Skip to content

Build fix for Linux and trivial typo fixes #137

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jul 3, 2024
Merged

Build fix for Linux and trivial typo fixes #137

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 10 additions & 10 deletions kleidiai-examples/llama_cpp/0001-Use-KleidiAI-Int4-Matmul-micro-kernels-in-llama.cpp.patch
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,9 +3,9 @@ From: Gian Marco Iodice <gianmarco.iodice@arm.com>
Date: Fri, 14 Jun 2024 14:55:47 +0100
Subject: [PATCH] Use KleidiAI Int4 Matmul micro-kernels in llama.cpp

- Update CMake file to fetch the Int4 micro-kernels from the KleidiAI
* Update CMake file to fetch the Int4 micro-kernels from the KleidiAI
repository
- Implement a KleidiAI backend for llama.cpp
* Implement a KleidiAI backend for llama.cpp

Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
---
Expand All @@ -20,7 +20,7 @@ Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
create mode 100644 ggml-kleidiai.h

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 08481334..5c0458e9 100644
index 08481334..99382573 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -548,6 +548,53 @@ if (LLAMA_VULKAN)
Expand All @@ -31,9 +31,9 @@ index 08481334..5c0458e9 100644
+
+ # Fetch KleidiAI sources:
+ include(FetchContent)
+ set(KLEIDIAI_COMMIT_SHA "b0911c80b35e41dc9c22075a63e83c217fd0a106")
+ set(KLEIDIAI_COMMIT_SHA "d6c3b987e445e5e1daeda94e3c2888efaa07ca50")
+ set(KLEIDIAI_DOWNLOAD_URL "https://gitlab.arm.com/kleidi/kleidiai/-/archive/${KLEIDIAI_COMMIT_SHA}/kleidiai-${KLEIDIAI_COMMIT_SHA}.tar.gz")
+ set(KLEIDIAI_ARCHIVE_MD5 "8b54226586eb18957c374a6d1434f4f2")
+ set(KLEIDIAI_ARCHIVE_MD5 "8e94e73bfa00ea038fd6e3d13f59080f")
+
+ if (POLICY CMP0135)
+ cmake_policy(SET CMP0135 NEW)
Expand Down Expand Up @@ -86,7 +86,7 @@ index 08481334..5c0458e9 100644

target_include_directories(ggml PUBLIC . ${LLAMA_EXTRA_INCLUDES})
diff --git a/ggml-alloc.c b/ggml-alloc.c
index bd367c42..ac099392 100644
index bd367c42..ed4ce0ae 100644
--- a/ggml-alloc.c
+++ b/ggml-alloc.c
@@ -9,6 +9,10 @@
Expand All @@ -105,12 +105,12 @@ index bd367c42..ac099392 100644
if (t->data == NULL && t->view_src == NULL) {
this_size = GGML_PAD(ggml_backend_buft_get_alloc_size(buft, t), alignment);
+#if defined(GGML_USE_KLEIDIAI)
+ // Temporary solution to allocate more memore if needed for packing the weights.
+ // Temporary solution to allocate more memory if needed for packing the weights.
+ // This method is not sufficient as we assume that the weights are for matmul only.
+ // However, weights could belong to other operations
+ const int64_t iai_diff = (ggml_kai_get_const_workspace_size_matmul(t) - this_size);
+ if (iai_diff > 0) {
+ this_size += iai_diff;
+ const int64_t kai_diff = (ggml_kai_get_const_workspace_size_matmul(t) - this_size);
+ if (kai_diff > 0) {
+ this_size += kai_diff;
+ }
+#endif
}
Expand Down