diff --git a/cuda-oxide/crates/cuda-host/README.md b/cuda-oxide/crates/cuda-host/README.md
index 7d921a8bb6..292c527a4f 100644
--- a/cuda-oxide/crates/cuda-host/README.md
+++ b/cuda-oxide/crates/cuda-host/README.md
@@ -20,8 +20,9 @@ mod kernels {
     #[kernel]
     pub fn vecadd(a: &[f32], b: &[f32], mut c: DisjointSlice<f32>) {
         let idx = thread::index_1d();
+        let i = idx.get();
         if let Some(c_elem) = c.get_mut(idx) {
-            *c_elem = a[idx.get()] + b[idx.get()];
+            *c_elem = a[i] + b[i];
         }
     }
 }
diff --git a/cuda-oxide/crates/rustc-codegen-cuda/examples/cross_crate_kernel/README.md b/cuda-oxide/crates/rustc-codegen-cuda/examples/cross_crate_kernel/README.md
index aff5decf2b..a5b9914bca 100644
--- a/cuda-oxide/crates/rustc-codegen-cuda/examples/cross_crate_kernel/README.md
+++ b/cuda-oxide/crates/rustc-codegen-cuda/examples/cross_crate_kernel/README.md
@@ -87,8 +87,9 @@ use core::ops::Mul;
 #[kernel]
 pub fn scale<T: Copy + Mul<Output = T>>(factor: T, input: &[T], mut out: DisjointSlice<T>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(out_elem) = out.get_mut(idx) {
-        *out_elem = input[idx.get()] * factor;
+        *out_elem = input[i] * factor;
     }
 }
 ```
diff --git a/cuda-oxide/crates/rustc-codegen-cuda/examples/debug/README.md b/cuda-oxide/crates/rustc-codegen-cuda/examples/debug/README.md
index 13639ba78a..cb118a58d3 100644
--- a/cuda-oxide/crates/rustc-codegen-cuda/examples/debug/README.md
+++ b/cuda-oxide/crates/rustc-codegen-cuda/examples/debug/README.md
@@ -42,8 +42,9 @@ pub fn clock_test(mut output: DisjointSlice<u64>) {
 #[kernel]
 pub fn trap_test(input: &[i32], mut output: DisjointSlice<i32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(output_elem) = output.get_mut(idx) {
-        let val = input[idx.get()];
+        let val = input[i];
 
         if val < 0 {
             debug::trap();  // Kernel aborts here
@@ -60,8 +61,9 @@ pub fn trap_test(input: &[i32], mut output: DisjointSlice<i32>) {
 #[kernel]
 pub fn assert_test(input: &[i32], mut output: DisjointSlice<i32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(output_elem) = output.get_mut(idx) {
-        let val = input[idx.get()];
+        let val = input[i];
 
         // With message
         gpu_assert!(val >= 0, "Expected non-negative value");
@@ -86,11 +88,12 @@ The two forms intentionally use different failure mechanisms:
 #[kernel]
 pub fn breakpoint_test(mut output: DisjointSlice<i32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(output_elem) = output.get_mut(idx) {
-        if idx.get() == 0 {
+        if i == 0 {
             debug::breakpoint();  // cuda-gdb stops here
         }
-        *output_elem = idx.get() as i32;
+        *output_elem = i as i32;
     }
 }
 ```
@@ -101,10 +104,11 @@ pub fn breakpoint_test(mut output: DisjointSlice<i32>) {
 #[kernel]
 pub fn profiler_test(input: &[f32], mut output: DisjointSlice<f32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(output_elem) = output.get_mut(idx) {
         debug::prof_trigger::<0>();  // Region start marker
 
-        let val = input[idx.get()];
+        let val = input[i];
