feat(gpu): implement CUDA-based Radix Integer compression #1424

pdroalves · 2024-07-25T19:49:40Z

Implements CudaCompressedCiphertextList, which can be used to compress many Cuda(Unsigned/Signed)RadixCiphertext.

Tests for the changes have been added (for bug fixes / features)
Docs have been added / updated (for bug fixes / features)
Relevant issues are marked as resolved/closed, related issues are linked in the description
Check for breaking changes (including serialization changes) and add them to commit message following the conventional commit specification

IceTDrinker

What I could see, note that compression and CUDA in TFHE are not necessarily my areas of expertise in the code base

backends/tfhe-cuda-backend/src/cuda_bind.rs

tfhe/benches/integer/bench.rs

tfhe/src/integer/gpu/list_compression/server_keys.rs

tfhe/src/core_crypto/gpu/algorithms/test/lwe_packing_keyswitch.rs

tfhe/src/core_crypto/gpu/mod.rs

tfhe/src/integer/client_key/radix.rs

tfhe/src/integer/gpu/client_key/radix.rs

tfhe/src/integer/gpu/list_compression/server_keys.rs

tfhe/src/integer/gpu/mod.rs

tfhe/src/integer/gpu/list_compression/server_keys.rs

tmontaigu · 2024-08-16T11:16:22Z

tfhe/src/integer/gpu/list_compression/server_keys.rs

+        start_block_index: usize,
+        end_block_index: usize,


start_block_index: usize, end_block_index: usize,

could directly be block_range: RangeInclusive<usize>

How this line would look like?

let blocks = packed_list.block_info[start_block_index..=end_block_index].to_vec();

let blocks = packed_list.block_info[block_range].to_vec();

…functional packing keyswitch

cla-bot bot added the cla-signed label Jul 25, 2024

pdroalves force-pushed the pa/feat/compression branch from a4b5851 to 207a76a Compare July 25, 2024 20:19

pdroalves changed the base branch from main to pa/refactor/sampleextraction July 25, 2024 20:19

pdroalves force-pushed the pa/feat/compression branch from 207a76a to 9c29d8d Compare July 25, 2024 20:22

pdroalves force-pushed the pa/refactor/sampleextraction branch 12 times, most recently from 249a7f0 to 40ee415 Compare July 29, 2024 13:58

pdroalves force-pushed the pa/feat/compression branch 4 times, most recently from cff04e5 to 7efa2c6 Compare August 1, 2024 20:50

pdroalves force-pushed the pa/refactor/sampleextraction branch from 40ee415 to a6fda27 Compare August 1, 2024 20:59

Base automatically changed from pa/refactor/sampleextraction to main August 2, 2024 09:10

pdroalves force-pushed the pa/feat/compression branch 3 times, most recently from 18533aa to c6b45c6 Compare August 6, 2024 13:53

pdroalves changed the title ~~Pa/feat/compression~~ feat(gpu): implement CUDA-based Radix Integer compression Aug 6, 2024

pdroalves force-pushed the pa/feat/compression branch 4 times, most recently from dbfbf48 to 9dbd262 Compare August 6, 2024 20:39

pdroalves force-pushed the pa/feat/compression branch from f40eaef to f7a5f4b Compare August 9, 2024 20:51

IceTDrinker reviewed Aug 12, 2024

View reviewed changes

pdroalves force-pushed the pa/feat/compression branch 9 times, most recently from 0d690d0 to 3b19e61 Compare August 13, 2024 13:34

pdroalves requested a review from IceTDrinker August 14, 2024 12:15

pdroalves force-pushed the pa/feat/compression branch 5 times, most recently from 416bb9f to 08c4750 Compare August 14, 2024 16:19

pdroalves requested a review from tmontaigu August 14, 2024 17:03