diff --git a/tfhe/benches/high_level_api/erc20.rs b/tfhe/benches/high_level_api/erc20.rs
index a53d325f11..295a40c6ea 100644
--- a/tfhe/benches/high_level_api/erc20.rs
+++ b/tfhe/benches/high_level_api/erc20.rs
@@ -221,8 +221,12 @@ fn bench_transfer_throughput<FheType, F>(
     F: for<'a> Fn(&'a FheType, &'a FheType, &'a FheType) -> (FheType, FheType) + Sync,
 {
     let mut rng = thread_rng();
+    #[cfg(not(feature = "gpu"))]
+    let num_gpus = 1u64;
+    #[cfg(feature = "gpu")]
+    let num_gpus = unsafe { cuda_get_number_of_gpus() } as u64;
 
-    for num_elems in [10, 100, 500] {
+    for num_elems in [10 * num_gpus, 100 * num_gpus, 500 * num_gpus] {
         group.throughput(Throughput::Elements(num_elems));
         let bench_id = format!("{bench_name}::{fn_name}::{type_name}::{num_elems}_elems");
         group.bench_with_input(&bench_id, &num_elems, |b, &num_elems| {
@@ -262,6 +266,8 @@ fn bench_transfer_throughput<FheType, F>(
 
 #[cfg(feature = "pbs-stats")]
 use pbs_stats::print_transfer_pbs_counts;
+#[cfg(feature = "gpu")]
+use tfhe_cuda_backend::cuda_bind::cuda_get_number_of_gpus;
 
 fn main() {
     #[cfg(not(feature = "gpu"))]