         let result = val * val;
 
         debug::prof_trigger::<1>();  // Region end marker
diff --git a/cuda-oxide/crates/rustc-codegen-cuda/examples/device_closures/src/main.rs b/cuda-oxide/crates/rustc-codegen-cuda/examples/device_closures/src/main.rs
index 5b4f1fe1cc..c604dd98e6 100644
--- a/cuda-oxide/crates/rustc-codegen-cuda/examples/device_closures/src/main.rs
+++ b/cuda-oxide/crates/rustc-codegen-cuda/examples/device_closures/src/main.rs
@@ -49,8 +49,9 @@
 //! fn scale_kernel(factor: u32, input: &[u32], mut out: DisjointSlice<u32>) {
 //!     // 'factor' is the closure capture, passed as scalar argument
 //!     let idx = thread::index_1d();
+//!     let i = idx.get();
 //!     if let Some(slot) = out.get_mut(idx) {
-//!         *slot = input[idx.get()] * factor;
+//!         *slot = input[i] * factor;
 //!     }
 //! }
 //! ```
diff --git a/cuda-oxide/crates/rustc-codegen-cuda/examples/error/README.md b/cuda-oxide/crates/rustc-codegen-cuda/examples/error/README.md
index 2b8d3298f3..52e3e7a186 100644
--- a/cuda-oxide/crates/rustc-codegen-cuda/examples/error/README.md
+++ b/cuda-oxide/crates/rustc-codegen-cuda/examples/error/README.md
@@ -25,9 +25,10 @@ This is a **negative test** - it verifies that:
 #[kernel]
 pub fn valid_f64_to_f32_kernel(a: &[f64], b: &[f64], mut c: DisjointSlice<f32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(c_elem) = c.get_mut(idx) {
         // Valid: explicit f64 to f32 cast
-        *c_elem = (a[idx.get()] + b[idx.get()]) as f32;
+        *c_elem = (a[i] + b[i]) as f32;
     }
 }
 ```
@@ -38,13 +39,14 @@ pub fn valid_f64_to_f32_kernel(a: &[f64], b: &[f64], mut c: DisjointSlice<f32>)
 #[kernel]
 pub fn unsupported_format_kernel(a: &[f32], mut c: DisjointSlice<f32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(c_elem) = c.get_mut(idx) {
         // INVALID: format_args! uses core::fmt which requires:
         // - Dynamic dispatch (dyn Trait)
         // - Heap allocation
         // - Complex runtime machinery
-        let _formatted = core::format_args!("{}", a[idx.get()]);
-        *c_elem = a[idx.get()];
+        let _formatted = core::format_args!("{}", a[i]);
+        *c_elem = a[i];
     }
 }
 ```
diff --git a/cuda-oxide/crates/rustc-codegen-cuda/examples/generic/README.md b/cuda-oxide/crates/rustc-codegen-cuda/examples/generic/README.md
index cb8d5ce2db..eb942b55d5 100644
--- a/cuda-oxide/crates/rustc-codegen-cuda/examples/generic/README.md
+++ b/cuda-oxide/crates/rustc-codegen-cuda/examples/generic/README.md
@@ -19,16 +19,18 @@ Tests whether the collector correctly handles monomorphized generic kernels. Thi
 #[kernel]
 pub fn scale<T: Copy + Mul<Output = T>>(factor: T, input: &[T], mut out: DisjointSlice<T>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(out_elem) = out.get_mut(idx) {
-        *out_elem = input[idx.get()] * factor;
+        *out_elem = input[i] * factor;
     }
 }
 
 #[kernel]
 pub fn add<T: Copy + Add<Output = T>>(a: &[T], b: &[T], mut c: DisjointSlice<T>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(c_elem) = c.get_mut(idx) {
-        *c_elem = a[idx.get()] + b[idx.get()];
+        *c_elem = a[i] + b[i];
     }
 }
 ```
