diff --git a/ggml/include/ggml-alloc.h b/ggml/include/ggml-alloc.h
index 434c13b34a929c..f5d5cf8c8d2c9d 100644
--- a/ggml/include/ggml-alloc.h
+++ b/ggml/include/ggml-alloc.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #pragma once
 
 #include "ggml.h"
@@ -74,3 +76,5 @@ GGML_API struct ggml_backend_buffer * ggml_backend_alloc_ctx_tensors(struct ggml
 #ifdef  __cplusplus
 }
 #endif
+
+#endif
\ No newline at end of file
diff --git a/ggml/include/ggml-backend.h b/ggml/include/ggml-backend.h
index 5f3f1e286990e4..a280682fb79e22 100644
--- a/ggml/include/ggml-backend.h
+++ b/ggml/include/ggml-backend.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #pragma once
 
 #include "ggml.h"
@@ -236,3 +238,5 @@ extern "C" {
 #ifdef  __cplusplus
 }
 #endif
+
+#endif
\ No newline at end of file
diff --git a/ggml/src/ggml-aarch64.c b/ggml/src/ggml-aarch64.c
index 26535b1c432ba6..56ea5b2f3a2244 100644
--- a/ggml/src/ggml-aarch64.c
+++ b/ggml/src/ggml-aarch64.c
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 // SPDX-FileCopyrightText: Copyright 2024 Arm Ltd.
 #define GGML_COMMON_IMPL_C
 #include "ggml-common.h"
@@ -2191,3 +2193,5 @@ void ggml_gemm_q4_0_8x8_q8_0(int n, float * restrict s, size_t bs, const void *
     }
 #endif
 }
+
+#endif
\ No newline at end of file
diff --git a/ggml/src/ggml-aarch64.h b/ggml/src/ggml-aarch64.h
index 517babaf1691bb..a7e6c65fb537d2 100644
--- a/ggml/src/ggml-aarch64.h
+++ b/ggml/src/ggml-aarch64.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 // SPDX-FileCopyrightText: Copyright 2024 Arm Ltd.
 #pragma once
 
@@ -37,3 +39,4 @@ void ggml_gemm_q4_0_8x8_q8_0(int n, float * GGML_RESTRICT s, size_t bs, const vo
 }
 #endif
 
+#endif
\ No newline at end of file
diff --git a/ggml/src/ggml-alloc.c b/ggml/src/ggml-alloc.c
index e176b883e38c50..cbcbbaa6351858 100644
--- a/ggml/src/ggml-alloc.c
+++ b/ggml/src/ggml-alloc.c
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #include "ggml-alloc.h"
 #include "ggml-backend-impl.h"
 #include "ggml.h"
@@ -1040,3 +1042,5 @@ ggml_backend_buffer_t ggml_backend_alloc_ctx_tensors_from_buft(struct ggml_conte
 ggml_backend_buffer_t ggml_backend_alloc_ctx_tensors(struct ggml_context * ctx, ggml_backend_t backend) {
     return ggml_backend_alloc_ctx_tensors_from_buft(ctx, ggml_backend_get_default_buffer_type(backend));
 }
+
+#endif
\ No newline at end of file
diff --git a/ggml/src/ggml-backend.c b/ggml/src/ggml-backend.c
index d39cfed8886f42..2803e235a6dc2b 100644
--- a/ggml/src/ggml-backend.c
+++ b/ggml/src/ggml-backend.c
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #include "ggml-backend-impl.h"
 #include "ggml-alloc.h"
 #include "ggml-impl.h"
@@ -2232,3 +2234,5 @@ bool ggml_backend_compare_graph_backend(ggml_backend_t backend1, ggml_backend_t
 
     return true;
 }
+
+#endif
\ No newline at end of file
diff --git a/ggml/src/ggml-blas.cpp b/ggml/src/ggml-blas.cpp
index a37aa407282b94..7d4f24c4c09d17 100644
--- a/ggml/src/ggml-blas.cpp
+++ b/ggml/src/ggml-blas.cpp
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #include "ggml-blas.h"
 #include "ggml-backend-impl.h"
 
