Hi.
Fairly heavy user of this on amd64 -current (Vulkan).
Many useful improvements including support for newer models (Gemini3,
QWEN3 etc.) and llama-server now exits gracefully from ^C.
I don't have much clue about ports, so apologies for any errors. No
current maintainer so sending directly to ports@ - hope that's okay.
(https://github.com/ggml-org/llama.cpp/pull/13541)
Percy.
Index: Makefile
===================================================================
RCS file: /cvs/ports/misc/llama.cpp/Makefile,v
diff -u -p -r1.7 Makefile
--- Makefile 18 Feb 2025 00:02:17 -0000 1.7
+++ Makefile 14 May 2025 16:05:05 -0000
@@ -8,7 +8,7 @@ COMMENT = LLM inference system
GH_ACCOUNT = ggerganov
GH_PROJECT = llama.cpp
-GH_TAGNAME = b4706
+GH_TAGNAME = b5372
PKGNAME = llama-cpp-0.0.${GH_TAGNAME:S/b//}
SHARED_LIBS += ggml-base 0.0
Index: distinfo
===================================================================
RCS file: /cvs/ports/misc/llama.cpp/distinfo,v
diff -u -p -r1.2 distinfo
--- distinfo 13 Feb 2025 12:21:58 -0000 1.2
+++ distinfo 14 May 2025 16:05:05 -0000
@@ -1,2 +1,2 @@
-SHA256 (llama.cpp-b4706.tar.gz) =
jpINppeW9Vu/jeqf9gnJPsZ1Hkpkj6YWOHbJSAcPwxc=
-SIZE (llama.cpp-b4706.tar.gz) = 20705861
+SHA256 (llama.cpp-b5372.tar.gz) =
28q/8fqCc/rtzo8qbUtQ6njW/Zs3mzaLH0ynCWIkuk8=
+SIZE (llama.cpp-b5372.tar.gz) = 21147804
Index: patches/patch-common_common_cpp
===================================================================
RCS file: patches/patch-common_common_cpp
diff -N patches/patch-common_common_cpp
--- /dev/null 1 Jan 1970 00:00:00 -0000
+++ patches/patch-common_common_cpp 14 May 2025 16:05:05 -0000
@@ -0,0 +1,12 @@
+Index: common/common.cpp
+--- common/common.cpp.orig
++++ common/common.cpp
+@@ -830,7 +830,7 @@ std::string fs_get_cache_directory() {
+ if (getenv("LLAMA_CACHE")) {
+ cache_directory = std::getenv("LLAMA_CACHE");
+ } else {
+-#if defined(__linux__) || defined(__FreeBSD__) || defined(_AIX)
++#if defined(__linux__) || defined(__FreeBSD__) || defined(_AIX) ||
defined(__OpenBSD__)
+ if (std::getenv("XDG_CACHE_HOME")) {
+ cache_directory = std::getenv("XDG_CACHE_HOME");
+ } else {
Index: patches/patch-tools_rpc_rpc-server_cpp
===================================================================
RCS file: patches/patch-tools_rpc_rpc-server_cpp
diff -N patches/patch-tools_rpc_rpc-server_cpp
--- /dev/null 1 Jan 1970 00:00:00 -0000
+++ patches/patch-tools_rpc_rpc-server_cpp 14 May 2025 16:05:05 -0000
@@ -0,0 +1,12 @@
+Index: tools/rpc/rpc-server.cpp
+--- tools/rpc/rpc-server.cpp.orig
++++ tools/rpc/rpc-server.cpp
+@@ -111,7 +111,7 @@ static std::string fs_get_cache_directory() {
+ if (getenv("LLAMA_CACHE")) {
+ cache_directory = std::getenv("LLAMA_CACHE");
+ } else {
+-#if defined(__linux__) || defined(__FreeBSD__) || defined(_AIX)
++#if defined(__linux__) || defined(__FreeBSD__) || defined(_AIX) ||
defined(__OpenBSD__)
