vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

GPU accel for rwkv is slow, disable it

Browse source
This commit is contained in:
Concedo 2023-07-02 00:41:46 +08:00
parent e1a7042943
commit ef3b8dc0d9
2 changed files with 4 additions and 46 deletions
Download patch file Download diff file Expand all files Collapse all files

8
gpttype_adapter.cpp
View file

@ -432,10 +432,10 @@ ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in
{ {
rwkv_ctx_v3 = rwkv_init_from_file(modelname.c_str(), n_threads); rwkv_ctx_v3 = rwkv_init_from_file(modelname.c_str(), n_threads);
if(inputs.gpulayers>0) // if(inputs.gpulayers>0)
{ // {
rwkv_gpu_offload_layers(rwkv_ctx_v3,inputs.gpulayers); // rwkv_gpu_offload_layers(rwkv_ctx_v3,inputs.gpulayers);
} // }
const struct rwkv_file_header & header = rwkv_ctx_v3->instance->model.header; const struct rwkv_file_header & header = rwkv_ctx_v3->instance->model.header;
const size_t n_vocab = header.n_vocab; const size_t n_vocab = header.n_vocab;

42
otherarch/rwkv_v3.cpp
View file

@ -6,13 +6,6 @@
#include "rwkv_v3.h" #include "rwkv_v3.h"
#include "ggml.h" #include "ggml.h"
#ifdef GGML_USE_CUBLAS
#include "ggml-cuda.h"
#endif
#if defined(GGML_USE_CLBLAST)
#include "ggml-opencl.h"
#endif
#include <string> #include <string>
#include <vector> #include <vector>
#include <cstring> #include <cstring>
@ -1065,11 +1058,7 @@ struct rwkv_future_tensor rwkv_future_graph_work(struct rwkv_future_ctx & ctx,
const size_t n_threads, const size_t n_threads,
const size_t sequence_len = 1 const size_t sequence_len = 1
) { ) {
#if defined(GGML_USE_CLBLAST) || defined(GGML_USE_CUBLAS)
enum ggml_type mul_mat_type = type == GGML_TYPE_F32 ? GGML_TYPE_F32 : GGML_TYPE_F16;
#else
enum ggml_type mul_mat_type = ggml_is_quantized(type) ? GGML_TYPE_Q8_1 : type; enum ggml_type mul_mat_type = ggml_is_quantized(type) ? GGML_TYPE_Q8_1 : type;
#endif
return ctx.alloc(GGML_TYPE_I8, rwkv_future_tensor::size(mul_mat_type, ffn_key_height, sequence_len) * n_threads + 64 * (n_threads - 1)); return ctx.alloc(GGML_TYPE_I8, rwkv_future_tensor::size(mul_mat_type, ffn_key_height, sequence_len) * n_threads + 64 * (n_threads - 1));
} }
@ -1556,38 +1545,7 @@ struct rwkv_context * rwkv_clone_context(struct rwkv_context * ctx, const uint32
} }
bool rwkv_gpu_offload_layers(struct rwkv_context * ctx, const uint32_t n_layers) { bool rwkv_gpu_offload_layers(struct rwkv_context * ctx, const uint32_t n_layers) {
#if defined(GGML_USE_CLBLAST) || defined(GGML_USE_CUBLAS)
printf("\nRWKV: Attempting offload of %u layers",n_layers);
const auto offload = [&](struct ggml_tensor * tensor) {
// TODO support multi-GPU
tensor->backend = GGML_BACKEND_GPU;
#if defined(GGML_USE_CLBLAST)
ggml_cl_transform_tensor(tensor->data, tensor);
#else
ggml_cuda_transform_tensor(tensor->data, tensor);
#endif
};
const size_t n_gpu = std::min(n_layers, ctx->instance->model.header.n_layer);
if (ctx->gpu_layers < n_gpu) {
for (size_t & i = ctx->gpu_layers; i < n_gpu; i++) {
const struct rwkv_layer & layer = ctx->instance->model.layers[i];
// TODO also offload other operations to GPU with ggml_cuda_assign_buffers
offload(layer.att_key);
offload(layer.att_value);
offload(layer.att_receptance);
offload(layer.att_output);
offload(layer.ffn_key);
offload(layer.ffn_value);
offload(layer.ffn_receptance);
}
return true;
}
#endif
return false; return false;
} }