diff --git a/cpu-miner.c b/cpu-miner.c
index 933b443..5401839 100644
--- a/cpu-miner.c
+++ b/cpu-miner.c
@@ -207,6 +207,7 @@ Options:\n\
                         Device IDs start counting from 0! Alternatively takes\n\
                         string names of your cards like gtx780ti or gt640#2\n\
                         (matching 2nd gt640 in the PC)\n\
+  -f, --diff            Divide difficulty by this factor (std is 1) \n\
   -v, --vote=VOTE       block reward vote (for HeavyCoin)\n\
   -m, --trust-pool      trust the max block reward vote (maxvote) sent by the pool\n\
   -o, --url=URL         URL of mining server\n\
diff --git a/heavy/cuda_blake512.cu b/heavy/cuda_blake512.cu
index dd1c737..9e394fd 100644
--- a/heavy/cuda_blake512.cu
+++ b/heavy/cuda_blake512.cu
@@ -20,6 +20,8 @@ uint32_t *d_hash5output[8];
 // die Message (112 bzw. 116 Bytes) mit Padding zur Berechnung auf der GPU
 __constant__ uint64_t c_PaddedMessage[16]; // padded message (80/84+32 bytes + padding)
 
+#include "cuda_helper.h"
+
 // ---------------------------- BEGIN CUDA blake512 functions ------------------------------------
 
 __constant__ uint8_t c_sigma[16][16];
@@ -136,8 +138,6 @@ template <int BLOCKSIZE> __device__ void blake512_compress( uint64_t *h, const u
     for( i = 0; i < 16; ++i )  h[i % 8] ^= v[i];
 }
 
-#include "cuda_helper.h"
-
 template <int BLOCKSIZE> __global__ void blake512_gpu_hash(int threads, uint32_t startNounce, void *outputHash, uint32_t *heftyHashes, uint32_t *nonceVector)
 {
 	int thread = (blockDim.x * blockIdx.x + threadIdx.x);