From 511182eabb36f6ec9776e2b3c4d7e16d93d0ac0d Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Sat, 18 May 2024 13:40:39 +0300
Subject: [PATCH] android : use "ci-android" branch for CI (#7341)

* android : use "ci-android" branch for CI

* ggml : disable SIMD exp and silu for 32-bit ARM

ggml-ci

* android : do not fetch, use add_subdirectory instead

* cmake : provide binary dir
---
 .../app/src/main/cpp/CMakeLists.txt            | 18 ++++++++++--------
 ggml.c                                         |  6 +++---
 2 files changed, 13 insertions(+), 11 deletions(-)

diff --git a/examples/llama.android/app/src/main/cpp/CMakeLists.txt b/examples/llama.android/app/src/main/cpp/CMakeLists.txt
index 85139329a..688a4750d 100644
--- a/examples/llama.android/app/src/main/cpp/CMakeLists.txt
+++ b/examples/llama.android/app/src/main/cpp/CMakeLists.txt
@@ -12,15 +12,17 @@ cmake_minimum_required(VERSION 3.22.1)
 # build script scope).
 project("llama-android")
 
-include(FetchContent)
-FetchContent_Declare(
-        llama
-        GIT_REPOSITORY https://github.com/ggerganov/llama.cpp
-        GIT_TAG        master
-)
+#include(FetchContent)
+#FetchContent_Declare(
+#        llama
+#        GIT_REPOSITORY https://github.com/ggerganov/llama.cpp
+#        GIT_TAG        ci-android
+#)
+#
+## Also provides "common"
+#FetchContent_MakeAvailable(llama)
 
-# Also provides "common"
-FetchContent_MakeAvailable(llama)
+add_subdirectory(../../../../../../ please-work)
 
 # Creates and names a library, sets it as either STATIC
 # or SHARED, and provides the relative paths to its source code.
diff --git a/ggml.c b/ggml.c
index 55152bce4..a04c74ddd 100644
--- a/ggml.c
+++ b/ggml.c
@@ -2076,7 +2076,7 @@ inline static float ggml_silu_f32(float x) {
     return x/(1.0f + expf(-x));
 }
 
-#if defined(__ARM_NEON)
+#if defined(__ARM_NEON) && defined(__aarch64__)
 
 // adapted from arm limited optimized routine
 // the maximum error is 1.45358 plus 0.5 ulps
@@ -2288,7 +2288,7 @@ static void ggml_vec_silu_f32(const int n, float * y, const float * x) {
     for (; i + 3 < n; i += 4) {
         _mm_storeu_ps(y + i, ggml_v_silu(_mm_loadu_ps(x + i)));
     }
-#elif defined(__ARM_NEON)
+#elif defined(__ARM_NEON) && defined(__aarch64__)
     for (; i + 3 < n; i += 4) {
         vst1q_f32(y + i, ggml_v_silu(vld1q_f32(x + i)));
     }
@@ -2335,7 +2335,7 @@ static ggml_float ggml_vec_soft_max_f32(const int n, float * y, const float * x,
 #endif
         sum += (ggml_float)_mm_cvtss_f32(val);
     }
-#elif defined(__ARM_NEON)
+#elif defined(__ARM_NEON) && defined(__aarch64__)
     for (; i + 3 < n; i += 4) {
         float32x4_t val = ggml_v_expf(vsubq_f32(vld1q_f32(x + i),
                                                 vdupq_n_f32(max)));