Add Vulkan longhaul support

src/llama-longhaul.h

@@ -6,6 +6,7 @@
 #include <condition_variable>
 #include <cstdint>
 #include <deque>
+#include <map>
 #include <memory>
 #include <mutex>
 #include <string>
@@ -43,6 +44,8 @@ private:
         const llama_model_loader::longhaul_source * source;
         int32_t expert_id;
         int slot;
+        bool direct;
+        std::vector<uint8_t> staging;
         bool ok = false;
         std::string error;
     };
@@ -74,7 +77,6 @@ private:
     std::vector<int32_t> available_slots;
     std::vector<std::pair<int32_t, int32_t>> load_plan;
     std::vector<int32_t> id_buffer;
-    std::vector<uint8_t> read_buffer;
 
     std::vector<std::thread> io_workers;
     std::vector<io_job> io_jobs;
@@ -85,7 +87,10 @@ private:
     size_t io_pending = 0;
     bool io_stopping = false;
 
+    std::map<ggml_backend_dev_t, ggml_backend_ptr> upload_backends;
+
     bool source_is_direct(const llama_model_loader::longhaul_source & source) const;
+    ggml_backend_t upload_backend(const llama_model_loader::longhaul_source & source);
     bool load_plan_sources();
     void invalidate_plan(int layer);
     void io_worker();