ -76,23 +76,111 @@ struct rpc_tensor { char padding[4]; }; -#pragma pack(pop) static_assert(sizeof(rpc_tensor) % 8 == 0, "rpc_tensor size must be multiple of 8"); // RPC commands enum rpc_cmd { - ALLOC_BUFFER = 0, - GET_ALIGNMENT, - GET_MAX_SIZE, - BUFFER_GET_BASE, - FREE_BUFFER, - BUFFER_CLEAR, - SET_TENSOR, - GET_TENSOR, - COPY_TENSOR, - GRAPH_COMPUTE, - GET_DEVICE_MEMORY, + RPC_CMD_ALLOC_BUFFER = 0, + RPC_CMD_GET_ALIGNMENT, + RPC_CMD_GET_MAX_SIZE, + RPC_CMD_BUFFER_GET_BASE, + RPC_CMD_FREE_BUFFER, + RPC_CMD_BUFFER_CLEAR, + RPC_CMD_SET_TENSOR, + RPC_CMD_GET_TENSOR, + RPC_CMD_COPY_TENSOR, + RPC_CMD_GRAPH_COMPUTE, + RPC_CMD_GET_DEVICE_MEMORY, + RPC_CMD_INIT_TENSOR, + RPC_CMD_GET_ALLOC_SIZE, + RPC_CMD_COUNT, +}; + +#pragma pack(1) +struct rpc_msg_get_alloc_size_req { + rpc_tensor tensor; +}; + +#pragma pack(1) +struct rpc_msg_get_alloc_size_rsp { + uint64_t alloc_size; +}; + +#pragma pack(1) +struct rpc_msg_init_tensor_req { + rpc_tensor tensor; +}; + +#pragma pack(1) +struct rpc_msg_alloc_buffer_req { + uint64_t size; +}; + +#pragma pack(1) +struct rpc_msg_alloc_buffer_rsp { + uint64_t remote_ptr; + uint64_t remote_size; +}; + +#pragma pack(1) +struct rpc_msg_get_alignment_rsp { + uint64_t alignment; +}; + +#pragma pack(1) +struct rpc_msg_get_max_size_rsp { + uint64_t max_size; +}; + +#pragma pack(1) +struct rpc_msg_buffer_get_base_req { + uint64_t remote_ptr; +}; + +#pragma pack(1) +struct rpc_msg_buffer_get_base_rsp { + uint64_t base_ptr; +}; + +#pragma pack(1) +struct rpc_msg_free_buffer_req { + uint64_t remote_ptr; +}; + +#pragma pack(1) +struct rpc_msg_buffer_clear_req { + uint64_t remote_ptr; + uint8_t value; +}; + +#pragma pack(1) +struct rpc_msg_get_tensor_req { + rpc_tensor tensor; + uint64_t offset; + uint64_t size; +}; + +#pragma pack(1) +struct rpc_msg_copy_tensor_req { + rpc_tensor src; + rpc_tensor dst; +}; + +#pragma pack(1) +struct rpc_msg_copy_tensor_rsp { + uint8_t result; +}; + +#pragma pack(1) +struct rpc_msg_graph_compute_rsp { + uint8_t result; +}; + +#pragma pack(1) +struct rpc_msg_get_device_memory_rsp { + uint64_t free_mem; + uint64_t total_mem; }; // RPC data structures @@ -239,6 +327,38 @@ static bool recv_data(sockfd_t sockfd, void * data, size_t size) { return true; } +static bool send_msg(sockfd_t sockfd, const void * msg, size_t msg_size) { + if (!send_data(sockfd, &msg_size, sizeof(msg_size))) { + return false; + } + return send_data(sockfd, msg, msg_size); +} + +static bool recv_msg(sockfd_t sockfd, void * msg, size_t msg_size) { + uint64_t size; + if (!recv_data(sockfd, &size, sizeof(size))) { + return false; + } + if (size != msg_size) { + return false; + } + return recv_data(sockfd, msg, msg_size); +} + +static bool recv_msg(sockfd_t sockfd, std::vector & input) { + uint64_t size; + if (!recv_data(sockfd, &size, sizeof(size))) { + return false; + } + try { + input.resize(size); + } catch (const std::bad_alloc & e) { + fprintf(stderr, "Failed to allocate input buffer of size %" PRIu64 "\n", size); + return false; + } + return recv_data(sockfd, input.data(), size); +} + static bool parse_endpoint(const std::string & endpoint, std::string & host, int & port) { size_t pos = endpoint.find(':'); if (pos == std::string::npos) { @@ -251,28 +371,27 @@ static bool parse_endpoint(const std::string & endpoint, std::string & host, int // RPC request : | rpc_cmd (1 byte) | request_size (8 bytes) | request_data (request_size bytes) | // RPC response: | response_size (8 bytes) | response_data (response_size bytes) | -static bool send_rpc_cmd(const std::shared_ptr & sock, enum rpc_cmd cmd, const std::vector & input, std::vector & output) { +static bool send_rpc_cmd(const std::shared_ptr & sock, enum rpc_cmd cmd, const void * input, size_t input_size, void * output, size_t output_size) { uint8_t cmd_byte = cmd; if (!send_data(sock->fd, &cmd_byte, sizeof(cmd_byte))) { return false; } - uint64_t input_size = input.size(); if (!send_data(sock->fd, &input_size, sizeof(input_size))) { return false; } - if (!send_data(sock->fd, input.data(), input.size())) { + if (!send_data(sock->fd, input, input_size)) { return false; } - uint64_t output_size; - if (!recv_data(sock->fd, &output_size, sizeof(output_size))) { + // TODO: currently the output_size is always known, do we need support for commands with variable output size? + // even if we do, we can skip sending output_size from the server for commands with known output size + uint64_t out_size; + if (!recv_data(sock->fd, &out_size, sizeof(out_size))) { return false; } - if (output_size == 0) { - output.clear(); - return true; + if (out_size != output_size) { + return false; } - output.resize(output_size); - if (!recv_data(sock->fd, output.data(), output_size)) { + if (!recv_data(sock->fd, output, output_size)) { return false; } return true; @@ -325,14 +444,9 @@ GGML_CALL static const char * ggml_backend_rpc_buffer_get_name(ggml_backend_buff GGML_CALL static void ggml_backend_rpc_buffer_free_buffer(ggml_backend_buffer_t buffer) { ggml_backend_rpc_buffer_context * ctx = (ggml_backend_rpc_buffer_context *)buffer->context; - // input serialization format: | remote_ptr (8 bytes) | - std::vector input(sizeof(uint64_t), 0); - uint64_t remote_ptr = ctx->remote_ptr; - memcpy(input.data(), &remote_ptr, sizeof(remote_ptr)); - std::vector output; - bool status = send_rpc_cmd(ctx->sock, FREE_BUFFER, input, output); + rpc_msg_free_buffer_req request = {ctx->remote_ptr}; + bool status = send_rpc_cmd(ctx->sock, RPC_CMD_FREE_BUFFER, &request, sizeof(request), nullptr, 0); GGML_ASSERT(status); - GGML_ASSERT(output.empty()); delete ctx; } @@ -341,20 +455,13 @@ GGML_CALL static void * ggml_backend_rpc_buffer_get_base(ggml_backend_buffer_t b if (ctx->base_cache.find(buffer) != ctx->base_cache.end()) { return ctx->base_cache[buffer]; } - // input serialization format: | remote_ptr (8 bytes) | - std::vector input(sizeof(uint64_t), 0); - uint64_t remote_ptr = ctx->remote_ptr; - memcpy(input.data(), &remote_ptr, sizeof(remote_ptr)); - std::vector output; - bool status = send_rpc_cmd(ctx->sock, BUFFER_GET_BASE, input, output); + rpc_msg_buffer_get_base_req request = {ctx->remote_ptr}; + rpc_msg_buffer_get_base_rsp response; + bool status = send_rpc_cmd(ctx->sock, RPC_CMD_BUFFER_GET_BASE, &request, sizeof(request), &response, sizeof(response)); GGML_ASSERT(status); - GGML_ASSERT(output.size() == sizeof(uint64_t)); - // output serialization format: | base_ptr (8 bytes) | - uint64_t base_ptr; - memcpy(&base_ptr, output.data(), sizeof(base_ptr)); - void * base = reinterpret_cast(base_ptr); - ctx->base_cache[buffer] = base; - return base; + void * base_ptr = reinterpret_cast(response.base_ptr); + ctx->base_cache[buffer] = base_ptr; + return base_ptr; } static rpc_tensor serialize_tensor(const ggml_tensor * tensor) { @@ -388,10 +495,18 @@ static rpc_tensor serialize_tensor(const ggml_tensor * tensor) { } GGML_CALL static void ggml_backend_rpc_buffer_init_tensor(ggml_backend_buffer_t buffer, ggml_tensor * tensor) { - UNUSED(buffer); - if (ggml_is_quantized(tensor->type)) { - // TODO: this check is due to MATRIX_ROW_PADDING in CUDA and should be generalized - GGML_ASSERT(tensor->ne[0] % 512 == 0 && "unsupported quantized tensor"); + ggml_backend_rpc_buffer_context * ctx = (ggml_backend_rpc_buffer_context *)buffer->context; + + // CUDA backend on the server pads everything to 512 due to CUDA limitations. + // Due to bandwidth constraints, we only call the server init tensor functions if necessary. + // In particular, only quantized tensors need padding + if (ggml_is_quantized(tensor->type) && (tensor->ne[0] % 512 != 0) && (tensor->view_src == nullptr)) { + rpc_msg_init_tensor_req request; + + request.tensor = serialize_tensor(tensor); + + bool status = send_rpc_cmd(ctx->sock, RPC_CMD_INIT_TENSOR, &request, sizeof(request), nullptr, 0); + GGML_ASSERT(status); } } @@ -404,26 +519,18 @@ GGML_CALL static void ggml_backend_rpc_buffer_set_tensor(ggml_backend_buffer_t b memcpy(input.data(), &rpc_tensor, sizeof(rpc_tensor)); memcpy(input.data() + sizeof(rpc_tensor), &offset, sizeof(offset)); memcpy(input.data() + sizeof(rpc_tensor) + sizeof(offset), data, size); - std::vector output; - bool status = send_rpc_cmd(ctx->sock, SET_TENSOR, input, output); + bool status = send_rpc_cmd(ctx->sock, RPC_CMD_SET_TENSOR, input.data(), input.size(), nullptr, 0); GGML_ASSERT(status); } GGML_CALL static void ggml_backend_rpc_buffer_get_tensor(ggml_backend_buffer_t buffer, const ggml_tensor * tensor, void * data, size_t offset, size_t size) { ggml_backend_rpc_buffer_context * ctx = (ggml_backend_rpc_buffer_context *)buffer->context; - // input serialization format: | rpc_tensor | offset (8 bytes) | size (8 bytes) | - int input_size = sizeof(rpc_tensor) + 2*sizeof(uint64_t); - std::vector input(input_size, 0); - rpc_tensor rpc_tensor = serialize_tensor(tensor); - memcpy(input.data(), &rpc_tensor, sizeof(rpc_tensor)); - memcpy(input.data() + sizeof(rpc_tensor), &offset, sizeof(offset)); - memcpy(input.data() + sizeof(rpc_tensor) + sizeof(offset), &size, sizeof(size)); - std::vector output; - bool status = send_rpc_cmd(ctx->sock, GET_TENSOR, input, output); + rpc_msg_get_tensor_req request; + request.tensor = serialize_tensor(tensor); + request.offset = offset; + request.size = size; + bool status = send_rpc_cmd(ctx->sock, RPC_CMD_GET_TENSOR, &request, sizeof(request), data, size); GGML_ASSERT(status); - GGML_ASSERT(output.size() == size); - // output serialization format: | data (size bytes) | - memcpy(data, output.data(), size); } GGML_CALL static bool ggml_backend_rpc_buffer_cpy_tensor(ggml_backend_buffer_t buffer, const ggml_tensor * src, ggml_tensor * dst) { @@ -436,30 +543,19 @@ GGML_CALL static bool ggml_backend_rpc_buffer_cpy_tensor(ggml_backend_buffer_t b return false; } ggml_backend_rpc_buffer_context * ctx = (ggml_backend_rpc_buffer_context *)buffer->context; - // input serialization format: | rpc_tensor src | rpc_tensor dst | - int input_size = 2*sizeof(rpc_tensor); - std::vector input(input_size, 0); - rpc_tensor rpc_src = serialize_tensor(src); - rpc_tensor rpc_dst = serialize_tensor(dst); - memcpy(input.data(), &rpc_src, sizeof(rpc_src)); - memcpy(input.data() + sizeof(rpc_src), &rpc_dst, sizeof(rpc_dst)); - std::vector output; - bool status = send_rpc_cmd(ctx->sock, COPY_TENSOR, input, output); + rpc_msg_copy_tensor_req request; + request.src = serialize_tensor(src); + request.dst = serialize_tensor(dst); + rpc_msg_copy_tensor_rsp response; + bool status = send_rpc_cmd(ctx->sock, RPC_CMD_COPY_TENSOR, &request, sizeof(request), &response, sizeof(response)); GGML_ASSERT(status); - // output serialization format: | result (1 