Skip to content

Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.

ml/backend/ggml: use default CUDA compression mode (#10314)

Jeffrey Morgan09bb2e30f694
diff --git a/CMakeLists.txt b/CMakeLists.txt
index b38426996..5343d8774 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -24,7 +24,7 @@ set(GGML_LLAMAFILE ON)
set(GGML_CUDA_PEER_MAX_BATCH_SIZE 128)
set(GGML_CUDA_GRAPHS ON)
set(GGML_CUDA_FA ON)
-set(GGML_CUDA_COMPRESSION_MODE none)
+set(GGML_CUDA_COMPRESSION_MODE default)
if((CMAKE_OSX_ARCHITECTURES AND NOT CMAKE_OSX_ARCHITECTURES MATCHES "arm64")
OR (NOT CMAKE_OSX_ARCHITECTURES AND NOT CMAKE_SYSTEM_PROCESSOR MATCHES "arm|aarch64|ARM64|ARMv[0-9]+"))