diff --git a/fhevm-engine/Cargo.toml b/fhevm-engine/Cargo.toml
index 829df784..aa5243c3 100644
--- a/fhevm-engine/Cargo.toml
+++ b/fhevm-engine/Cargo.toml
@@ -20,8 +20,10 @@ rayon = "1.10.0"
 [profile.dev.package.tfhe]
 overflow-checks = false
 
-# for testing in release mode due to too big
-# binary inside mac
 [profile.release]
-opt-level = "z"
+# for testing in release mode due to too big binary inside mac:
+#  set opt-level = "z"
+# however, this leads to 2-4x slower execution due to loss of loop
+# vectorization
+opt-level = 3
 lto = "fat"
diff --git a/fhevm-engine/fhevm-engine-common/Cargo.toml b/fhevm-engine/fhevm-engine-common/Cargo.toml
index 237f8d1f..7c837673 100644
--- a/fhevm-engine/fhevm-engine-common/Cargo.toml
+++ b/fhevm-engine/fhevm-engine-common/Cargo.toml
@@ -4,9 +4,9 @@ version = "0.1.0"
 edition = "2021"
 
 [target.'cfg(target_arch = "x86_64")'.dependencies]
-tfhe = { version = "0.8.2", features = ["boolean", "shortint", "integer", "x86_64-unix", "zk-pok", "experimental-force_fft_algo_dif4"] }
+tfhe = { version = "0.8.2", features = ["boolean", "shortint", "integer", "x86_64-unix", "zk-pok", "experimental-force_fft_algo_dif4", "nightly-avx512"] }
 [target.'cfg(target_arch = "aarch64")'.dependencies]
-tfhe = { version = "0.8.2", features = ["boolean", "shortint", "integer", "aarch64-unix", "zk-pok", "experimental-force_fft_algo_dif4"] }
+tfhe = { version = "0.8.2", features = ["boolean", "shortint", "integer", "aarch64-unix", "zk-pok", "experimental-force_fft_algo_dif4", "nightly-avx512"] }
 
 [dependencies]
 sha3.workspace = true