byte) | - GGML_ASSERT(output.size() == 1); - return output[0]; + return response.result; } GGML_CALL static void ggml_backend_rpc_buffer_clear(ggml_backend_buffer_t buffer, uint8_t value) { ggml_backend_rpc_buffer_context * ctx = (ggml_backend_rpc_buffer_context *)buffer->context; - // serialization format: | bufptr (8 bytes) | value (1 byte) | - int input_size = sizeof(uint64_t) + sizeof(uint8_t); - std::vector input(input_size, 0); - memcpy(input.data(), &ctx->remote_ptr, sizeof(ctx->remote_ptr)); - memcpy(input.data() + sizeof(ctx->remote_ptr), &value, sizeof(value)); - std::vector output; - bool status = send_rpc_cmd(ctx->sock, BUFFER_CLEAR, input, output); + rpc_msg_buffer_clear_req request = {ctx->remote_ptr, value}; + bool status = send_rpc_cmd(ctx->sock, RPC_CMD_BUFFER_CLEAR, &request, sizeof(request), nullptr, 0); GGML_ASSERT(status); } @@ -482,25 +578,16 @@ GGML_CALL static const char * ggml_backend_rpc_buffer_type_name(ggml_backend_buf GGML_CALL static ggml_backend_buffer_t ggml_backend_rpc_buffer_type_alloc_buffer(ggml_backend_buffer_type_t buft, size_t size) { ggml_backend_rpc_buffer_type_context * buft_ctx = (ggml_backend_rpc_buffer_type_context *)buft->context; - // input serialization format: | size (8 bytes) | - int input_size = sizeof(uint64_t); - std::vector input(input_size, 0); - memcpy(input.data(), &size, sizeof(size)); - std::vector output; + rpc_msg_alloc_buffer_req request = {size}; + rpc_msg_alloc_buffer_rsp response; auto sock = get_socket(buft_ctx->endpoint); - bool status = send_rpc_cmd(sock, ALLOC_BUFFER, input, output); + bool status = send_rpc_cmd(sock, RPC_CMD_ALLOC_BUFFER, &request, sizeof(request), &response, sizeof(response)); GGML_ASSERT(status); - GGML_ASSERT(output.size() == 2*sizeof(uint64_t)); - // output serialization format: | remote_ptr (8 bytes) | remote_size (8 bytes) | - uint64_t remote_ptr; - memcpy(&remote_ptr, output.data(), sizeof(remote_ptr)); - size_t remote_size; - memcpy(&remote_size, output.data() + sizeof(uint64_t), sizeof(remote_size)); - if (remote_ptr != 0) { + if (response.remote_ptr != 0) { ggml_backend_buffer_t buffer = ggml_backend_buffer_init(buft, ggml_backend_rpc_buffer_interface, - new ggml_backend_rpc_buffer_context{sock, {}, remote_ptr, "RPC[" + std::string(buft_ctx->endpoint) + "]"}, - remote_size); + new ggml_backend_rpc_buffer_context{sock, {}, response.remote_ptr, "RPC[" + std::string(buft_ctx->endpoint) + "]"}, + response.remote_size); return buffer; } else { return nullptr; @@ -508,16 +595,10 @@ GGML_CALL static ggml_backend_buffer_t ggml_backend_rpc_buffer_type_alloc_buffer } static size_t get_alignment(const std::shared_ptr & sock) { - // input serialization format: | 0 bytes | - std::vector input; - std::vector output; - bool status = send_rpc_cmd(sock, GET_ALIGNMENT, input, output); + rpc_msg_get_alignment_rsp response; + bool status = send_rpc_cmd(sock, RPC_CMD_GET_ALIGNMENT, nullptr, 0, &response, sizeof(response)); GGML_ASSERT(status); - GGML_ASSERT(output.size() == sizeof(uint64_t)); - // output serialization format: | alignment (8 bytes) | - uint64_t alignment; - memcpy(&alignment, output.data(), sizeof(alignment)); - return alignment; + return response.alignment; } GGML_CALL static size_t ggml_backend_rpc_buffer_type_get_alignment(ggml_backend_buffer_type_t buft) { @@ -526,16 +607,10 @@ GGML_CALL static size_t ggml_backend_rpc_buffer_type_get_alignment(ggml_backend_ } static size_t get_max_size(const std::shared_ptr & sock) { - // input serialization format: | 0 bytes | - std::vector input; - std::vector output; - bool status = send_rpc_cmd(sock, GET_MAX_SIZE, input, output); + rpc_msg_get_max_size_rsp response; + bool status = send_rpc_cmd(sock, RPC_CMD_GET_MAX_SIZE, nullptr, 0, &response, sizeof(response)); GGML_ASSERT(status); - GGML_ASSERT(output.size() == sizeof(uint64_t)); - // output serialization format: | max_size (8 bytes) | - uint64_t max_size; - memcpy(&max_size, output.data(), sizeof(max_size)); - return max_size; + return response.max_size; } GGML_CALL static size_t ggml_backend_rpc_get_max_size(ggml_backend_buffer_type_t buft) { @@ -544,8 +619,23 @@ GGML_CALL static size_t ggml_backend_rpc_get_max_size(ggml_backend_buffer_type_t } GGML_CALL static size_t ggml_backend_rpc_buffer_type_get_alloc_size(ggml_backend_buffer_type_t buft, const ggml_tensor * tensor) { - UNUSED(buft); - return ggml_nbytes(tensor); + // See comments in init_tensor. + if (ggml_is_quantized(tensor->type) && (tensor->ne[0] % 512 != 0) && (tensor->view_src == nullptr)) { + ggml_backend_rpc_buffer_type_context * buft_ctx = (ggml_backend_rpc_buffer_type_context *)buft->context; + auto sock = get_socket(buft_ctx->endpoint); + + rpc_msg_get_alloc_size_req request; + + request.tensor = serialize_tensor(tensor); + + rpc_msg_get_alloc_size_rsp response; + bool status = send_rpc_cmd(sock, RPC_CMD_GET_ALLOC_SIZE, &request, sizeof(request), &response, sizeof(response)); + GGML_ASSERT(status); + + return response.alloc_size; + } else { + return ggml_nbytes(tensor); + } } static ggml_backend_buffer_type_i ggml_backend_rpc_buffer_type_interface = { @@ -620,12 +710,11 @@ GGML_CALL static enum ggml_status ggml_backend_rpc_graph_compute(ggml_backend_t ggml_backend_rpc_context * rpc_ctx = (ggml_backend_rpc_context *)backend->context; std::vector input; serialize_graph(cgraph, input); - std::vector