@@ -98,8 +100,9 @@ pub fn saxpy<T: Copy + Mul<Output = T> + Add<Output = T>>(
     a: T, x: &[T], y: &[T], mut out: DisjointSlice<T>
 ) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(out_elem) = out.get_mut(idx) {
-        *out_elem = a * x[idx.get()] + y[idx.get()];
+        *out_elem = a * x[i] + y[i];
     }
 }
 
@@ -116,8 +119,9 @@ pub fn saxpy<T: Copy + Mul<Output = T> + Add<Output = T>>(
 #[kernel]
 pub fn map<T: Copy, F: Fn(T) -> T + Copy>(f: F, input: &[T], mut out: DisjointSlice<T>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(out_elem) = out.get_mut(idx) {
-        *out_elem = f(input[idx.get()]);
+        *out_elem = f(input[i]);
     }
 }
 ```
diff --git a/cuda-oxide/crates/rustc-codegen-cuda/examples/helper_fn/README.md b/cuda-oxide/crates/rustc-codegen-cuda/examples/helper_fn/README.md
index 193e0b8edc..915c46e86a 100644
--- a/cuda-oxide/crates/rustc-codegen-cuda/examples/helper_fn/README.md
+++ b/cuda-oxide/crates/rustc-codegen-cuda/examples/helper_fn/README.md
@@ -18,8 +18,8 @@ Demonstrates the `#[device]` attribute for helper functions that are called from
 #[device]
 pub fn vecadd_device(a: &[f32], b: &[f32], mut c: DisjointSlice<f32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(c_elem) = c.get_mut(idx) {
-        let i = idx.get();
         *c_elem = a[i] + b[i];
     }
 }
@@ -90,11 +90,12 @@ fn clamp(x: f32, lo: f32, hi: f32) -> f32 {
 #[kernel]
 pub fn apply_clamp(input: &[f32], mut out: DisjointSlice<f32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(out_elem) = out.get_mut(idx) {
         // Call the device function by its original name. The macro
         // expands this to the reserved internal symbol behind the
         // scenes; users never type the mangled name themselves.
-        *out_elem = clamp(input[idx.get()], 0.0, 1.0);
+        *out_elem = clamp(input[i], 0.0, 1.0);
     }
 }
 ```
diff --git a/cuda-oxide/crates/rustc-codegen-cuda/examples/host_closure/README.md b/cuda-oxide/crates/rustc-codegen-cuda/examples/host_closure/README.md
index e20dd4890a..15ed94e347 100644
--- a/cuda-oxide/crates/rustc-codegen-cuda/examples/host_closure/README.md
+++ b/cuda-oxide/crates/rustc-codegen-cuda/examples/host_closure/README.md
@@ -18,8 +18,9 @@ Tests passing closures from host code to generic GPU kernels. This enables funct
 #[kernel]
 pub fn map<T: Copy, F: Fn(T) -> T + Copy>(f: F, input: &[T], mut out: DisjointSlice<T>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(out_elem) = out.get_mut(idx) {
-        *out_elem = f(input[idx.get()]);
+        *out_elem = f(input[i]);
     }
 }
 ```
diff --git a/cuda-oxide/crates/rustc-codegen-cuda/src/lib.rs b/cuda-oxide/crates/rustc-codegen-cuda/src/lib.rs
index 8564f84dbe..dcfc592fc2 100644
--- a/cuda-oxide/crates/rustc-codegen-cuda/src/lib.rs
+++ b/cuda-oxide/crates/rustc-codegen-cuda/src/lib.rs
@@ -246,8 +246,9 @@
 //! #[kernel]
 //! pub fn vecadd(a: &[f32], b: &[f32], mut c: DisjointSlice<f32>) {
 //!     let idx = thread::index_1d();
+//!     let i = idx.get();
 //!     if let Some(c_elem) = c.get_mut(idx) {
-//!         *c_elem = a[idx.get()] + b[idx.get()];
+//!         *c_elem = a[i] + b[i];
 //!     }
 //! }
 //!