@@ -366,3 +368,5 @@ void ggml_backend_blas_set_n_threads(ggml_backend_t backend_blas, int n_threads)
     ggml_backend_blas_context * ctx = (ggml_backend_blas_context *)backend_blas->context;
     ctx->n_threads = n_threads;
 }
+
+#endif
\ No newline at end of file
diff --git a/ggml/src/ggml-cann.cpp b/ggml/src/ggml-cann.cpp
index 9bf7e332a7e06e..d69564a34ab9ac 100644
--- a/ggml/src/ggml-cann.cpp
+++ b/ggml/src/ggml-cann.cpp
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 /*
  * Copyright (c) 2023-2024 The ggml authors
  *
@@ -2021,3 +2023,5 @@ GGML_CALL int ggml_backend_cann_reg_devices() {
     }
     return device_count;
 }
+
+#endif
\ No newline at end of file
diff --git a/ggml/src/ggml-common.h b/ggml/src/ggml-common.h
index fafd5fa7ae000d..16de4fb4af0a60 100644
--- a/ggml/src/ggml-common.h
+++ b/ggml/src/ggml-common.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #ifndef GGML_COMMON_DECL
 
 #if defined(GGML_COMMON_DECL_C)
@@ -1827,3 +1829,5 @@ GGML_TABLE_END()
 
 #endif // GGML_COMMON_IMPL
 #endif // GGML_COMMON_IMPL
+
+#endif
\ No newline at end of file
diff --git a/ggml/src/ggml-impl.h b/ggml/src/ggml-impl.h
index a2c8dbec0824f0..747985ea783559 100644
--- a/ggml/src/ggml-impl.h
+++ b/ggml/src/ggml-impl.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #pragma once
 
 #include "ggml.h"
@@ -653,3 +655,5 @@ size_t ggml_hash_find_or_insert(      struct ggml_hash_set hash_set, struct ggml
 #ifdef __cplusplus
 }
 #endif
+
+#endif
\ No newline at end of file
diff --git a/ggml/src/ggml-metal.m b/ggml/src/ggml-metal.m
index 6a906fbf79a58a..93d0e51a504f60 100644
--- a/ggml/src/ggml-metal.m
+++ b/ggml/src/ggml-metal.m
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #import "ggml-metal.h"
 
 #import "ggml-backend-impl.h"
@@ -3267,3 +3269,5 @@ GGML_CALL ggml_backend_t ggml_backend_reg_metal_init(const char * params, void *
     GGML_UNUSED(params);
     GGML_UNUSED(user_data);
 }
+
+#endif
\ No newline at end of file
diff --git a/ggml/src/ggml-quants.c b/ggml/src/ggml-quants.c
index 47418597c00d8f..5cc9bec777aa2a 100644
--- a/ggml/src/ggml-quants.c
+++ b/ggml/src/ggml-quants.c
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #define GGML_COMMON_IMPL_C
 #include "ggml-common.h"
 
@@ -14858,3 +14860,5 @@ bool ggml_validate_row_data(enum ggml_type type, const void * data, size_t nbyte
 
     return true;
 }
+
+#endif
\ No newline at end of file
diff --git a/ggml/src/ggml-quants.h b/ggml/src/ggml-quants.h
index 88b1f3269646d8..c30125887eae63 100644
--- a/ggml/src/ggml-quants.h
+++ b/ggml/src/ggml-quants.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #pragma once
 
 #define GGML_COMMON_DECL_C
@@ -130,3 +132,5 @@ void iq3xs_free_impl(int grid_size);
 #ifdef __cplusplus
 }
 #endif
+
+#endif
diff --git a/ggml/src/ggml.c b/ggml/src/ggml.c
index dbb3a3ebe1cca4..d6770f8febd2a3 100644
--- a/ggml/src/ggml.c
+++ b/ggml/src/ggml.c
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #define _CRT_SECURE_NO_DEPRECATE // Disables ridiculous "unsafe" warnings on Windows
 #define _USE_MATH_DEFINES // For M_PI on MSVC
 