output; + rpc_msg_graph_compute_rsp response; auto sock = get_socket(rpc_ctx->endpoint); - bool status = send_rpc_cmd(sock, GRAPH_COMPUTE, input, output); + bool status = send_rpc_cmd(sock, RPC_CMD_GRAPH_COMPUTE, input.data(), input.size(), &response, sizeof(response)); GGML_ASSERT(status); - GGML_ASSERT(output.size() == 1); - return (enum ggml_status)output[0]; + return (enum ggml_status)response.result; } GGML_CALL static bool ggml_backend_rpc_supports_op(ggml_backend_t backend, const ggml_tensor * op) { @@ -716,19 +805,11 @@ GGML_API GGML_CALL bool ggml_backend_is_rpc(ggml_backend_t backend) { } static void get_device_memory(const std::shared_ptr & sock, size_t * free, size_t * total) { - // input serialization format: | 0 bytes | - std::vector input; - std::vector output; - bool status = send_rpc_cmd(sock, GET_DEVICE_MEMORY, input, output); + rpc_msg_get_device_memory_rsp response; + bool status = send_rpc_cmd(sock, RPC_CMD_GET_DEVICE_MEMORY, nullptr, 0, &response, sizeof(response)); GGML_ASSERT(status); - GGML_ASSERT(output.size() == 2*sizeof(uint64_t)); - // output serialization format: | free (8 bytes) | total (8 bytes) | - uint64_t free_mem; - memcpy(&free_mem, output.data(), sizeof(free_mem)); - uint64_t total_mem; - memcpy(&total_mem, output.data() + sizeof(uint64_t), sizeof(total_mem)); - *free = free_mem; - *total = total_mem; + *free = response.free_mem; + *total = response.total_mem; } GGML_API GGML_CALL void ggml_backend_rpc_get_device_memory(const char * endpoint, size_t * free, size_t * total) { @@ -748,16 +829,18 @@ class rpc_server { rpc_server(ggml_backend_t backend) : backend(backend) {} ~rpc_server(); - bool alloc_buffer(const std::vector & input, std::vector & output); - void get_alignment(std::vector & output); - void get_max_size(std::vector & output); - bool buffer_get_base(const std::vector & input, std::vector & output); - bool free_buffer(const std::vector & input); - bool buffer_clear(const std::vector & input); + void alloc_buffer(const rpc_msg_alloc_buffer_req & request, rpc_msg_alloc_buffer_rsp & response); + void get_alignment(rpc_msg_get_alignment_rsp & response); + void get_max_size(rpc_msg_get_max_size_rsp & response); + bool buffer_get_base(const rpc_msg_buffer_get_base_req & request, rpc_msg_buffer_get_base_rsp & response); + bool free_buffer(const rpc_msg_free_buffer_req & request); + bool buffer_clear(const rpc_msg_buffer_clear_req & request); bool set_tensor(const std::vector & input); - bool get_tensor(const std::vector & input, std::vector & output); - bool copy_tensor(const std::vector & input, std::vector & output); - bool graph_compute(const std::vector & input, std::vector & output); + bool get_tensor(const rpc_msg_get_tensor_req & request, std::vector & response); + bool copy_tensor(const rpc_msg_copy_tensor_req & request, rpc_msg_copy_tensor_rsp & response); + bool graph_compute(const std::vector & input, rpc_msg_graph_compute_rsp & response); + bool init_tensor(const rpc_msg_init_tensor_req & request); + bool get_alloc_size(const rpc_msg_get_alloc_size_req & request, rpc_msg_get_alloc_size_rsp & response); private: ggml_tensor * deserialize_tensor(struct ggml_context * ctx, const rpc_tensor * tensor); @@ -771,80 +854,80 @@ class rpc_server { std::unordered_set buffers; }; -bool rpc_server::alloc_buffer(const std::vector & input, std::vector & output) { - // input serialization format: | size (8 bytes) | - if (input.size() != sizeof(uint64_t)) { +bool rpc_server::get_alloc_size(const rpc_msg_get_alloc_size_req & request, rpc_msg_get_alloc_size_rsp & response) { + ggml_backend_buffer_type_t buft; + struct ggml_init_params params { + /*.mem_size =*/ ggml_tensor_overhead(), + /*.mem_buffer =*/ NULL, + /*.no_alloc =*/ true, + }; + + struct ggml_context * ctx = ggml_init(params); + ggml_tensor * tensor = deserialize_tensor(ctx, &request.tensor); + + if (tensor == nullptr) { + GGML_PRINT_DEBUG("Null tensor pointer passed to server get_alloc_size function.\n"); + ggml_free(ctx); return false; } - uint64_t size; - memcpy(&size, input.data(), sizeof(size)); + + if (tensor->buffer == nullptr) { + //No buffer allocated. + buft = ggml_backend_get_default_buffer_type(backend); + } else { + buft = tensor->buffer->buft; + } + + response.alloc_size = ggml_backend_buft_get_alloc_size(buft,tensor); + + ggml_free(ctx); + return true; +} + +void rpc_server::alloc_buffer(const rpc_msg_alloc_buffer_req & request, rpc_msg_alloc_buffer_rsp & response) { ggml_backend_buffer_type_t buft = ggml_backend_get_default_buffer_type(backend); - ggml_backend_buffer_t buffer = ggml_backend_buft_alloc_buffer(buft, size); - uint64_t remote_ptr = 0; - uint64_t remote_size = 0; + ggml_backend_buffer_t buffer = ggml_backend_buft_alloc_buffer(buft, request.size); + response.remote_ptr = 0; + response.remote_size = 0; if (buffer != nullptr) { - remote_ptr = reinterpret_cast(buffer); - remote_size = buffer->size; - GGML_PRINT_DEBUG("[%s] size: %" PRIu64 " -> remote_ptr: %" PRIx64 ", remote_size: %" PRIu64 "\n", __func__, size, remote_ptr, remote_size); + response.remote_ptr = reinterpret_cast(buffer); + response.remote_size = buffer->size; + GGML_PRINT_DEBUG("[%s] size: %" PRIu64 " -> remote_ptr: %" PRIx64 ", remote_size: %" PRIu64 "\n", __func__, request.size, response.remote_ptr, response.remote_size); buffers.insert(buffer); } else { - GGML_PRINT_DEBUG("[%s] size: %" PRIu64 " -> failed\n", __func__, size); + GGML_PRINT_DEBUG("[%s] size: %" PRIu64 " -> failed\n", __func__, request.size); } - // output serialization format: | remote_ptr (8 bytes) | remote_size (8 bytes) | - output.resize(2*sizeof(uint64_t), 0); - memcpy(output.data(), &remote_ptr, sizeof(remote_ptr)); - memcpy(output.data() + sizeof(uint64_t), &remote_size, sizeof(remote_size)); - return true; } -void rpc_server::get_alignment(std::vector & output) { +void rpc_server::get_alignment(rpc_msg_get_alignment_rsp & response) { ggml_backend_buffer_type_t buft = ggml_backend_get_default_buffer_type(backend); size_t alignment = ggml_backend_buft_get_alignment(buft); GGML_PRINT_DEBUG("[%s] alignment: %lu\n", __func__, alignment); - // output serialization format: | alignment (8 bytes) | - output.resize(sizeof(uint64_t), 0); - memcpy(output.data(), &alignment, sizeof(alignment)); + response.alignment = alignment; } -void rpc_server::get_max_size(std::vector & output) { +void rpc_server::get_max_size(rpc_msg_get_max_size_rsp & response) { ggml_backend_buffer_type_t buft = ggml_backend_get_default_buffer_type(backend); size_t max_size = ggml_backend_buft_get_max_size(buft); GGML_PRINT_DEBUG("[%s] max_size: %lu\n", __func__, max_size); - // output serialization format: | max_size (8 bytes) | - output.resize(sizeof(uint64_t), 0); - memcpy(output.data(), &max_size, sizeof(max_size)); + response.max_size = max_size; } -bool rpc_server::buffer_get_base(const std::vector & input, std::vector & output) { - // input serialization format: | remote_ptr (8 bytes) | - if (input.size() != sizeof(uint64_t)) { - return false; - } - uint64_t remote_ptr; - memcpy(&remote_ptr, input.data(), sizeof(remote_ptr)); - GGML_PRINT_DEBUG("[%s] remote_ptr: %" PRIx64 "\n", __func__, remote_ptr); - ggml_backend_buffer_t buffer = reinterpret_cast(remote_ptr); +bool rpc_server::buffer_get_base(const rpc_msg_buffer_get_base_req & request, rpc_msg_buffer_get_base_rsp & response) { + GGML_PRINT_DEBUG("[%s] remote_ptr: %" PRIx64 "\n", __func__, request.remote_ptr); + ggml_backend_buffer_t buffer = reinterpret_cast(request.remote_ptr); if (buffers.find(buffer) == buffers.end()) { GGML_PRINT_DEBUG("[%s] buffer not found\n", __func__); return false; } void * base = ggml_backend_buffer_get_base(buffer); - // output serialization format: | base_ptr (8 bytes) | - uint64_t base_ptr = reinterpret_cast(base); - output.resize(sizeof(uint64_t), 0); - memcpy(output.data(), &base_ptr, sizeof(base_ptr)); + response.base_ptr = reinterpret_cast(base); return true; } -bool rpc_server::free_buffer(const std::vector & input) { - // input serialization format: | remote_ptr (8 bytes) | - if (input.size() != sizeof(uint64_t)) { - return false; - } - uint64_t remote_ptr; - memcpy(&remote_ptr, input.data(), sizeof(remote_ptr)); - GGML_PRINT_DEBUG("[%s] remote_ptr: %" PRIx64 "\n", __func__, remote_ptr); - ggml_backend_buffer_t buffer = reinterpret_cast(remote_ptr); +bool rpc_server::free_buffer(const rpc_msg_free_buffer_req & request) { + GGML_PRINT_DEBUG("[%s] remote_ptr: %" PRIx64 "\n", __func__, request.remote_ptr); + ggml_backend_buffer_t buffer = reinterpret_cast(request.remote_ptr); if (buffers.find(buffer) == buffers.end()) { GGML_PRINT_DEBUG("[%s] buffer not found\n", __func__); return false; @@ -854,22 +937,14 @@ bool rpc_server::free_buffer(const std::vector & input) { return true; } -bool rpc_server::buffer_clear(const std::vector & input) { - // input serialization format: | remote_ptr (8 bytes) | value (1 byte) | - if (input.size() != sizeof(uint64_t) + sizeof(uint8_t)) { - return false; - } - uint64_t remote_ptr; - memcpy(&remote_ptr, input.data(), sizeof(remote_ptr)); - uint8_t value; - memcpy(&value, input.data() + sizeof(uint64_t), sizeof(value)); - GGML_PRINT_DEBUG("[%s] remote_ptr: %" PRIx64 ", value: %u\n", __func__, remote_ptr, value); - ggml_backend_buffer_t buffer = reinterpret_cast(remote_ptr); +bool rpc_server::buffer_clear(const rpc_msg_buffer_clear_req & request) { + GGML_PRINT_DEBUG("[%s] remote_ptr: %" PRIx64 ", value: %u\n", __func__, request.remote_ptr, request.value); + ggml_backend_buffer_t buffer = reinterpret_cast(request.remote_ptr); if (buffers.find(buffer) == buffers.end()) { GGML_PRINT_DEBUG("[%s] buffer not found\n", __func__); return false; } - ggml_backend_buffer_clear(buffer, value); + ggml_backend_buffer_clear(buffer, request.value); return true; } @@ -942,74 +1017,89 @@ bool rpc_server::set_tensor(const std::vector & input) { return true; } -bool rpc_server::get_tensor(const std::vector & input, std::vector & output) { - // serialization format: | rpc_tensor | offset (8 bytes) | size (8 bytes) | - if (input.size() != sizeof(rpc_tensor) + 2*sizeof(uint64_t)) { +bool rpc_server::init_tensor(const rpc_msg_init_tensor_req & request) { + struct ggml_init_params params { + /*.mem_size =*/ ggml_tensor_overhead(), + /*.mem_buffer =*/ NULL, + /*.no_alloc =*/ true, + }; + struct ggml_context * ctx = ggml_init(params); + ggml_tensor * tensor = deserialize_tensor(ctx, &request.tensor); + if (tensor == nullptr) { + GGML_PRINT_DEBUG("Null tensor pointer passed to server init_tensor function.