diff --git a/.gitignore b/.gitignore
index b862a0415..1f841c830 100644
--- a/.gitignore
+++ b/.gitignore
@@ -52,6 +52,7 @@ models-mnt
 /server
 /simple
 /speculative
+/parallel
 /train-text-from-scratch
 /vdot
 build-info.h
diff --git a/Makefile b/Makefile
index dc8ae3807..9b631c2a5 100644
--- a/Makefile
+++ b/Makefile
@@ -1,5 +1,5 @@
 # Define the default target now so that it is always the first target
-BUILD_TARGETS = main quantize quantize-stats perplexity embedding vdot train-text-from-scratch convert-llama2c-to-ggml simple save-load-state server embd-input-test gguf llama-bench baby-llama beam-search speculative tests/test-c.o
+BUILD_TARGETS = main quantize quantize-stats perplexity embedding vdot train-text-from-scratch convert-llama2c-to-ggml simple save-load-state server embd-input-test gguf llama-bench baby-llama beam-search speculative parallel tests/test-c.o
 
 # Binaries only useful for tests
 TEST_TARGETS = tests/test-llama-grammar tests/test-grammar-parser tests/test-double-float tests/test-grad0 tests/test-opt tests/test-quantize-fns tests/test-quantize-perf tests/test-sampling tests/test-tokenizer-0-llama tests/test-tokenizer-0-falcon tests/test-tokenizer-1-llama
@@ -563,6 +563,9 @@ beam-search: examples/beam-search/beam-search.cpp build-info.h ggml.o llama.o co
 speculative: examples/speculative/speculative.cpp build-info.h ggml.o llama.o common.o grammar-parser.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
 
+parallel: examples/parallel/parallel.cpp build-info.h ggml.o llama.o common.o $(OBJS)
+	$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
+
 ifdef LLAMA_METAL
 metal: examples/metal/metal.cpp ggml.o $(OBJS)
 	$(CXX) $(CXXFLAGS) $^ -o $@ $(LDFLAGS)
diff --git a/llama.cpp b/llama.cpp
index df0b39bfb..abfc16c1a 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1318,7 +1318,7 @@ static bool llama_kv_cache_find_slot(
 }
 
 // find how many cells are currently in use
-int32_t llama_kv_cache_cell_max(const struct llama_kv_cache & cache) {
+static int32_t llama_kv_cache_cell_max(const struct llama_kv_cache & cache) {
     for (uint32_t i = cache.size - 2; i > 0; --i) {
         if (cache.cells[i].pos >= 0 && !cache.cells[i].seq_id.empty()) {
             return i + 1;
@@ -1328,7 +1328,7 @@ int32_t llama_kv_cache_cell_max(const struct llama_kv_cache & cache) {
     return 0;
 }
 
-void llama_kv_cache_rm_tokens(struct llama_kv_cache & cache, int32_t c0, int32_t c1) {
+static void llama_kv_cache_rm_tokens(struct llama_kv_cache & cache, int32_t c0, int32_t c1) {
     if (c0 < 0) c0 = 0;
     if (c1 < 0) c1 = cache.size;
 
@@ -1338,7 +1338,7 @@ void llama_kv_cache_rm_tokens(struct llama_kv_cache & cache, int32_t c0, int32_t
     }
 }
 
-void llama_kv_cache_rm_seq(
+static void llama_kv_cache_rm_seq(
              struct llama_kv_cache & cache,
                       llama_seq_id   seq_id,
                          llama_pos   p0,
@@ -1353,7 +1353,7 @@ void llama_kv_cache_rm_seq(
     }
 }
 
-void llama_kv_cache_keep_seq(struct llama_kv_cache & cache, llama_seq_id seq_id) {
+static void llama_kv_cache_keep_seq(struct llama_kv_cache & cache, llama_seq_id seq_id) {
     for (uint32_t i = 0; i < cache.size; ++i) {
         if (!cache.cells[i].has_seq_id(seq_id)) {
             cache.cells[i].pos = -1;
@@ -1362,7 +1362,7 @@ void llama_kv_cache_keep_seq(struct llama_kv_cache & cache, llama_seq_id seq_id)
     }
 }
 
-void llama_kv_cache_shift_seq(
+static void llama_kv_cache_shift_seq(
              struct llama_kv_cache & cache,
                       llama_seq_id   seq_id,
                          llama_pos   p0,