Kawrakow e1164e1fd8 Adding IQ1_KT - 1.75 bpw SOTA quants (#616) ...

* iq1_kt: basics

* iq1_kt: CUDA dequantize

Testing with LlaMA-3.1-8B-Instruct, we get almost the same PPL
as iq2_xxs, so about 0.2 bpw fewer bits for the same quality.

* iq1_kt: CUDA MMQ

* iq1_kt: CUDA MMVQ

* iq1_kt: AVX2 GEMM/GEMV

* iq1_kt: convert/repack to q8_0_r8 (AVX2)

* iq1_kt: slightly faster GEMV

18.6 t/s -> 19.4 t/s

* iq1_kt: NEON GEMM/GEMV

Pathetic as usual

* iq1_kt: slightly faster NEON - still pathetic

* iq1_kt: tiny bit better GEMV on NEON

* iq1_kt: convert/repack to q8_0_r8 (NEON)

* iq1_kt: very slightly faster convert/repack to q8_0_r8 on NEON

* Adding frgotten file

* iq1_kt: add to constants.py

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>

2025-07-20 10:05:23 +02:00

..

cmake

Merge mainline llama.cpp (#3)

2024-07-27 07:55:01 +02:00

include

Adding IQ1_KT - 1.75 bpw SOTA quants (#616)

2025-07-20 10:05:23 +02:00

src

Adding IQ1_KT - 1.75 bpw SOTA quants (#616)

2025-07-20 10:05:23 +02:00

.gitignore

Merge mainline llama.cpp (#3)

2024-07-27 07:55:01 +02:00

CMakeLists.txt

Add GGML_MAX_CONTEXTS definition in CMakeLists.txt (#622)

2025-07-17 08:50:42 +02:00