\n"); + ggml_free(ctx); return false; } - const rpc_tensor * in_tensor = (const rpc_tensor *)input.data(); - uint64_t offset; - memcpy(&offset, input.data() + sizeof(rpc_tensor), sizeof(offset)); - uint64_t size; - memcpy(&size, input.data() + sizeof(rpc_tensor) + sizeof(offset), sizeof(size)); + // Call the backend's buffer_init_tensor function + ggml_backend_buffer_t buffer = tensor->buffer; + if (buffer && buffer->iface.init_tensor) { + buffer->iface.init_tensor(buffer, tensor); + } else { + GGML_PRINT_DEBUG("Null buffer for tensor passed to init_tensor function\n"); + } + + if (tensor->extra != nullptr) { + // This pointer can either be passed around client/server, or probably better stored server-side and kept track of. + // Currently unimplemented. + GGML_PRINT_DEBUG("tensor->extra populated by the backend, this is currently unsupported.\n"); + ggml_free(ctx); + return false; + } + + ggml_free(ctx); + return true; +} + +bool rpc_server::get_tensor(const rpc_msg_get_tensor_req & request, std::vector & response) { struct ggml_init_params params { /*.mem_size =*/ ggml_tensor_overhead(), /*.mem_buffer =*/ NULL, /*.no_alloc =*/ true, }; struct ggml_context * ctx = ggml_init(params); - ggml_tensor * tensor = deserialize_tensor(ctx, in_tensor); + ggml_tensor * tensor = deserialize_tensor(ctx, &request.tensor); if (tensor == nullptr) { GGML_PRINT_DEBUG("[%s] error deserializing tensor\n", __func__); ggml_free(ctx); return false; } - GGML_PRINT_DEBUG("[%s] buffer: %p, data: %p, offset: %" PRIu64 ", size: %" PRIu64 "\n", __func__, (void*)tensor->buffer, tensor->data, offset, size); + GGML_PRINT_DEBUG("[%s] buffer: %p, data: %p, offset: %" PRIu64 ", size: %" PRIu64 "\n", __func__, (void*)tensor->buffer, tensor->data, request.offset, request.size); // sanitize tensor->data { const size_t p0 = (size_t) ggml_backend_buffer_get_base(tensor->buffer); const size_t p1 = p0 + ggml_backend_buffer_get_size(tensor->buffer); - if (in_tensor->data + offset < p0 || in_tensor->data + offset >= p1 || size > (p1 - in_tensor->data - offset)) { - GGML_ABORT("[%s] tensor->data out of bounds\n", __func__); + if (request.tensor.data + request.offset < p0 || + request.tensor.data + request.offset >= p1 || + request.size > (p1 - request.tensor.data - request.offset)) { + GGML_ABORT("[%s] tensor->data out of bounds\n", __func__); } } - // output serialization format: | data (size bytes) | - output.resize(size, 0); - ggml_backend_tensor_get(tensor, output.data(), offset, size); + response.resize(request.size, 0); + ggml_backend_tensor_get(tensor, response.data(), request.offset, request.size); ggml_free(ctx); return true; } -bool rpc_server::copy_tensor(const std::vector & input, std::vector & output) { - // serialization format: | rpc_tensor src | rpc_tensor dst | - if (input.size() != 2*sizeof(rpc_tensor)) { - return false; - } - const rpc_tensor * rpc_src = (const rpc_tensor *)input.data(); - const rpc_tensor * rpc_dst = (const rpc_tensor *)(input.data() + sizeof(rpc_src)); - +bool rpc_server::copy_tensor(const rpc_msg_copy_tensor_req & request, rpc_msg_copy_tensor_rsp & response) { struct ggml_init_params params { /*.mem_size =*/ 2*ggml_tensor_overhead(), /*.mem_buffer =*/ NULL, /*.no_alloc =*/ true, }; struct ggml_context * ctx = ggml_init(params); - ggml_tensor * src = deserialize_tensor(ctx, rpc_src); - ggml_tensor * dst = deserialize_tensor(ctx, rpc_dst); + ggml_tensor * src = deserialize_tensor(ctx, &request.src); + ggml_tensor * dst = deserialize_tensor(ctx, &request.dst); if (src == nullptr || dst == nullptr) { GGML_PRINT_DEBUG("[%s] error deserializing tensors\n", __func__); ggml_free(ctx); return false; } GGML_PRINT_DEBUG("[%s] src->buffer: %p, dst->buffer: %p\n", __func__, (void*)src->buffer, (void*)dst->buffer); - bool result = ggml_backend_buffer_copy_tensor(src, dst); - // output serialization format: | result (1 byte) | - output.resize(1, 0); - output[0] = result; + response.result = ggml_backend_buffer_copy_tensor(src, dst); ggml_free(ctx); return true; } @@ -1038,7 +1128,7 @@ ggml_tensor * rpc_server::create_node(uint64_t id, return result; } -bool rpc_server::graph_compute(const std::vector & input, std::vector & output) { +bool rpc_server::graph_compute(const std::vector & input, rpc_msg_graph_compute_rsp & response) { // serialization format: // | n_nodes (4 bytes) | nodes (n_nodes * sizeof(uint64_t) | n_tensors (4 bytes) | tensors (n_tensors * sizeof(rpc_tensor)) | if (input.size() < sizeof(uint32_t)) { @@ -1078,9 +1168,7 @@ bool rpc_server::graph_compute(const std::vector & input, std::vectornodes[i] = create_node(id, ctx, tensor_ptrs, tensor_map); } ggml_status status = ggml_backend_graph_compute(backend, graph); - // output serialization format: | status (1 byte) | - output.resize(1, 0); - output[0] = status; + response.result = status; ggml_free(ctx); return true; } @@ -1098,80 +1186,183 @@ static void rpc_serve_client(ggml_backend_t backend, sockfd_t sockfd, size_t fre if (!recv_data(sockfd, &cmd, 1)) { break; } - std::vector input; - std::vector output; - uint64_t input_size; - if (!recv_data(sockfd, &input_size, sizeof(input_size))) { - break; - } - input.resize(input_size); - if (!recv_data(sockfd, input.data(), input_size)) { + if (cmd >= RPC_CMD_COUNT) { + // fail fast if the command is invalid + fprintf(stderr, "Unknown command: %d\n", cmd); break; } - bool ok = true; switch (cmd) { - case ALLOC_BUFFER: { - ok = server.alloc_buffer(input, output); + case RPC_CMD_ALLOC_BUFFER: { + rpc_msg_alloc_buffer_req request; + if (!recv_msg(sockfd, &request, sizeof(request))) { + return; + } + rpc_msg_alloc_buffer_rsp response; + server.alloc_buffer(request, response); + if (!send_msg(sockfd, &response, sizeof(response))) { + return; + } break; } - case GET_ALIGNMENT: { - server.get_alignment(output); + case RPC_CMD_GET_ALLOC_SIZE: { + rpc_msg_get_alloc_size_req request; + if (!recv_msg(sockfd, &request, sizeof(request))) { + return; + } + rpc_msg_get_alloc_size_rsp response; + server.get_alloc_size(request, response); + if (!send_msg(sockfd, &response, sizeof(response))) { + return; + } break; } - case GET_MAX_SIZE: { - server.get_max_size(output); + case RPC_CMD_GET_ALIGNMENT: { + if (!recv_msg(sockfd, nullptr, 0)) { + return; + } + rpc_msg_get_alignment_rsp response; + server.get_alignment(response); + if (!send_msg(sockfd, &response, sizeof(response))) { + return; + } break; } - case BUFFER_GET_BASE: { - ok = server.buffer_get_base(input, output); + case RPC_CMD_GET_MAX_SIZE: { + if (!recv_msg(sockfd, nullptr, 0)) { + return; + } + rpc_msg_get_max_size_rsp response; + server.get_max_size(response); + if (!send_msg(sockfd, &response, sizeof(response))) { + return; + } break; } - case FREE_BUFFER: { - ok = server.free_buffer(input); + case RPC_CMD_BUFFER_GET_BASE: { + rpc_msg_buffer_get_base_req request; + if (!recv_msg(sockfd, &request, sizeof(request))) { + return; + } + rpc_msg_buffer_get_base_rsp response; + if (!server.buffer_get_base(request, response)) { + return; + } + if (!send_msg(sockfd, &response, sizeof(response))) { + return; + } break; } - case BUFFER_CLEAR: { - ok = server.buffer_clear(input); + case RPC_CMD_FREE_BUFFER: { + rpc_msg_free_buffer_req request; + if (!recv_msg(sockfd, &request, sizeof(request))) { + return; + } + if (!server.free_buffer(request)) { + return; + } + if (!send_msg(sockfd, nullptr, 0)) { + return; + } break; } - case SET_TENSOR: { - ok = server.set_tensor(input); + case RPC_CMD_BUFFER_CLEAR: { + rpc_msg_buffer_clear_req request; + if (!recv_msg(sockfd, &request, sizeof(request))) { + return; + } + if (!server.buffer_clear(request)) { + return; + } + if (!send_msg(sockfd, nullptr, 0)) { + return; + } break; } - case GET_TENSOR: { - ok = server.get_tensor(input, output); + case RPC_CMD_SET_TENSOR: { + std::vector input; + if (!recv_msg(sockfd, input)) { + return; + } + if (!server.set_tensor(input)) { + return; + } + if (!send_msg(sockfd, nullptr, 0)) { + return; + } break; } - case COPY_TENSOR: { - ok = server.copy_tensor(input, output); + case RPC_CMD_INIT_TENSOR: { + rpc_msg_init_tensor_req request; + if (!recv_msg(sockfd, &request,sizeof(request))) { + return; + } + if (!server.init_tensor(request)) { + return; + } + if (!send_msg(sockfd, nullptr, 0)) { + return; + } break; } - case GRAPH_COMPUTE: { - ok = server.graph_compute(input, output); + case RPC_CMD_GET_TENSOR: { + rpc_msg_get_tensor_req request; + if (!recv_msg(sockfd, &request, sizeof(request))) { + return; + } + std::vector response; + if (!server.get_tensor(request, response)) { + return; + } + if (!send_msg(sockfd, response.data(), response.size())) { + return; + } break; } - case GET_DEVICE_MEMORY: { - // output serialization format: | free (8 bytes) | total (8 bytes) | - output.resize(2*sizeof(uint64_t), 0); - memcpy(output.data(), &free_mem, sizeof(free_mem)); - memcpy(output.data() + sizeof(uint64_t), &total_mem, sizeof(total_mem)); + case RPC_CMD_COPY_TENSOR: { + rpc_msg_copy_tensor_req request; + if (!recv_msg(sockfd, &request, sizeof(request))) { + return; + } + rpc_msg_copy_tensor_rsp response; + if (!server.copy_tensor(request, response)) { + return; + } + if (!send_msg(sockfd, &response, sizeof(response))) { + return; + } + break; + } + case RPC_CMD_GRAPH_COMPUTE: { + std::vector input; + if (!recv_msg(sockfd, input)) { + return; + } + rpc_msg_graph_compute_rsp response; + if (!server.graph_compute(input, response)) { + return; + } + if (!send_msg(sockfd, &response, sizeof(response))) { + return; + } + break; + } + case RPC_CMD_GET_DEVICE_MEMORY: { + if (!recv_msg(sockfd, nullptr, 0)) { + return; + } + rpc_msg_get_device_memory_rsp response; + response.free_mem = free_mem; + response.total_mem = total_mem; + if (!send_msg(sockfd, &response, sizeof(response))) { + return; + } break; } default: { fprintf(stderr, "Unknown command: %d\n", cmd); - ok = false; + return; } } - if (!ok) { - break; - } - uint64_t output_size = output.size(); - if (!send_data(sockfd, &output_size, sizeof(output_size))) { - break; - } - if (!send_data(sockfd, output.data(), output_size)) { - break; - } } } @@ -1203,8 +1394,10 @@ void start_rpc_server(ggml_backend_t backend, const char * endpoint, size_t free return; } printf("Accepted client connection, free_mem=%zu, total_mem=%zu\n", free_mem, total_mem); + fflush(stdout); rpc_serve_client(backend, client_socket->fd, free_mem, total_mem); printf("Client connection closed\n"); + fflush(stdout); } #ifdef _WIN32 WSACleanup(); diff --git a/ggml/src/iqk/iqk_mul_mat.cpp b/ggml/src/iqk/iqk_mul_mat.cpp index 0955f15d6..9b0e8d8a2 100644 --- a/ggml/src/iqk/iqk_mul_mat.cpp +++ b/ggml/src/iqk/iqk_mul_mat.cpp @@ -403,10 +403,10 @@ struct MulMat { } -bool iqk_mul_mat(long Nx, long Ny, long ne00, - int typeA, const void * A, long