diff --git a/include/bench/bench.h b/include/bench/bench.h
index 9698f50..cbf6935 100755
--- a/include/bench/bench.h
+++ b/include/bench/bench.h
@@ -37,24 +37,24 @@ static inline int get_time_hi() {return *(volatile int*) GET_TIME_HI_ADDR;}
 
 #if !defined(ARCHI_HAS_FC)
 
-static inline void start_timer()
+static inline void start_timer(int cid)
 {
-  timer_start(timer_base_cl(0, 0, 1));
+  timer_start(timer_base_cl(cid, 0, 1));
 }
 
-static inline void stop_timer()
+static inline void stop_timer(int cid)
 {
-  timer_conf_set(timer_base_cl(0, 0, 1), 0);
+  timer_conf_set(timer_base_cl(cid, 0, 1), 0);
 }
 
-static inline void reset_timer()
+static inline void reset_timer(int cid)
 {
-  timer_reset(timer_base_cl(0, 0, 1));
+  timer_reset(timer_base_cl(cid, 0, 1));
 }
 
-static inline int get_time()
+static inline int get_time(int cid)
 {
-  return timer_count_get(timer_base_cl(0, 0, 1));
+  return timer_count_get(timer_base_cl(cid, 0, 1));
 }
 
 #else
diff --git a/include/data/data.h b/include/data/data.h
index 8d8acbd..3e59ce6 100644
--- a/include/data/data.h
+++ b/include/data/data.h
@@ -39,7 +39,7 @@
 #define PI_L2 __attribute__((section(".l2_data")))
 #define L2_DATA PI_L2
 
-#define L1_GLOBAL_DATA __attribute__((section(".data_l1")))
+#define L1_GLOBAL_DATA __attribute__((section("__data_l1")))
 #define L1_DATA L1_GLOBAL_DATA
 
 #ifdef USE_CLUSTER
diff --git a/include/hal/timer/timer_v2.h b/include/hal/timer/timer_v2.h
index d6923f1..6961a2a 100644
--- a/include/hal/timer/timer_v2.h
+++ b/include/hal/timer/timer_v2.h
@@ -39,7 +39,7 @@ static inline unsigned int timer_base_fc(int id, int sub_id)
 
 static inline unsigned int timer_base_cl(int cid, int id, int sub_id)
 {
-  return ARCHI_CLUSTER_PERIPHERALS_GLOBAL_ADDR(0) + ARCHI_TIMER_OFFSET + id * ARCHI_TIMER_SIZE + sub_id * 4;
+  return ARCHI_CLUSTER_PERIPHERALS_GLOBAL_ADDR(cid) + ARCHI_TIMER_OFFSET + id * ARCHI_TIMER_SIZE + sub_id * 4;
 }
 
 #else
diff --git a/kernel/bench.c b/kernel/bench.c
index 8bc820d..e8f6378 100644
--- a/kernel/bench.c
+++ b/kernel/bench.c
@@ -29,7 +29,7 @@ void bench_disable_printf(void) {
 
 void bench_timer_start(void) {
   if (get_core_id()==0)
-    start_timer();
+    start_timer((int) get_cluster_id());
 #ifdef PROFILE
   perf_start();
 #endif
@@ -37,7 +37,7 @@ void bench_timer_start(void) {
 
 void bench_timer_stop(void) {
   if (get_core_id()==0)
-    stop_timer();
+    stop_timer((int) get_cluster_id());
 
 #ifdef PROFILE
   perf_stop();
@@ -47,8 +47,8 @@ void bench_timer_stop(void) {
 
 void bench_timer_reset(void) {
   if (get_core_id()==0) {
-    stop_timer();
-    reset_timer();
+    stop_timer((int) get_cluster_id());
+    reset_timer((int) get_cluster_id());
   }
 #ifdef PROFILE
   perf_reset();
@@ -79,7 +79,7 @@ void print_result(testcase_t *test, testresult_t *result)
 void print_summary(unsigned int errors)
 {
   #ifdef RTL_SDK
-  volatile int* ptr = (int*)(0x10001000+get_core_id()*4*2);
+  volatile int* ptr = (int*)(pos_l1_base+get_core_id()*4*2);
   ptr[1] = errors;
   #endif
 
@@ -110,7 +110,7 @@ void run_benchmark(testcase_t *test, testresult_t *result)
 
   test->test(result, bench_timer_start, bench_timer_stop);
 
-  result->time = get_time();
+  result->time = get_time((int) get_cluster_id());
 }
 
 int run_suite(testcase_t *tests)
diff --git a/kernel/chips/pulp_cluster/link.ld b/kernel/chips/pulp_cluster/link.ld
index 534bb11..bdad1f7 100644
--- a/kernel/chips/pulp_cluster/link.ld
+++ b/kernel/chips/pulp_cluster/link.ld
@@ -221,12 +221,16 @@ SECTIONS
     . = ALIGN(4);
   } > L2
 
+  __l2_data_end = .;
+
   .l1cluster_g : {
     . = ALIGN(4);
+    __heap_sram = .;
     *(.heapsram)
     *(.heapsram.*)
     *(.l1cluster_g)
     *(.l1cluster_g.*)
+    __data_l1 = .;
     *(.data_l1)
     *(.data_l1.*)
     . = ALIGN(4);
@@ -245,7 +249,7 @@ SECTIONS
 
   __l1_end = ALIGN(4);
 
-  __l2_shared_end = LOADADDR(.bss_l1) + SIZEOF(.bss_l1);
+  __l2_shared_end = __l2_data_end + SIZEOF(.l1cluster_g) + SIZEOF(.bss_l1);
 
 
 
diff --git a/kernel/crt0.S b/kernel/crt0.S
index c6c755a..fc846a0 100644
--- a/kernel/crt0.S
+++ b/kernel/crt0.S
@@ -30,9 +30,10 @@ pos_init_entry:
     # PEs from 1 to 7 will go to sync_loop and wait. PE0 will reach them
     # later after the pos_init_start. Then, they'll set up their stack into
     # the L1 and jump to cluster_entry_stub
-    li t0, 0x10080000
-    sw x0, 0(t0)
-    bnez    a0,  sync_loop
+    # We check if the offset of the core is zero, so that even if the cluster
+    # ID is not zero, the execution does not break here
+    andi    a2, a0, 0x0f
+    bnez    a2, sync_loop
 #else   
     srli    a0, a0, 5
 #ifdef ARCHI_CL_BOOT
@@ -76,9 +77,6 @@ pos_init_entry:
 #ifdef ARCHI_NO_FC
     csrr    a0, 0xF14
     andi    a1, a0, 0x1f
-    li t0, 0x10080000
-    li t1, 0x1
-    sw t1, 0(t0)
     j pe_start
 #endif