Auto-endianness integrated!

Author: VyoJ

buildenv

@@ -7,6 +7,8 @@ export ZOPEN_DEV_TAG="master"
 export ZOPEN_NAME="llamacpp-master"
 export ZOPEN_RUNTIME_DEPS="ncurses"
 
+export ZOPEN_SKIP_ZOSLIB_ENV_HOOK=1
+
 rm -f "llama"
 ln -s "llama.cpp" "llama"
 ln -s "llama.cpp" $ZOPEN_NAME
@@ -17,7 +19,7 @@ export ZOPEN_COMP="CLANG"
 export BLAS_HOME="/usr/lpp/cbclib"
 
 export ZOPEN_CONFIGURE="cmake"
-export ZOPEN_CONFIGURE_OPTS="-B ../build --install-prefix \"\$ZOPEN_INSTALL_DIR/\" -DCURL_LIBRARY=\$CURL_HOME/lib/libcurl.a -DCURL_INCLUDE_DIR=\$CURL_HOME/include -DBUILD_SHARED_LIBS_DEFAULT=OFF -DBUILD_SHARED_LIBS=OFF -DGGML_STATIC=ON -DGGML_BACKEND_DL=OFF -DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS -DBLAS_INCLUDE_DIRS=\$BLAS_HOME/include/openblas -DBLAS_LIBRARIES=\$BLAS_HOME/lib/libopenblas.so -DLLAMA_BUILD_TESTS=ON ."
+export ZOPEN_CONFIGURE_OPTS="-B ../build --install-prefix \"\$ZOPEN_INSTALL_DIR/\" -DCMAKE_C_FLAGS=\"-fzvector -m64 -march=z15 -lmass.arch13\" -DCMAKE_C_STANDARD=11 -DCMAKE_C_STANDARD_REQUIRED=ON -DCMAKE_C_EXTENSIONS=OFF -DCMAKE_BUILD_TYPE=Release -DLLAMA_SERVER_SSL=ON -DOPENSSL_ROOT_DIR=\$OPENSSL_HOME -DCURL_LIBRARY=\$CURL_HOME/lib/libcurl.a -DCURL_INCLUDE_DIR=\$CURL_HOME/include -DBUILD_SHARED_LIBS_DEFAULT=OFF -DBUILD_SHARED_LIBS=OFF -DGGML_STATIC=ON -DGGML_BACKEND_DL=OFF -DGGML_OPENBLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS -DBLAS_INCLUDE_DIRS=/usr/lpp/cbclib/include/openblas -DBLAS_LIBRARIES=/usr/lpp/cbclib/lib/libopenblas.so -DLLAMA_BUILD_TESTS=ON -DLLAMA_CURL=ON ."
 
 export ZOPEN_MAKE="cmake"
 export ZOPEN_MAKE_OPTS="--build ../build --parallel \$ZOPEN_NUM_JOBS --config Release"
@@ -61,11 +63,27 @@ zopen_check_results()
   echo "expectedTotalTests:$total"
 }
 
+zopen_pre_check()
+{
+unset SSL_CERT_FILE
+unset SSL_CERT_PATH
+export CFLAGS="$CFLAGS -march=z15 -mzvector"
+export CPPFLAGS="$CPPFLAGS -march=z15"
+}
+
 zopen_append_to_env()
 {
+export SSL_CERT_PATH=$(curl-config --ca)
   # echo envars outside of PATH, MANPATH, LIBPATH
 }
 
+zopen_append_to_zoslib_env()
+{
+cat<<EOF
+SSL_CERT_PATH|set|$(curl-config --ca)
+EOF
+}
+
 zopen_append_to_setup()
 {
   # echo commands that will run when installing via setup.sh

patches/arg.cpp.patch

@@ -1,5 +1,5 @@
 diff --git a/common/arg.cpp b/common/arg.cpp
-index 40af7e57..46fec792 100644
+index 06005359..28220aac 100644
 --- a/common/arg.cpp
 +++ b/common/arg.cpp
 @@ -36,6 +36,9 @@
@@ -30,7 +30,17 @@ index 40af7e57..46fec792 100644
      http_headers.ptr = curl_slist_append(http_headers.ptr, "User-Agent: llama-cpp");
      // Check if hf-token or bearer-token was specified
      if (!bearer_token.empty()) {
-@@ -569,6 +574,7 @@ std::pair<long, std::vector<char>> common_remote_get_content(const std::string &
+@@ -506,7 +511,8 @@ static bool common_download_model(
+         struct gguf_init_params gguf_params = {
+             /*.no_alloc = */ true,
+             /*.ctx      = */ NULL,
+-        };
++            /* .allow_byteswapping = */ true,
++	};
+         auto * ctx_gguf = gguf_init_from_file(model.path.c_str(), gguf_params);
+         if (!ctx_gguf) {
+             LOG_ERR("\n%s:  failed to load input GGUF from %s\n", __func__, model.path.c_str());
+@@ -569,6 +575,7 @@ std::pair<long, std::vector<char>> common_remote_get_content(const std::string &
      curl_easy_setopt(curl.get(), CURLOPT_URL, url.c_str());
      curl_easy_setopt(curl.get(), CURLOPT_NOPROGRESS, 1L);
      curl_easy_setopt(curl.get(), CURLOPT_FOLLOWLOCATION, 1L);

