diff --git a/cache_utils/Cargo.toml b/cache_utils/Cargo.toml
index 110e4b7..c05b938 100644
--- a/cache_utils/Cargo.toml
+++ b/cache_utils/Cargo.toml
@@ -19,6 +19,8 @@ nix = { version = "0.20.0", optional = true }
 libc = { version = "0.2.92", optional = true }
 hashbrown = { version = "0.11.2", optional = true }
 turn_lock = { path = "../turn_lock", optional = true}
+lazy_static = "1.4.0"
+bitvec = "0.22.3"
 
 [features]
 use_std = ["nix", "itertools/use_std", "libc", "cpuid/use_std", "turn_lock"]
diff --git a/cache_utils/src/bin/cache_level_cal.rs b/cache_utils/src/bin/cache_level_cal.rs
index a7a8d88..d7ec654 100644
--- a/cache_utils/src/bin/cache_level_cal.rs
+++ b/cache_utils/src/bin/cache_level_cal.rs
@@ -14,6 +14,7 @@ use nix::unistd::Pid;
 
 use core::arch::x86_64 as arch_x86;
 
+use cache_utils::ip_tool::Function;
 use core::cmp::min;
 use std::cmp::Ordering;
 use std::collections::HashMap;
@@ -53,6 +54,53 @@ use std::str::from_utf8;
    - [ ] Make the plots
 */
 
+unsafe fn function_call(f: &Function, addr: *const u8) -> u64 {
+    unsafe { (f.fun)(addr) }
+}
+
+unsafe fn prepare_RAM(p: *const u8) {
+    unsafe { flush(p) };
+}
+
+unsafe fn prepare_pL3(p: *const u8) {
+    unsafe { maccess(p) };
+    unsafe { arch_x86::_mm_mfence() };
+    unsafe { flush(p) };
+    unsafe { arch_x86::_mm_mfence() };
+    unsafe { arch_x86::_mm_prefetch::<{ arch_x86::_MM_HINT_T2 }>(p as *const i8) };
+    unsafe { arch_x86::__cpuid_count(0, 0) };
+}
+
+unsafe fn prepare_pL2(p: *const u8) {
+    unsafe { maccess(p) };
+    unsafe { arch_x86::_mm_mfence() };
+    unsafe { flush(p) };
+    unsafe { arch_x86::_mm_mfence() };
+    unsafe { arch_x86::_mm_prefetch::<{ arch_x86::_MM_HINT_T1 }>(p as *const i8) };
+    unsafe { arch_x86::__cpuid_count(0, 0) };
+}
+
+unsafe fn prepare_pL1(p: *const u8) {
+    unsafe { maccess(p) };
+    unsafe { arch_x86::_mm_mfence() };
+    unsafe { flush(p) };
+    unsafe { arch_x86::_mm_mfence() };
+    unsafe { arch_x86::_mm_prefetch::<{ arch_x86::_MM_HINT_T0 }>(p as *const i8) };
+    unsafe { arch_x86::__cpuid_count(0, 0) };
+}
+
+unsafe fn prepare_L1(p: *const u8) {
+    unsafe { only_reload(p) };
+}
+
+unsafe fn prepare_eL2(p: *const u8) {
+    unimplemented!()
+}
+
+unsafe fn prepare_eL3(p: *const u8) {
+    unimplemented!()
+}
+
 unsafe fn multiple_access(p: *const u8) {
     unsafe {
         maccess::<u8>(p);
@@ -105,26 +153,11 @@ struct Threshold {
     pub num_false_miss: u32,
 }
 
-unsafe fn only_flush_wrap(_: &(), addr: *const u8) -> u64 {
-    unsafe { only_flush(addr) }
-}
-
-unsafe fn only_reload_wrap(_: &(), addr: *const u8) -> u64 {
-    unsafe { only_reload(addr) }
-}
-
-unsafe fn load_and_flush_wrap(_: &(), addr: *const u8) -> u64 {
-    unsafe { load_and_flush(addr) }
-}
-unsafe fn flush_and_reload_wrap(_: &(), addr: *const u8) -> u64 {
-    unsafe { flush_and_reload(addr) }
-}
-
-unsafe fn reload_and_flush_wrap(_: &(), addr: *const u8) -> u64 {
-    unsafe { reload_and_flush(addr) }
-}
-
 fn main() {
+    let measure_reload =
+        cache_utils::ip_tool::Function::try_new(1, 0, cache_utils::ip_tool::TIMED_MACCESS).unwrap();
+    let measure_nop =
+        cache_utils::ip_tool::Function::try_new(1, 0, cache_utils::ip_tool::TIMED_NOP).unwrap();
     // Grab a slice of memory
 
     let core_per_socket_out = Command::new("sh")
@@ -172,13 +205,50 @@ fn main() {
         panic!("not aligned nicely");
     }
 
