Direct I/O and Transparent HugePages

--direct-io for bypassing page cache (and using THP on Linux)

Up to 3-6x faster uncached loading, fewer pageouts, no page cache pollution.

common/common.h

@@ -160,6 +160,7 @@ struct gpt_params {
     bool instruct          = false; // instruction mode (used for Alpaca models)
     bool logits_all        = false; // return logits for all tokens in the batch
     bool use_mmap          = true;  // use mmap for faster loads
+    bool use_direct_io     = false; // use direct I/O
     bool use_mlock         = false; // use mlock to keep model in memory
     bool verbose_prompt    = false; // print prompt tokens before generation
     bool display_prompt    = true;  // print prompt before generation