+ if (std::getenv("XDG_CACHE_HOME")) {
+ cache_directory = std::getenv("XDG_CACHE_HOME");
+ } else {
Index: pkg/PLIST
===================================================================
RCS file: /cvs/ports/misc/llama.cpp/pkg/PLIST,v
diff -u -p -r1.3 PLIST
--- pkg/PLIST 13 Feb 2025 12:21:59 -0000 1.3
+++ pkg/PLIST 14 May 2025 16:05:05 -0000
@@ -8,28 +8,23 @@ bin/convert_hf_to_gguf.py
@bin bin/llama-embedding
@bin bin/llama-eval-callback
@bin bin/llama-export-lora
-@bin bin/llama-gbnf-validator
+@bin bin/llama-finetune
@bin bin/llama-gen-docs
@bin bin/llama-gguf
@bin bin/llama-gguf-hash
@bin bin/llama-gguf-split
@bin bin/llama-gritlm
@bin bin/llama-imatrix
-@bin bin/llama-infill
-@bin bin/llama-llava-cli
-@bin bin/llama-llava-clip-quantize-cli
@bin bin/llama-lookahead
@bin bin/llama-lookup
@bin bin/llama-lookup-create
@bin bin/llama-lookup-merge
@bin bin/llama-lookup-stats
-@bin bin/llama-minicpmv-cli
+@bin bin/llama-mtmd-cli
@bin bin/llama-parallel
@bin bin/llama-passkey
@bin bin/llama-perplexity
@bin bin/llama-quantize
-@bin bin/llama-quantize-stats
-@bin bin/llama-qwen2vl-cli
@bin bin/llama-retrieval
@bin bin/llama-run
@bin bin/llama-save-load-state
@@ -45,6 +40,7 @@ include/ggml-alloc.h
include/ggml-backend.h
include/ggml-blas.h
include/ggml-cann.h
+include/ggml-cpp.h
include/ggml-cpu.h
include/ggml-cuda.h
include/ggml-kompute.h
@@ -68,5 +64,5 @@ lib/cmake/llama/llama-version.cmake
@lib lib/libggml-vulkan.so.${LIBggml-vulkan_VERSION}
@lib lib/libggml.so.${LIBggml_VERSION}
@lib lib/libllama.so.${LIBllama_VERSION}
-@lib lib/libllava_shared.so.${LIBllava_shared_VERSION}
+@so lib/libmtmd_shared.so
lib/pkgconfig/llama.pc
Index: Makefile
===================================================================
RCS file: /cvs/ports/misc/llama.cpp/Makefile,v
diff -u -p -r1.7 Makefile
--- Makefile 18 Feb 2025 00:02:17 -0000 1.7
+++ Makefile 14 May 2025 16:05:05 -0000
@@ -8,7 +8,7 @@ COMMENT = LLM inference system
GH_ACCOUNT = ggerganov
GH_PROJECT = llama.cpp
-GH_TAGNAME = b4706
+GH_TAGNAME = b5372
PKGNAME = llama-cpp-0.0.${GH_TAGNAME:S/b//}
SHARED_LIBS += ggml-base 0.0
Index: distinfo
===================================================================
RCS file: /cvs/ports/misc/llama.cpp/distinfo,v
diff -u -p -r1.2 distinfo
--- distinfo 13 Feb 2025 12:21:58 -0000 1.2
+++ distinfo 14 May 2025 16:05:05 -0000
@@ -1,2 +1,2 @@
-SHA256 (llama.cpp-b4706.tar.gz) = jpINppeW9Vu/jeqf9gnJPsZ1Hkpkj6YWOHbJSAcPwxc=
-SIZE (llama.cpp-b4706.tar.gz) = 20705861
+SHA256 (llama.cpp-b5372.tar.gz) = 28q/8fqCc/rtzo8qbUtQ6njW/Zs3mzaLH0ynCWIkuk8=
+SIZE (llama.cpp-b5372.tar.gz) = 21147804
Index: patches/patch-common_common_cpp
===================================================================
RCS file: patches/patch-common_common_cpp
diff -N patches/patch-common_common_cpp
--- /dev/null 1 Jan 1970 00:00:00 -0000
+++ patches/patch-common_common_cpp 14 May 2025 16:05:05 -0000
@@ -0,0 +1,12 @@
