fix(gpu): fix end index in gpu compression

zama-ai · Sep 30, 2024 · 03154d5 · 03154d5
1 parent 576bc57
commit 03154d5
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 1 deletion.
diff --git a/backends/tfhe-cuda-backend/cuda/include/device.h b/backends/tfhe-cuda-backend/cuda/include/device.h
@@ -45,6 +45,9 @@ void cuda_memcpy_async_to_gpu(void *dest, void *src, uint64_t size,
 void cuda_memcpy_async_gpu_to_gpu(void *dest, void *src, uint64_t size,
                                   cudaStream_t stream, uint32_t gpu_index);
 
+void cuda_memcpy_gpu_to_gpu(void *dest, void *src, uint64_t size,
+                            uint32_t gpu_index);
+
 void cuda_memcpy_async_to_cpu(void *dest, const void *src, uint64_t size,
                               cudaStream_t stream, uint32_t gpu_index);
 

diff --git a/tfhe/src/integer/gpu/ciphertext/compressed_ciphertext_list.rs b/tfhe/src/integer/gpu/ciphertext/compressed_ciphertext_list.rs
@@ -94,7 +94,7 @@ impl CudaCompressedCiphertextList {
             .map(DataKind::num_blocks)
             .sum();
 
-        let end_block_index = start_block_index + current_info.num_blocks();
+        let end_block_index = start_block_index + current_info.num_blocks() - 1;
 
         Some((
             decomp_key.unpack(