github/tfhe-rs
1
1
Fork 0
mirror of https://github.com/zama-ai/tfhe-rs.git synced 2026-01-08 22:28:01 -05:00
Code Issues Packages Projects Releases Wiki Activity

DO NOT MERGE: bench without device sync in drop

Browse Source
This commit is contained in:
Agnes Leroy
2025-10-14 10:11:41 +02:00
parent 538e886f13
commit 998e1f2a36
1 changed files with 2 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
tfhe/src/core_crypto/gpu/vec.rs
View File

@@ -1,6 +1,6 @@
use super::get_number_of_gpus;
use crate::core_crypto::gpu::slice::{CudaSlice, CudaSliceMut};
use crate::core_crypto::gpu::{synchronize_device, CudaStreams};
use crate::core_crypto::gpu::CudaStreams;
use crate::core_crypto::prelude::Numeric;
use std::collections::Bound::{Excluded, Included, Unbounded};
use std::ffi::c_void;
@@ -471,19 +471,16 @@ impl<T: Numeric> CudaVec<T> {
//
// clippy complains that we impl Send on CudaVec while CudaPtr is non Send.
// This is ok for us, as CudaPtr is meant to be a wrapper type that serves
// as distinguishing ptr that points to cuda memory from pointers pointing to
// as distinguishing pointers that point to cuda memory from pointers pointing to
// CPU memory.
#[allow(clippy::non_send_fields_in_send_ty)]
unsafe impl<T> Send for CudaVec<T> where T: Send + Numeric {}
unsafe impl<T> Sync for CudaVec<T> where T: Sync + Numeric {}
impl<T: Numeric> Drop for CudaVec<T> {
/// Free memory for pointer `ptr` synchronously
fn drop(&mut self) {
for (i, &ptr) in self.ptr.iter().enumerate() {
// Synchronizes the device to be sure no stream is still using this pointer
let gpu_index = self.gpu_indexes[i];
synchronize_device(gpu_index.0);
unsafe { cuda_drop(ptr, gpu_index.0) };
}
}