vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 01:24:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fixes for sdcpp

Browse source
This commit is contained in:
Concedo 2025-04-12 10:08:23 +08:00
parent a0ae187563
commit 7e1289ade8
7 changed files with 26 additions and 16 deletions
Download patch file Download diff file Expand all files Collapse all files

18
examples/llava/clip.cpp
View file

@ -363,26 +363,26 @@ struct clip_ctx {
if(enable_gpu_clip)
{
#ifdef GGML_USE_CUDA
backend = ggml_backend_cuda_init(0);
backend = ggml_backend_ptr(ggml_backend_cuda_init(0));
LOG_INF("%s: CLIP using CUDA backend\n", __func__);
#endif
#ifdef GGML_USE_METAL
backend = ggml_backend_metal_init();
backend = ggml_backend_ptr(ggml_backend_metal_init());
LOG_INF("%s: CLIP using Metal backend\n", __func__);
#endif
#ifdef GGML_USE_VULKAN
backend = ggml_backend_vk_init(0);
backend = ggml_backend_ptr(ggml_backend_vk_init(0));
LOG_INF("%s: CLIP using Vulkan backend\n", __func__);
#endif
}
if (!backend) {
backend = ggml_backend_cpu_init();
backend = ggml_backend_ptr(ggml_backend_cpu_init());
LOG_INF("%s: CLIP using CPU backend\n", __func__);
}
backend_ptrs.push_back(backend);
backend_buft.push_back(ggml_backend_get_default_buffer_type(backend));
backend_ptrs.push_back(backend.get());
backend_buft.push_back(ggml_backend_get_default_buffer_type(backend.get()));
sched.reset(
ggml_backend_sched_new(backend_ptrs.data(), backend_buft.data(), backend_ptrs.size(), 8192, false)
@ -1228,7 +1228,7 @@ struct clip_model_loader {
// print gguf info
try {
std::string name;
get_string(KEY_NAME, name, false);
std::string description;
@ -2950,8 +2950,8 @@ bool clip_image_batch_encode(clip_ctx * ctx, const int n_threads, const clip_ima
if (window_mask) ggml_backend_tensor_set(window_mask, mask.data(), 0, ggml_nbytes(window_mask));
}
if (ggml_backend_is_cpu(ctx->backend)) {
ggml_backend_cpu_set_n_threads(ctx->backend, n_threads);
if (ggml_backend_is_cpu(ctx->backend.get())) {
ggml_backend_cpu_set_n_threads(ctx->backend.get(), n_threads);
}
auto status = ggml_backend_sched_graph_compute(ctx->sched.get(), gf);