Expand description
ZFP codec implementation for the numcodecs
API.
This implementation uses ZFP’s
ZFP_ROUNDING_MODE=ZFP_ROUND_FIRST
and
ZFP_WITH_TIGHT_ERROR=ON
experimental features to reduce the bias and correlation in ZFP’s errors
(see https://zfp.readthedocs.io/en/release1.0.1/faq.html#zfp-rounding).
This implementation also rejects non-reversibly compressing non-finite (infinite or NaN) values, since ZFP’s behaviour for them is undefined (see https://zfp.readthedocs.io/en/release1.0.1/faq.html#q-valid).
Please see the numcodecs-zfp-classic
codec for an implementation that
uses ZFP without these modifications.
Structs§
- ZfpCodec
- Codec providing compression using ZFP
- ZfpHeader
Error - Opaque error for when encoding or decoding the header fails
Enums§
- ZfpCodec
Error - Errors that may occur when applying the
ZfpCodec
. - ZfpCompression
Mode - ZFP compression mode
- ZfpD
Type - Dtypes that Zfp can compress and decompress
Functions§
- compress
- Compress the
data
array using ZFP with the providedmode
. - decompress
- Decompress the
encoded
data into an array using ZFP. - decompress_
into - Decompress the
encoded
data into adecoded
array using ZFP.