diff --git a/cuda-oxide/cuda-oxide-book/compiler/mir-importer.md b/cuda-oxide/cuda-oxide-book/compiler/mir-importer.md
index d6c3281ade..181a824718 100644
--- a/cuda-oxide/cuda-oxide-book/compiler/mir-importer.md
+++ b/cuda-oxide/cuda-oxide-book/compiler/mir-importer.md
@@ -401,8 +401,9 @@ pieces connect. Here is a vector addition kernel:
 #[kernel]
 pub fn vecadd(a: &[f32], b: &[f32], mut c: DisjointSlice<f32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(c_elem) = c.get_mut(idx) {
-        *c_elem = a[idx.get()] + b[idx.get()];
+        *c_elem = a[i] + b[i];
     }
 }
 ```
diff --git a/cuda-oxide/cuda-oxide-book/gpu-programming/closures-and-generics.md b/cuda-oxide/cuda-oxide-book/gpu-programming/closures-and-generics.md
index 7055e8100c..8826017482 100644
--- a/cuda-oxide/cuda-oxide-book/gpu-programming/closures-and-generics.md
+++ b/cuda-oxide/cuda-oxide-book/gpu-programming/closures-and-generics.md
@@ -342,8 +342,9 @@ pub fn apply_transform(input: &[f32], mut out: DisjointSlice<f32>) {
         clamped * clamped
     };
 
+    let i = idx.get();
     if let Some(out_elem) = out.get_mut(idx) {
-        *out_elem = transform(input[idx.get()]);
+        *out_elem = transform(input[i]);
     }
 }
 ```
diff --git a/cuda-oxide/cuda-oxide-book/gpu-programming/error-handling-and-debugging.md b/cuda-oxide/cuda-oxide-book/gpu-programming/error-handling-and-debugging.md
index f48e28e4b5..c49bc2c7c7 100644
--- a/cuda-oxide/cuda-oxide-book/gpu-programming/error-handling-and-debugging.md
+++ b/cuda-oxide/cuda-oxide-book/gpu-programming/error-handling-and-debugging.md
@@ -45,8 +45,9 @@ pub fn debug_kernel(data: &[f32], mut out: DisjointSlice<f32>) {
     if idx.get() < 4 {
         gpu_printf!("Thread {} sees value {}\n", idx.get(), data[idx.get()]);
     }
+    let i = idx.get();
     if let Some(out_elem) = out.get_mut(idx) {
-        *out_elem = data[idx.get()] * 2.0;
+        *out_elem = data[i] * 2.0;
     }
 }
 ```
@@ -81,7 +82,7 @@ the GPU. `gpu_printf!` bypasses all of this by lowering directly to a CUDA
 For fatal error checking on the device, use `gpu_assert!` or `debug::trap()`:
 
 ```rust
-use cuda_device::{DisjointSlice, debug, gpu_assert, kernel, thread};
+use cuda_device::{DisjointSlice, gpu_assert, kernel, thread};
 
 #[kernel]
 pub fn checked_kernel(data: &[f32], len: u32, mut out: DisjointSlice<f32>) {
@@ -96,8 +97,9 @@ pub fn checked_kernel(data: &[f32], len: u32, mut out: DisjointSlice<f32>) {
         "expected non-negative input"
     );
 
+    let i = idx.get();
     if let Some(out_elem) = out.get_mut(idx) {
-        *out_elem = data[idx.get()];
+        *out_elem = data[i];
     }
 }
 ```
diff --git a/cuda-oxide/cuda-oxide-book/gpu-programming/execution-model.md b/cuda-oxide/cuda-oxide-book/gpu-programming/execution-model.md
index 7694d271f5..b014d86292 100644
--- a/cuda-oxide/cuda-oxide-book/gpu-programming/execution-model.md
+++ b/cuda-oxide/cuda-oxide-book/gpu-programming/execution-model.md
@@ -46,8 +46,9 @@ use cuda_device::{DisjointSlice, kernel, thread};
 #[kernel]
 pub fn vecadd(a: &[f32], b: &[f32], mut c: DisjointSlice<f32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(c_elem) = c.get_mut(idx) {