patches/clip.cpp.patch

@@ -0,0 +1,12 @@
+diff --git a/tools/mtmd/clip.cpp b/tools/mtmd/clip.cpp
+index a4b62f9a..2d89d4b1 100644
+--- a/tools/mtmd/clip.cpp
++++ b/tools/mtmd/clip.cpp
+@@ -2028,6 +2028,7 @@ struct clip_model_loader {
+         struct gguf_init_params params = {
+             /*.no_alloc = */ true,
+             /*.ctx      = */ &meta,
++	    /*.allow_byteswapping = */ true,
+         };
+ 
+         ctx_gguf = gguf_context_ptr(gguf_init_from_file(fname, params));

patches/common.cpp.patch

@@ -1,8 +1,8 @@
 diff --git a/common/common.cpp b/common/common.cpp
-index 4cc40ed..234ad95 100644
+index d8c4d988..00aa7d43 100644
 --- a/common/common.cpp
 +++ b/common/common.cpp
-@@ -851,7 +851,7 @@ std::string fs_get_cache_directory() {
+@@ -874,7 +874,7 @@ std::string fs_get_cache_directory() {
      if (getenv("LLAMA_CACHE")) {
          cache_directory = std::getenv("LLAMA_CACHE");
      } else {
@@ -11,3 +11,11 @@ index 4cc40ed..234ad95 100644
          if (std::getenv("XDG_CACHE_HOME")) {
              cache_directory = std::getenv("XDG_CACHE_HOME");
          } else {
+@@ -1436,6 +1436,7 @@ static common_control_vector_data common_control_vector_load_one(const common_co
+     struct gguf_init_params meta_gguf_params = {
+         /* .no_alloc = */ false,
+         /* .ctx      = */ &ctx,
++	/* .allow_byteswapping = */ true,
+     };
+     struct gguf_context * ctx_gguf = gguf_init_from_file(load_info.fname.c_str(), meta_gguf_params);
+     if (!ctx_gguf) {

patches/convert-llama2c-to-ggml.cpp.patch

@@ -0,0 +1,12 @@
+diff --git a/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp b/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp
+index bdf0eed2..b36170bc 100644
+--- a/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp
++++ b/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp
+@@ -535,6 +535,7 @@ static void load_vocab(const char * filename, const Config * config, struct my_l
+         struct gguf_init_params params = {
+             /*.no_alloc = */ false,
+             /*.ctx      = */ &ctx_data,
++	    /*.allow_byteswapping = */ true,
+         };
+ 
+         struct gguf_context * ctx = gguf_init_from_file(filename, params);

patches/examples_gguf.cpp.patch

@@ -0,0 +1,12 @@
+diff --git a/examples/gguf/gguf.cpp b/examples/gguf/gguf.cpp
+index f31989c8..94ac3e5a 100644
+--- a/examples/gguf/gguf.cpp
++++ b/examples/gguf/gguf.cpp
+@@ -87,6 +87,7 @@ static bool gguf_ex_read_0(const std::string & fname) {
+     struct gguf_init_params params = {
+         /*.no_alloc = */ false,
+         /*.ctx      = */ NULL,
++	/*.allow_byteswapping = */ true,
+     };
+ 
+     struct gguf_context * ctx = gguf_init_from_file(fname.c_str(), params);

patches/export-lora.cpp.patch

@@ -0,0 +1,12 @@
+diff --git a/tools/export-lora/export-lora.cpp b/tools/export-lora/export-lora.cpp
+index f038019b..bf750ab9 100644
+--- a/tools/export-lora/export-lora.cpp
++++ b/tools/export-lora/export-lora.cpp
+@@ -50,6 +50,7 @@ static struct gguf_context * load_gguf(std::string & fname, struct ggml_context
+     struct gguf_init_params params = {
+         /*.no_alloc = */ true,
+         /*.ctx      = */ ctx_ggml,
++	/*.allow_byteswapping = */ true,
+     };
+     struct gguf_context * ctx_gguf = gguf_init_from_file(fname.c_str(), params);
+     if (!ctx_gguf) {

patches/ggml-backend-reg.cpp.patch

@@ -1,8 +1,8 @@
 diff --git a/ggml/src/ggml-backend-reg.cpp b/ggml/src/ggml-backend-reg.cpp
-index 405d8e3..b3682a9 100644
+index f0cdac31..29247c6a 100644
 --- a/ggml/src/ggml-backend-reg.cpp
 +++ b/ggml/src/ggml-backend-reg.cpp
-@@ -556,7 +556,9 @@ static ggml_backend_reg_t ggml_backend_load_best(const char * name, bool silent,
+@@ -561,7 +561,9 @@ static ggml_backend_reg_t ggml_backend_load_best(const char * name, bool silent,
  }
 
  void ggml_backend_load_all() {

patches/ggml-cpu.c.patch

@@ -1,5 +1,5 @@
 diff --git a/ggml/src/ggml-cpu/ggml-cpu.c b/ggml/src/ggml-cpu/ggml-cpu.c
-index c7426df..e074799 100644
+index c5271b77..6a547e4b 100644
 --- a/ggml/src/ggml-cpu/ggml-cpu.c
 +++ b/ggml/src/ggml-cpu/ggml-cpu.c
 @@ -17,7 +17,7 @@

patches/ggml-cpu.cpp.patch

@@ -1,5 +1,5 @@
 diff --git a/ggml/src/ggml-cpu/ggml-cpu.cpp b/ggml/src/ggml-cpu/ggml-cpu.cpp
-index e013e8b..6290b22 100644
+index c9daa4c3..63839d07 100644
 --- a/ggml/src/ggml-cpu/ggml-cpu.cpp
 +++ b/ggml/src/ggml-cpu/ggml-cpu.cpp
 @@ -33,6 +33,38 @@