github/tfhe-rs
1
1
Fork 0
mirror of https://github.com/zama-ai/tfhe-rs.git synced 2026-01-09 22:57:59 -05:00
Code Issues Packages Projects Releases Wiki Activity

feat(hpu): integer bench update (adds mod, div -> div_mod), erc20_simd simd batch size read from iop prototype

Browse Source
This commit is contained in:
pgardratzama
2025-09-09 17:21:14 +02:00
committed by Pierre Gardrat
parent 1530f52c79
commit 4ff0d6cac2
2 changed files with 18 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
tfhe-benchmark/benches/high_level_api/erc20.rs
View File

@@ -18,8 +18,6 @@ use tfhe::prelude::*;
use tfhe::GpuIndex;
use tfhe::{set_server_key, ClientKey, CompressedServerKey, FheBool, FheUint64};
pub const HPU_SIMD_N: usize = 9;
/// Transfer as written in the original FHEvm white-paper,
/// it uses a comparison to check if the sender has enough,
/// and cmuxes based on the comparison result
@@ -393,9 +391,8 @@ fn bench_transfer_latency_simd<FheType, F>(
FheType: FheWait,
F: for<'a> Fn(&'a Vec<FheType>, &'a Vec<FheType>, &'a Vec<FheType>) -> Vec<FheType>,
{
#[cfg(feature = "gpu")]
configure_gpu(client_key);
use tfhe::tfhe_hpu_backend::prelude::hpu_asm;
let hpu_simd_n = hpu_asm::iop::IOP_ERC_20_SIMD.format().unwrap().proto.src.len()/3;
let bench_id = format!("{bench_name}::{fn_name}::{type_name}");
c.bench_function(&bench_id, |b| {
let mut rng = thread_rng();
@@ -403,7 +400,7 @@ fn bench_transfer_latency_simd<FheType, F>(
let mut from_amounts: Vec<FheType> = vec![];
let mut to_amounts: Vec<FheType> = vec![];
let mut amounts: Vec<FheType> = vec![];
for _i in 0..HPU_SIMD_N-1 {
for _i in 0..hpu_simd_n {
let from_amount = FheType::encrypt(rng.gen::<u64>(), client_key);
let to_amount = FheType::encrypt(rng.gen::<u64>(), client_key);
let amount = FheType::encrypt(rng.gen::<u64>(), client_key);
@@ -648,30 +645,32 @@ fn hpu_bench_transfer_throughput_simd<FheType, F>(
FheType: FheWait,
F: for<'a> Fn(&'a Vec<FheType>, &'a Vec<FheType>, &'a Vec<FheType>) -> Vec<FheType> + Sync,
{
use tfhe::tfhe_hpu_backend::prelude::hpu_asm;
let hpu_simd_n = hpu_asm::iop::IOP_ERC_20_SIMD.format().unwrap().proto.src.len()/3;
let mut rng = thread_rng();
for num_elems in [2, 10] {
group.throughput(Throughput::Elements(num_elems*(HPU_SIMD_N as u64)));
let real_num_elems = num_elems*(hpu_simd_n as u64);
group.throughput(Throughput::Elements(real_num_elems));
let bench_id =
format!("{bench_name}::throughput::{fn_name}::{type_name}::{num_elems}_elems");
format!("{bench_name}::throughput::{fn_name}::{type_name}::{real_num_elems}_elems");
group.bench_with_input(&bench_id, &num_elems, |b, &num_elems| {
let from_amounts = (0..num_elems)
.map(|_| {
(0..HPU_SIMD_N-1)
(0..hpu_simd_n)
.map(|_| FheType::encrypt(rng.gen::<u64>(), client_key))
.collect()
})
.collect::<Vec<_>>();
let to_amounts = (0..num_elems)
.map(|_| {
(0..HPU_SIMD_N-1)
(0..hpu_simd_n)
.map(|_| FheType::encrypt(rng.gen::<u64>(), client_key))
.collect()
})
.collect::<Vec<_>>();
let amounts = (0..num_elems)
.map(|_| {
(0..HPU_SIMD_N-1)
(0..hpu_simd_n)
.map(|_| FheType::encrypt(rng.gen::<u64>(), client_key))
.collect()
})

9
tfhe-benchmark/benches/integer/bench.rs
View File

@@ -3164,14 +3164,19 @@ mod hpu {
);
define_hpu_bench_default_fn!(
iop_name: div,
display_name: div
display_name: div_mod
);
define_hpu_bench_default_fn!(
iop_name: mod,
display_name: modulo
);
criterion_group!(
default_hpu_ops,
default_hpu_add,
default_hpu_sub,
default_hpu_mul,
default_hpu_div
default_hpu_div,
default_hpu_mod
);
// Alu Scalar -----------------------------------------------------------------