strideA, - int typeB, const void * B, long strideB, - float * C, long stride_C, int ith, int nth) { +bool iqk_mul_mat(long long Nx, long long Ny, long long ne00, + int typeA, const void * A, long long strideA, + int typeB, const void * B, long long strideB, + float * C, long long stride_C, int ith, int nth) { MulMat mm; if (!MulMat::prepare(typeA, typeB, ne00, mm, Ny)) { @@ -440,12 +440,12 @@ inline uint32_t simple_gcd(uint32_t a, uint32_t b) { } } -bool iqk_mul_mat_4d(long Nx, long Ny, long ne00, - long ne02, long ne03, long ne12, long ne13, - long nb02, long nb03, long nb12, long nb13, long nb2, long nb3, - int typeA, const void * A, long strideA, - int typeB, const void * B, long strideB, - float * C, long stride_C, int ith, int nth) { +bool iqk_mul_mat_4d(long long Nx, long long Ny, long long ne00, + long long ne02, long long ne03, long long ne12, long long ne13, + long long nb02, long long nb03, long long nb12, long long nb13, long long nb2, long long nb3, + int typeA, const void * A, long long strideA, + int typeB, const void * B, long long strideB, + float * C, long long stride_C, int ith, int nth) { auto r2 = ne12 / ne02; auto r3 = ne13 / ne03; @@ -500,10 +500,10 @@ bool iqk_mul_mat_4d(long Nx, long Ny, long ne00, return true; } -bool iqk_mul_mat_moe(long Nx, long Ny, long ne00, int ne11, - int typeA, const void * A, long strideA, - int typeB, const void * B, long strideB, - float * C, long nb1, long nb2, const void * vrow_mapping, int ith, int nth) { +bool iqk_mul_mat_moe(long long Nx, long long Ny, long long ne00, int ne11, + int typeA, const void * A, long long strideA, + int typeB, const void * B, long long strideB, + float * C, long long nb1, long long nb2, const void * vrow_mapping, int ith, int nth) { const mmid_row_mapping * row_mapping = (const mmid_row_mapping *)vrow_mapping; assert(row_mapping != nullptr); @@ -17340,20 +17340,20 @@ bool iqk_flash_attn_noalibi(int int_type_k, // type of k #else // IQK_IMPLEMENT -bool iqk_mul_mat(int, long, long, long, int, const void *, long, int, const void *, long, float *, long, int, int) { +bool iqk_mul_mat(int, long long, long long, long long, int, const void *, long long, int, const void *, long long, float *, long long, int, int) { return false; } -bool iqk_mul_mat_4d(long /*Nx*/, long /*Ny*/, long /*ne00*/, - long /*ne02*/, long /*ne03*/, long /*ne12*/, long /*ne13*/, - long /*nb02*/, long /*nb03*/, long /*nb12*/, long /*nb13*/, long /*nb2*/, long /*nb3*/, - int /*typeA*/, const void * /*A*/, long /*strideA*/, - int /*typeB*/, const void * /*B*/, long /*strideB*/, - float * /*C*/, long /*stride_C*/, int /*ith*/, int /*nth*/) { +bool iqk_mul_mat_4d(long long /*Nx*/, long long /*Ny*/, long long /*ne00*/, + long long /*ne02*/, long long /*ne03*/, long long /*ne12*/, long long /*ne13*/, + long long /*nb02*/, long long /*nb03*/, long long /*nb12*/, long long /*nb13*/, long long /*nb2*/, long long /*nb3*/, + int /*typeA*/, const void * /*A*/, long long /*strideA*/, + int /*typeB*/, const void * /*B*/, long long /*strideB*/, + float * /*C*/, long long /*stride_C*/, int /*ith*/, int /*nth*/) { return false; } -bool iqk_mul_mat_moe(long, long, long, int, int, const void *, long, int, const void *, long, float *, long, long, +bool iqk_mul_mat_moe(long long, long long, long long, int, int, const void *, long long, int, const void *, long long, float *, long long, long long, const void *, int, int) { return false; } diff --git a/ggml/src/iqk/iqk_mul_mat.h b/ggml/src/iqk/iqk_mul_mat.h index 767f89cff..825905f55 100644 --- a/ggml/src/iqk/iqk_mul_mat.h +++ b/ggml/src/iqk/iqk_mul_mat.h @@ -11,22 +11,22 @@ extern "C" { #endif -bool iqk_mul_mat(long Nx, long Ny, long ne00, - int typeA, const void * A, long strideA, - int typeB, const void * B, long strideB, - float * C, long stride_C, int ith, int nth); +bool iqk_mul_mat(long long Nx, long long Ny, long long ne00, + int typeA, const void * A, long long strideA, + int typeB, const void * B, long long strideB, + float * C, long long stride_C, int ith, int nth); -bool iqk_mul_mat_4d(long Nx, long Ny, long ne00, - long ne02, long ne03, long ne12, long ne13, - long nb02, long nb03, long nb12, long nb13, long nb2, long nb3, - int typeA, const void * A, long strideA, - int typeB, const void * B, long strideB, - float * C, long stride_C, int ith, int nth); +bool iqk_mul_mat_4d(long long Nx, long long Ny, long long ne00, + long long ne02, long long ne03, long long ne12, long long ne13, + long long nb02, long long nb03, long long nb12, long long nb13, long long nb2, long long nb3, + int typeA, const void * A, long long strideA, + int typeB, const void * B, long long strideB, + float * C, long long stride_C, int ith, int nth); -bool iqk_mul_mat_moe(long Nx, long Ny, long ne00, int ne11, - int typeA, const void * A, long strideA, - int typeB, const void * B, long strideB, - float * C, long nb1, long nb2, const void * vrow_mapping, int ith, int nth); +bool iqk_mul_mat_moe(long long Nx, long long Ny, long long ne00, int ne11, + int typeA, const void * A, long long strideA, + int typeB, const void * B, long long strideB, + float * C, long long nb1, long long nb2, const void * vrow_mapping, int ith, int nth); bool iqk_moe_fused_up_gate(long Nx, long Ny, long ne00, int ne11, int unary_op, int typeA, const void * Aup, const void * Agate, long strideA,