diff --git a/src/int/sdiv.rs b/src/int/sdiv.rs
index c62831d9..fa12d746 100644
--- a/src/int/sdiv.rs
+++ b/src/int/sdiv.rs
@@ -1,6 +1,6 @@
 use super::specialized_div_rem::*;
 
-// NOTE: there are panics inside the specialized_div_rem functions if division by 0
+// NOTE: there are aborts inside the specialized_div_rem functions if division by 0
 // is encountered, however these should be unreachable and optimized away unless
 // uses of `std/core::intrinsics::unchecked_div/rem` do not have a 0 check in front
 // of them.
diff --git a/src/int/specialized_div_rem/asymmetric.rs b/src/int/specialized_div_rem/asymmetric.rs
index 9d27b99d..133d8249 100644
--- a/src/int/specialized_div_rem/asymmetric.rs
+++ b/src/int/specialized_div_rem/asymmetric.rs
@@ -48,7 +48,8 @@ macro_rules! impl_asymmetric {
             let div_hi = (div >> n) as $uX;
             if div_hi == 0 {
                 if div_lo == 0 {
-                    panic!("division by zero");
+                    // division by zero
+                    ::abort();
                 }
                 if duo_hi < div_lo {
                     // plain $uD by $uX division that will fit into $uX
diff --git a/src/int/specialized_div_rem/binary_long.rs b/src/int/specialized_div_rem/binary_long.rs
index 7b10e952..dc80b75f 100644
--- a/src/int/specialized_div_rem/binary_long.rs
+++ b/src/int/specialized_div_rem/binary_long.rs
@@ -23,7 +23,8 @@ macro_rules! impl_binary_long {
         )*
         pub fn $unsigned_name(duo: $uX, div: $uX) -> ($uX,$uX) {
             if div == 0 {
-                panic!("division by zero")
+                // division by zero
+                ::abort();
             }
 
             // Full $uX binary long division. Use `leading_zeros` on the first round,
diff --git a/src/int/specialized_div_rem/delegate.rs b/src/int/specialized_div_rem/delegate.rs
index 9206e299..bd6fc117 100644
--- a/src/int/specialized_div_rem/delegate.rs
+++ b/src/int/specialized_div_rem/delegate.rs
@@ -38,7 +38,8 @@ macro_rules! impl_delegate {
 
             match (div_lo == 0, div_hi == 0, duo_hi == 0) {
                 (true, true, _) => {
-                    panic!("division by zero")
+                    // division by zero
+                    ::abort();
                 }
                 (_,false,true) => {
                     // `duo` < `div`
diff --git a/src/int/specialized_div_rem/trifecta.rs b/src/int/specialized_div_rem/trifecta.rs
index dec3cc7d..78b6af65 100644
--- a/src/int/specialized_div_rem/trifecta.rs
+++ b/src/int/specialized_div_rem/trifecta.rs
@@ -50,8 +50,12 @@ macro_rules! impl_trifecta {
 
             // the number of bits in a $uX
             let n = $n_h * 2;
-            // the number of bits in a $uD
-            let n_d = n * 2;
+
+            // This should be optimized away because of checks for zero upstream
+            if div == 0 {
+                // division by zero
+                ::abort();
+            }
 
             // Note that throughout this function, `lo` and `hi` refer to the high and low `n` bits
             // of a `$uD`, `0` to `3` refer to the 4 `n_h` bit parts of a `$uD`,
@@ -60,11 +64,6 @@ macro_rules! impl_trifecta {
             let div_lz = div.leading_zeros();
             let mut duo_lz = duo.leading_zeros();
 
-            // division by zero branch
-            if div_lz == n_d {
-                panic!("division by zero")
-            }
-
             // the possible ranges of `duo` and `div` at this point:
             // `0 <= duo < 2^n_d`
             // `1 <= div < 2^n_d`
diff --git a/src/int/udiv.rs b/src/int/udiv.rs
index 08905f25..a9415ed4 100644
--- a/src/int/udiv.rs
+++ b/src/int/udiv.rs
@@ -1,6 +1,6 @@
 use super::specialized_div_rem::*;
 
-// NOTE: there are panics inside the specialized_div_rem functions if division by 0
+// NOTE: there are aborts inside the specialized_div_rem functions if division by 0
 // is encountered, however these should be unreachable and optimized away unless
 // uses of `std/core::intrinsics::unchecked_div/rem` do not have a 0 check in front
 // of them.