@@ -22042,3 +22044,5 @@ int ggml_cpu_has_matmul_int8(void) {
 }
 
 ////////////////////////////////////////////////////////////////////////////////
+
+#endif
diff --git a/ggml/src/llamafile/sgemm.cpp b/ggml/src/llamafile/sgemm.cpp
index 6626ceb26213f2..a5cfc33140378f 100644
--- a/ggml/src/llamafile/sgemm.cpp
+++ b/ggml/src/llamafile/sgemm.cpp
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 // Copyright 2024 Mozilla Foundation
 //
 // Permission is hereby granted, free of charge, to any person obtaining
@@ -1025,3 +1027,5 @@ bool llamafile_sgemm(int64_t m, int64_t n, int64_t k, const void *A, int64_t lda
     (void)Btype;
     (void)Ctype;
 }
+
+#endif
\ No newline at end of file
diff --git a/include/llama.h b/include/llama.h
index e68cd807e63bde..96ca61d2f9b064 100644
--- a/include/llama.h
+++ b/include/llama.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #ifndef LLAMA_H
 #define LLAMA_H
 
@@ -1231,3 +1233,5 @@ llama_token llama_sample_token_with_rng(struct llama_context * ctx, llama_token_
 #endif // LLAMA_API_INTERNAL
 
 #endif // LLAMA_H
+
+#endif
\ No newline at end of file
diff --git a/src/llama-grammar.cpp b/src/llama-grammar.cpp
index bd9322e2fab09b..812ef363b5bc5d 100644
--- a/src/llama-grammar.cpp
+++ b/src/llama-grammar.cpp
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #include "llama-grammar.h"
 
 #include "llama-vocab.h"
@@ -537,3 +539,5 @@ void llama_grammar_accept_token_impl(struct llama_grammar * grammar, const struc
 
     smpl->t_sample_us += ggml_time_us() - t_start_sample_us;
 }
+
+#endif
\ No newline at end of file
diff --git a/src/llama-grammar.h b/src/llama-grammar.h
index 8e578e09f8c2dc..f6e334fec7bd2b 100644
--- a/src/llama-grammar.h
+++ b/src/llama-grammar.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #pragma once
 
 #include "llama-impl.h"
@@ -39,3 +41,5 @@ void llama_grammar_accept_token_impl(
           const struct llama_vocab * vocab,
        const struct llama_sampling * smpl,
                        llama_token   token);
+
+#endif
\ No newline at end of file
diff --git a/src/llama-impl.h b/src/llama-impl.h
index dcc8c1c15a1b1e..fc911a03e53021 100644
--- a/src/llama-impl.h
+++ b/src/llama-impl.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #pragma once
 
 #define LLAMA_API_INTERNAL
@@ -24,3 +26,5 @@ void llama_log_callback_default(ggml_log_level level, const char * text, void *
 #define LLAMA_LOG_INFO(...)  llama_log_internal(GGML_LOG_LEVEL_INFO , __VA_ARGS__)
 #define LLAMA_LOG_WARN(...)  llama_log_internal(GGML_LOG_LEVEL_WARN , __VA_ARGS__)
 #define LLAMA_LOG_ERROR(...) llama_log_internal(GGML_LOG_LEVEL_ERROR, __VA_ARGS__)
+
+#endif
\ No newline at end of file
diff --git a/src/llama-sampling.cpp b/src/llama-sampling.cpp
index 8910f6d6542e91..76fdf3811dea49 100644
--- a/src/llama-sampling.cpp
+++ b/src/llama-sampling.cpp
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #include "llama-sampling.h"
 
 #include <algorithm>
@@ -633,3 +635,5 @@ llama_token llama_sample_token_with_rng_impl(struct llama_sampling * smpl, llama
 llama_token llama_sample_token_impl(struct llama_sampling * smpl, llama_token_data_array * candidates) {
     return llama_sample_token_with_rng_impl(smpl, candidates, smpl->rng);
 }
+
+#endif
\ No newline at end of file
diff --git a/src/llama-sampling.h b/src/llama-sampling.h
index f7f8e3ef706bc8..1965e847a5d073 100644
--- a/src/llama-sampling.h
+++ b/src/llama-sampling.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #pragma once
 
