diff --git a/embassy-rp/src/pio.rs b/embassy-rp/src/pio.rs
index 5fed4f1fa..fe66d9aef 100644
--- a/embassy-rp/src/pio.rs
+++ b/embassy-rp/src/pio.rs
@@ -855,13 +855,14 @@ impl<'d, PIO: Instance> Common<'d, PIO> {
     {
         let mut used_mask = 0;
         for (i, instr) in instrs.enumerate() {
-            let addr = (i + start) as u8;
-            let mask = 1 << (addr as usize);
-            if self.instructions_used & mask != 0 {
-                return Err(addr as usize);
+            // wrapping around the end of program memory is valid, let's make use of that.
+            let addr = (i + start) % 32;
+            let mask = 1 << addr;
+            if (self.instructions_used | used_mask) & mask != 0 {
+                return Err(addr);
             }
             unsafe {
-                PIO::PIO.instr_mem(addr as usize).write(|w| {
+                PIO::PIO.instr_mem(addr).write(|w| {
                     w.set_instr_mem(instr);
                 });
             }
diff --git a/embassy-rp/src/relocate.rs b/embassy-rp/src/relocate.rs
index f36170e03..9cb279ccd 100644
--- a/embassy-rp/src/relocate.rs
+++ b/embassy-rp/src/relocate.rs
@@ -26,11 +26,7 @@ where
             Some(if instr & 0b1110_0000_0000_0000 == 0 {
                 // this is a JMP instruction -> add offset to address
                 let address = (instr & 0b1_1111) as u8;
-                let address = address + self.offset;
-                assert!(
-                    address < pio::RP2040_MAX_PROGRAM_SIZE as u8,
-                    "Invalid JMP out of the program after offset addition"
-                );
+                let address = address.wrapping_add(self.offset) % 32;
                 instr & (!0b11111) | address as u16
             } else {
                 instr
@@ -62,8 +58,8 @@ impl<'a, const PROGRAM_SIZE: usize> RelocatedProgram<'a, PROGRAM_SIZE> {
         let wrap = self.program.wrap;
         let origin = self.origin;
         Wrap {
-            source: wrap.source + origin,
-            target: wrap.target + origin,
+            source: wrap.source.wrapping_add(origin) % 32,
+            target: wrap.target.wrapping_add(origin) % 32,
         }
     }
 
diff --git a/examples/rp/src/bin/pio_hd44780.rs b/examples/rp/src/bin/pio_hd44780.rs
index 61c5565d3..088fd5649 100644
--- a/examples/rp/src/bin/pio_hd44780.rs
+++ b/examples/rp/src/bin/pio_hd44780.rs
@@ -95,6 +95,7 @@ impl<'l> HD44780<'l> {
         let prg = pio_proc::pio_asm!(
             r#"
                 .side_set 1 opt
+                .origin 20
 
                 loop:
                     out x,     24
@@ -148,7 +149,7 @@ impl<'l> HD44780<'l> {
         // many side sets are only there to free up a delay bit!
         let prg = pio_proc::pio_asm!(
             r#"
-                .origin 7
+                .origin 27
                 .side_set 1
 
                 .wrap_target