+Index: common/common.cpp
+--- common/common.cpp.orig
++++ common/common.cpp
+@@ -830,7 +830,7 @@ std::string fs_get_cache_directory() {
+ if (getenv("LLAMA_CACHE")) {
+ cache_directory = std::getenv("LLAMA_CACHE");
+ } else {
+-#if defined(__linux__) || defined(__FreeBSD__) || defined(_AIX)
++#if defined(__linux__) || defined(__FreeBSD__) || defined(_AIX) || defined(__OpenBSD__)
+ if (std::getenv("XDG_CACHE_HOME")) {
+ cache_directory = std::getenv("XDG_CACHE_HOME");
+ } else {
Index: patches/patch-tools_rpc_rpc-server_cpp
===================================================================
RCS file: patches/patch-tools_rpc_rpc-server_cpp
diff -N patches/patch-tools_rpc_rpc-server_cpp
--- /dev/null 1 Jan 1970 00:00:00 -0000
+++ patches/patch-tools_rpc_rpc-server_cpp 14 May 2025 16:05:05 -0000
@@ -0,0 +1,12 @@
+Index: tools/rpc/rpc-server.cpp
+--- tools/rpc/rpc-server.cpp.orig
++++ tools/rpc/rpc-server.cpp
+@@ -111,7 +111,7 @@ static std::string fs_get_cache_directory() {
+ if (getenv("LLAMA_CACHE")) {
+ cache_directory = std::getenv("LLAMA_CACHE");
+ } else {
+-#if defined(__linux__) || defined(__FreeBSD__) || defined(_AIX)
++#if defined(__linux__) || defined(__FreeBSD__) || defined(_AIX) || defined(__OpenBSD__)
+ if (std::getenv("XDG_CACHE_HOME")) {
+ cache_directory = std::getenv("XDG_CACHE_HOME");
+ } else {
Index: pkg/PLIST
===================================================================
RCS file: /cvs/ports/misc/llama.cpp/pkg/PLIST,v
diff -u -p -r1.3 PLIST
--- pkg/PLIST 13 Feb 2025 12:21:59 -0000 1.3
+++ pkg/PLIST 14 May 2025 16:05:05 -0000
@@ -8,28 +8,23 @@ bin/convert_hf_to_gguf.py
@bin bin/llama-embedding
@bin bin/llama-eval-callback
@bin bin/llama-export-lora
-@bin bin/llama-gbnf-validator
+@bin bin/llama-finetune
@bin bin/llama-gen-docs
@bin bin/llama-gguf
@bin bin/llama-gguf-hash
@bin bin/llama-gguf-split
@bin bin/llama-gritlm
@bin bin/llama-imatrix
-@bin bin/llama-infill
-@bin bin/llama-llava-cli
-@bin bin/llama-llava-clip-quantize-cli
@bin bin/llama-lookahead
@bin bin/llama-lookup
@bin bin/llama-lookup-create
@bin bin/llama-lookup-merge
@bin bin/llama-lookup-stats
-@bin bin/llama-minicpmv-cli
+@bin bin/llama-mtmd-cli
@bin bin/llama-parallel
@bin bin/llama-passkey
@bin bin/llama-perplexity
@bin bin/llama-quantize
-@bin bin/llama-quantize-stats
-@bin bin/llama-qwen2vl-cli
@bin bin/llama-retrieval
@bin bin/llama-run
@bin bin/llama-save-load-state
@@ -45,6 +40,7 @@ include/ggml-alloc.h
include/ggml-backend.h
include/ggml-blas.h
include/ggml-cann.h
+include/ggml-cpp.h
include/ggml-cpu.h
include/ggml-cuda.h
include/ggml-kompute.h
@@ -68,5 +64,5 @@ lib/cmake/llama/llama-version.cmake
@lib lib/libggml-vulkan.so.${LIBggml-vulkan_VERSION}
@lib lib/libggml.so.${LIBggml_VERSION}
@lib lib/libllama.so.${LIBllama_VERSION}
-@lib lib/libllava_shared.so.${LIBllava_shared_VERSION}
+@so lib/libmtmd_shared.so
lib/pkgconfig/llama.pc