-        *c_elem = a[idx.get()] + b[idx.get()];
+        *c_elem = a[i] + b[i];
     }
 }
 ```
@@ -238,8 +239,9 @@ returns `None` for out-of-bounds indices, so those threads simply do nothing:
 #[kernel]
 pub fn vecadd(a: &[f32], b: &[f32], mut c: DisjointSlice<f32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(c_elem) = c.get_mut(idx) {   // out-of-bounds threads skip
-        *c_elem = a[idx.get()] + b[idx.get()];
+        *c_elem = a[i] + b[i];
     }
 }
 ```
diff --git a/cuda-oxide/cuda-oxide-book/gpu-programming/kernels-and-device-functions.md b/cuda-oxide/cuda-oxide-book/gpu-programming/kernels-and-device-functions.md
index 303334a50e..33f12f7b39 100644
--- a/cuda-oxide/cuda-oxide-book/gpu-programming/kernels-and-device-functions.md
+++ b/cuda-oxide/cuda-oxide-book/gpu-programming/kernels-and-device-functions.md
@@ -23,8 +23,9 @@ use cuda_device::{DisjointSlice, kernel, thread};
 #[kernel]
 pub fn vecadd(a: &[f32], b: &[f32], mut c: DisjointSlice<f32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(c_elem) = c.get_mut(idx) {
-        *c_elem = a[idx.get()] + b[idx.get()];
+        *c_elem = a[i] + b[i];
     }
 }
 ```
@@ -152,8 +153,9 @@ fn clamp(x: f32, lo: f32, hi: f32) -> f32 {
 #[kernel]
 pub fn apply_clamp(input: &[f32], mut out: DisjointSlice<f32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(out_elem) = out.get_mut(idx) {
-        *out_elem = clamp(input[idx.get()], 0.0, 1.0);
+        *out_elem = clamp(input[i], 0.0, 1.0);
     }
 }
 ```
diff --git a/cuda-oxide/cuda-oxide-book/gpu-safety/the-safety-model.md b/cuda-oxide/cuda-oxide-book/gpu-safety/the-safety-model.md
index ed7f98b663..80d9168fe5 100644
--- a/cuda-oxide/cuda-oxide-book/gpu-safety/the-safety-model.md
+++ b/cuda-oxide/cuda-oxide-book/gpu-safety/the-safety-model.md
@@ -81,8 +81,8 @@ on multiple slices:
 #[kernel]
 pub fn vecadd(a: &[f32], b: &[f32], mut c: DisjointSlice<f32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(c_elem) = c.get_mut(idx) {
-        let i = idx.get();
         *c_elem = a[i] + b[i];
     }
 }
diff --git a/cuda-oxide/cuda-oxide-book/projects/async-mlp-pipeline.md b/cuda-oxide/cuda-oxide-book/projects/async-mlp-pipeline.md
index 7c12483443..fb538a0bbb 100644
--- a/cuda-oxide/cuda-oxide-book/projects/async-mlp-pipeline.md
+++ b/cuda-oxide/cuda-oxide-book/projects/async-mlp-pipeline.md
@@ -158,12 +158,13 @@ pub fn matvec_naive(
     mut vec_out: DisjointSlice<f32>,
 ) {
     let row = thread::index_1d();
+    let i = row.get();
     if let Some(out_elem) = vec_out.get_mut(row) {
         let n_sz = n as usize;
         let mut sum = 0.0f32;
         let mut j = 0usize;
         while j < n_sz {
-            sum = sum + mat[row.get() * n_sz + j] * vec_in[j];
+            sum = sum + mat[i * n_sz + j] * vec_in[j];
             j = j + 1;
         }
         *out_elem = sum;
@@ -181,8 +182,9 @@ interfaces.
 #[kernel]
 pub fn relu(input: &[f32], mut output: DisjointSlice<f32>) {
     let idx = thread::index_1d();
+    let i = idx.get();
     if let Some(out_elem) = output.get_mut(idx) {
-        let val = input[idx.get()];
+        let val = input[i];
         *out_elem = if val > 0.0f32 { val } else { 0.0f32 };
     }
 }