-    let operations = [CalibrateOperation2T {
-        prepare: maccess::<u8>,
-        op: only_flush_wrap,
-        name: "clflush_remote_hit",
-        display_name: "clflush remote hit",
-        t: &(),
-    }];
+    let operations = [
+        CalibrateOperation2T {
+            prepare: prepare_RAM,
+            op: function_call,
+            name: "RAM_load",
+            display_name: "Load from RAM",
+            t: &measure_reload,
+        },
+        CalibrateOperation2T {
+            prepare: prepare_pL3,
+            op: function_call,
+            name: "pL3_load",
+            display_name: "Load from L3 (prefetch)",
+            t: &measure_reload,
+        },
+        CalibrateOperation2T {
+            prepare: prepare_pL2,
+            op: function_call,
+            name: "pL2_load",
+            display_name: "Load from L2 (prefetch)",
+            t: &measure_reload,
+        },
+        CalibrateOperation2T {
+            prepare: prepare_pL1,
+            op: function_call,
+            name: "pL1_load",
+            display_name: "Load from L1 (prefetch)",
+            t: &measure_reload,
+        },
+        CalibrateOperation2T {
+            prepare: prepare_L1,
+            op: function_call,
+            name: "L1_load",
+            display_name: "Load from L1 (Reload)",
+            t: &measure_reload,
+        },
+        CalibrateOperation2T {
+            prepare: noop::<u8>,
+            op: function_call,
+            name: "pL3_load",
+            display_name: "Load from L3 (prefetch)",
+            t: &measure_nop,
+        },
+    ];
 
     let r = unsafe {
         calibrate_fixed_freq_2_thread(
diff --git a/prefetcher_reverse/src/ip_tool.rs b/cache_utils/src/ip_tool.rs
similarity index 92%
rename from prefetcher_reverse/src/ip_tool.rs
rename to cache_utils/src/ip_tool.rs
index e0cc893..6674c07 100644
--- a/prefetcher_reverse/src/ip_tool.rs
+++ b/cache_utils/src/ip_tool.rs
@@ -1,9 +1,10 @@
+use crate::mmap::MMappedMemory;
 use bitvec::prelude::*;
-use cache_utils::mmap::MMappedMemory;
 use lazy_static::lazy_static;
 use std::collections::LinkedList;
 use std::ptr::copy_nonoverlapping;
 use std::sync::Mutex;
+use std::vec::Vec;
 
 struct WXRange {
     start: usize,
@@ -56,6 +57,12 @@ pub const TIMED_CLFLUSH: FunctionTemplate = FunctionTemplate {
     end: timed_clflush_template_end as *const u8,
 };
 
+pub const TIMED_NOP: FunctionTemplate = FunctionTemplate {
+    start: timed_nop_template,
+    ip: timed_nop_template_ip as *const u8,
+    end: timed_nop_template_end as *const u8,
+};
+
 impl WXRange {
     unsafe fn allocate(
         &mut self,
@@ -274,6 +281,31 @@ global_asm!(
     ".global timed_clflush_template_end",
     "timed_clflush_template_end:",
     "nop",
+    ".global timed_nop_template",
+    "timed_nop_template:",
+    "mfence",
+    "lfence",
+    "rdtsc",
+    "shl rdx, 32",
+    "mov rsi, rdx",
+    "add rsi, rax",
+    "mfence",
+    "lfence",
+    ".global timed_nop_template_ip",
+    "timed_nop_template_ip:",
+    "nop",
+    "mfence",
+    "lfence",
+    "rdtsc",
+    "shl rdx, 32",
+    "add rax, rdx",
+    "mfence",
+    "lfence",
+    "sub rax, rsi",
+    "ret",
+    ".global timed_nop_template_end",
+    "timed_nop_template_end:",
+    "nop",
 );
 
 extern "C" {
@@ -283,6 +315,9 @@ extern "C" {
     fn timed_clflush_template(pointer: *const u8) -> u64;
     fn timed_clflush_template_ip();
     fn timed_clflush_template_end();
+    fn timed_nop_template(pointer: *const u8) -> u64;
+    fn timed_nop_template_ip();
+    fn timed_nop_template_end();
 }
 
 pub fn tmp_test() {
diff --git a/cache_utils/src/lib.rs b/cache_utils/src/lib.rs
index 8ec4180..d892938 100644
--- a/cache_utils/src/lib.rs
+++ b/cache_utils/src/lib.rs
@@ -1,8 +1,13 @@
 #![cfg_attr(feature = "no_std", no_std)]
 #![feature(ptr_internals)]
+#![feature(linked_list_cursors)]
+#![feature(global_asm)]
 #![allow(clippy::missing_safety_doc)]
 #![deny(unsafe_op_in_unsafe_fn)]
 
+use core::arch::x86_64 as arch_x86;
+use core::ptr;
+
 use static_assertions::assert_cfg;
 
 assert_cfg!(
@@ -25,8 +30,8 @@ pub mod frequency;
 #[cfg(feature = "use_std")]
 mod calibrate_2t;
 
-use core::arch::x86_64 as arch_x86;
-use core::ptr;
+#[cfg(feature = "use_std")]
+pub mod ip_tool;
 
 // rdtsc no fence
 pub unsafe fn rdtsc_nofence() -> u64 {