Commit Graph

16 Commits

Author SHA1 Message Date
ql 17c0115f3c ASTC-encoder update to 4.7.0
Signed-off-by: zhaonan287 <zhaonan34@huawei.com>
2024-05-16 10:11:13 +08:00
Pete Harris 2819e6203b Reduce ROM size (#323)
A number of ROM tables store entries that are never used in practice.

color_quant/unquant_tables store QUANT_2/3/4/5, which are illegal encodings. Truncate and bias.
quant_mode_table stores values for more than 20 color integers, which are illegal encodings. Truncate.
2022-04-19 21:43:22 +01:00
Pete Harris 79f6a6bd49 Use a ROM quant_mode_table to avoid parallel init (#284) 2021-08-20 07:37:01 +01:00
pete e6993d496d Fix /W4 warnings on VS2019 2021-06-17 15:55:20 +01:00
pete abd8955115 Use initializer not assignment 2021-06-14 15:36:46 +01:00
pete 34e4140b06 Make more use of unsigned ints 2 2021-05-22 14:24:06 +01:00
pete 5543738c58 Reflow comments on 100 character lines 2021-05-19 09:39:26 +01:00
pete c1e598ebe0 Refactor many function names 2021-05-18 23:47:36 +01:00
pete d4b6328b6c API cleanup for pick_best_endpoint_format 2021-05-16 21:31:01 +01:00
pete 7f857f16ab Update some copyright years 2021-02-12 21:11:14 +00:00
pete a1fb47e8b3 Shorten (un)quantize to quant in many places 2021-01-26 23:59:50 +00:00
pete 66fe3e9dbb Use a table-driven approach for computing ISE sizes 2021-01-26 23:52:15 +00:00
pete a325e533f4 Change quantization_mode_table to use int8_t 2021-01-12 18:48:18 +00:00
Pete Harris 9fce19699c Merge increment
* Update test data reference CSV

* Formatting cleanups

* Make image data type flexible
- No new types supported yet, but it's a precursor to just supporting f32 as an interface type.
2020-09-04 22:59:08 +01:00
Peter Harris 51d5963ba8 Allow decompression-only builds 2020-07-19 17:02:22 +01:00
Pete Harris 2b8fd08fc7 First phase of the front-end to codec API split (#125)
This phase introduces the first phase of the API split between the API front-end and the CODEC backed. However, some aspects of the final design are not yet supported:

This build only only supports automatic multi-threading; no user owned thread pools are supported.

Some memory allocations still tied to the compression pass, not the context, so reusing the context will have higher than planned CPU overhead due to the repeated memory allocation (although the impact is probably small unless compression small image).

The input image layout is still the legacy format and must be pre-padded by the caller for any pass using the area-based perceptual metrics.
2020-07-02 21:17:21 +01:00