How to use cuZFP's lossless compression to compress int？

Open xusiang opened this issue 3 months ago • 1 comments

Hello， I tried to use cuZFP (https://github.com/mclarsen/cuZFP,which is said that this work has now been added to the official zfp repository) to compress int data in lossless mode.However,I tried to run https://github.com/mclarsen/cuZFP/tree/develop/src/tests/t_encode_decode_1.cpp : TEST(encode_decode, test_encode_decode_int32) { run_test<int>(512); } but I found the result lossy:

Decode elapsed time: 0.00001 (s) Decode rate: 158.95 (MB / sec) Total absolute error 434.00000000000000000000 Average abosulte error 0.84765625000000000000 with 512 values. How can I use lossless mode to compress int data?

Oct 09 '25 08:10 xusiang

Reversible (lossless) mode is not yet supported by the CUDA backend. See https://zfp.readthedocs.io/en/release1.0.1/execution.html#setting-the-execution-policy.

Oct 09 '25 15:10 lindstro