Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fix(gpu): fix scalar ne #1864

Merged
merged 1 commit into from
Dec 12, 2024
Merged

fix(gpu): fix scalar ne #1864

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 6 additions & 2 deletions backends/tfhe-cuda-backend/cuda/src/integer/comparison.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -187,14 +187,18 @@ __host__ void is_at_least_one_comparisons_block_true(
uint32_t remaining_blocks = num_radix_blocks;
while (remaining_blocks > 0) {
// Split in max_value chunks
uint32_t chunk_length = std::min(max_value, remaining_blocks);
int num_chunks = remaining_blocks / chunk_length;
int num_chunks = (remaining_blocks + max_value - 1) / max_value;

// Since all blocks encrypt either 0 or 1, we can sum max_value of them
// as in the worst case we will be adding `max_value` ones
auto input_blocks = mem_ptr->tmp_lwe_array_out;
auto accumulator = buffer->tmp_block_accumulated;
uint32_t chunk_lengths[num_chunks];
auto begin_remaining_blocks = remaining_blocks;
for (int i = 0; i < num_chunks; i++) {
uint32_t chunk_length =
std::min(max_value, begin_remaining_blocks - i * max_value);
chunk_lengths[i] = chunk_length;
accumulate_all_blocks<Torus>(streams[0], gpu_indexes[0], accumulator,
input_blocks, big_lwe_dimension,
chunk_length);
Expand Down
Loading