diff --git a/Makefile b/Makefile
index 5b3baf695..4b98452f6 100644
--- a/Makefile
+++ b/Makefile
@@ -366,10 +366,14 @@ ifdef LLAMA_BLIS
 endif # LLAMA_BLIS
 
 ifdef LLAMA_CUBLAS
-	MK_CPPFLAGS  += -DGGML_USE_CUBLAS -I/usr/local/cuda/include -I/opt/cuda/include -L$(CUDA_PATH)/targets/x86_64-linux/include -I/usr/local/cuda/targets/aarch64-linux/include
+	MK_CPPFLAGS  += -DGGML_USE_CUBLAS -I/usr/local/cuda/include -I/opt/cuda/include -I$(CUDA_PATH)/targets/x86_64-linux/include -I/usr/local/cuda/targets/aarch64-linux/include
 	MK_LDFLAGS   += -lcublas -lculibos -lcudart -lcublasLt -lpthread -ldl -lrt -L/usr/local/cuda/lib64 -L/opt/cuda/lib64 -L$(CUDA_PATH)/targets/x86_64-linux/lib -L/usr/local/cuda/targets/aarch64-linux/lib
 	OBJS         += ggml-cuda.o
+ifdef JETSON_EOL_MODULE_DETECT	
 	MK_NVCCFLAGS  = -use_fast_math
+else
+	MK_NVCCFLAGS = --forward-unknown-to-host-compiler -use_fast_math
+endif # JETSON_EOL_MODULE_DETECT	
 
 ifdef LLAMA_DEBUG
 	MK_NVCCFLAGS += -lineinfo
diff --git a/ggml-cuda.cu b/ggml-cuda.cu
index b4396de71..d0eed3d2d 100644
--- a/ggml-cuda.cu
+++ b/ggml-cuda.cu
@@ -8403,7 +8403,7 @@ static void ggml_cuda_mul_mat_mat_batched_cublas(const ggml_tensor * src0, const
     half * dst_f16 = nullptr;
     char * dst_t   = nullptr;
 
-    cudaDataType_t cu_compute_type = CUBLAS_COMPUTE_16F;
+    cublasComputeType_t cu_compute_type = CUBLAS_COMPUTE_16F;
     cudaDataType_t      cu_data_type    = CUDA_R_16F;
 
     // dst strides