 #include "llama-impl.h"
@@ -54,3 +56,4 @@ llama_token llama_sample_token_greedy_impl     (struct llama_sampling * smpl, ll
 llama_token llama_sample_token_with_rng_impl   (struct llama_sampling * smpl, llama_token_data_array * candidates, std::mt19937 & rng);
 llama_token llama_sample_token_impl            (struct llama_sampling * smpl, llama_token_data_array * candidates);
 
+#endif
\ No newline at end of file
diff --git a/src/llama-vocab.cpp b/src/llama-vocab.cpp
index c482b36899a1cc..2a2562e5815872 100644
--- a/src/llama-vocab.cpp
+++ b/src/llama-vocab.cpp
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #include "llama-vocab.h"
 
 #include "unicode.h"
@@ -1719,3 +1721,5 @@ int32_t llama_detokenize_impl(
 
     return total <= text_len_max ? total : -total;
 }
+
+#endif
\ No newline at end of file
diff --git a/src/llama-vocab.h b/src/llama-vocab.h
index 30b565d55dad5b..911bbedffe4065 100644
--- a/src/llama-vocab.h
+++ b/src/llama-vocab.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #pragma once
 
 #include "llama-impl.h"
@@ -128,3 +130,5 @@ int32_t llama_detokenize_impl(
                          int32_t   text_len_max,
                             bool   remove_special,
                             bool   unparse_special);
+
+#endif
\ No newline at end of file
diff --git a/src/llama.cpp b/src/llama.cpp
index 40c5e8e8d4b3be..f5fee906f66cf8 100644
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #include "llama-impl.h"
 #include "llama-vocab.h"
 #include "llama-grammar.h"
@@ -19209,3 +19211,5 @@ void llama_log_callback_default(ggml_log_level level, const char * text, void *
     fputs(text, stderr);
     fflush(stderr);
 }
+
+#endif
\ No newline at end of file
diff --git a/src/unicode-data.cpp b/src/unicode-data.cpp
index 02bdf782380fe7..bed427306ceb43 100644
--- a/src/unicode-data.cpp
+++ b/src/unicode-data.cpp
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 // generated with scripts/gen-unicode-data.py
 
 #include "unicode-data.h"
@@ -7030,3 +7032,5 @@ const std::vector<range_nfd> unicode_ranges_nfd = {  // start, last, nfd
 {0x02FA1C, 0x02FA1C, 0x009F3B},
 {0x02FA1D, 0x02FA1D, 0x02A600},
 };
+
+#endif
diff --git a/src/unicode-data.h b/src/unicode-data.h
index e27fe1770710a3..1199bac807dc9e 100644
--- a/src/unicode-data.h
+++ b/src/unicode-data.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #pragma once
 
 #include <cstdint>
@@ -18,3 +20,5 @@ extern const std::unordered_set<uint32_t> unicode_set_whitespace;
 extern const std::unordered_map<uint32_t, uint32_t> unicode_map_lowercase;
 extern const std::unordered_map<uint32_t, uint32_t> unicode_map_uppercase;
 extern const std::vector<range_nfd> unicode_ranges_nfd;
+
+#endif
\ No newline at end of file
diff --git a/src/unicode.cpp b/src/unicode.cpp
index 46650bff06d15e..09671bcf2aaa40 100644
--- a/src/unicode.cpp
+++ b/src/unicode.cpp
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #if defined(_MSC_VER)
 #define _SILENCE_CXX17_CODECVT_HEADER_DEPRECATION_WARNING
 #endif
@@ -816,3 +818,5 @@ std::vector<std::string> unicode_regex_split(const std::string & text, const std
 
     return unicode_byte_encoding_process(bpe_words);
 }
+
+#endif
\ No newline at end of file
diff --git a/src/unicode.h b/src/unicode.h
index 008532a242ab8d..7f7d619923fa90 100644
--- a/src/unicode.h
+++ b/src/unicode.h
@@ -1,3 +1,5 @@
+#if defined(__arm64__)
+
 #pragma once
 
 #include <cstdint>
@@ -65,3 +67,5 @@ uint8_t unicode_utf8_to_byte(const std::string & utf8);
 uint32_t unicode_tolower(uint32_t cp);
 
 std::vector<std::string> unicode_regex_split(const std::string & text, const std::vector<std::string> & regex_exprs);
+
+#endif
\ No newline at end of file