nicolas
/
ik_llama_opt
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
ik_llama_opt/ggml/src
History
Kawrakow 82c4f27332
Fuse the attention gate in Step-3.5-Flash (#1244) 
* WIP

* This works but is slow

* Turn off the up / gate clamps for now

* OK we need the clamping

* Fuse the clamp (CUDA)

* Fuse the clamp (CPU)

* WIP

* Be able to use merged q, k, v

* Be able to use merged up/gate experts

* Fuse the clamp (CUDA mmvq)

* WIP: graph parallel for Step-3.5

* WIP

* This should be it

* Cleanup

* Fix merge

* Not working attempt to extend fused_mul_unary to the Step-3.5 case

* It works now, but performance gain is very minor
 		2026-02-07 07:56:58 +02:00
..
cmake Merge vulkan code from mainline up to commit of 6/28/2025 (#563) 2025-07-02 08:49:42 +02:00
ggml-cann Merge mainline - Aug 12 2024 (#17) 2024-08-12 15:14:32 +02:00
ggml-cuda Fuse the attention gate in Step-3.5-Flash (#1244) 2026-02-07 07:56:58 +02:00
ggml-sycl Merge mainline - Aug 12 2024 (#17) 2024-08-12 15:14:32 +02:00
iqk Step-3.5-Flash support (#1231) 2026-02-05 08:13:22 +02:00
kompute@4565194ed7 Merge mainline llama.cpp (#3) 2024-07-27 07:55:01 +02:00
kompute-shaders Merge mainline llama.cpp (#3) 2024-07-27 07:55:01 +02:00
llamafile Merge mainline llama.cpp (#3) 2024-07-27 07:55:01 +02:00
vulkan-shaders Port of Qwen3-VL support from mainline (#883) 2025-11-04 19:20:54 +02:00
CMakeLists.txt Remove llamafile remnants (#1179) 2026-01-22 13:20:23 +02:00
ggml-aarch64.c Merge mainline - Aug 12 2024 (#17) 2024-08-12 15:14:32 +02:00
ggml-aarch64.h Merge mainline llama.cpp (#3) 2024-07-27 07:55:01 +02:00
ggml-alloc.c Enable CUDA graphs for MoE models + GPT-OSS support (#689) 2025-08-15 09:18:07 +03:00
ggml-backend-impl.h Merge vulkan code from mainline up to commit of 6/28/2025 (#563) 2025-07-02 08:49:42 +02:00
ggml-backend.cpp Fix build failure when OpenMP is not available (#1171) 2026-01-22 12:26:23 +02:00
ggml-blas.cpp Merge mainline - Aug 12 2024 (#17) 2024-08-12 15:14:32 +02:00
ggml-cann.cpp Merge vulkan code from mainline up to commit of 6/28/2025 (#563) 2025-07-02 08:49:42 +02:00
ggml-common.h AVX512+AVXVNNI GEMM implementation for quants using Q8_K for activations (#710) 2025-08-22 06:27:07 +03:00
ggml-cuda.cu Change default FA offset to ln(2) (#1235) 2026-02-05 13:42:53 +02:00
ggml-impl.h MXFP4 (#682) 2025-08-09 08:40:18 +03:00
ggml-kompute.cpp Merge vulkan code from mainline up to commit of 6/28/2025 (#563) 2025-07-02 08:49:42 +02:00
ggml-metal.m MXFP4 (#682) 2025-08-09 08:40:18 +03:00
ggml-metal.metal MXFP4 (#682) 2025-08-09 08:40:18 +03:00
ggml-quants.c Fix avx2 GEMM mess (v2) (#724) 2025-08-27 08:03:47 +03:00
ggml-quants.h IQ1_M_R4: better 1.75 bpw quants (#187) 2025-02-06 14:08:52 +02:00
ggml-rpc.cpp server: improve speed of speculative decoding (#1119) 2026-01-10 08:01:22 +02:00
ggml-sycl.cpp Merge vulkan code from mainline up to commit of 6/28/2025 (#563) 2025-07-02 08:49:42 +02:00
ggml-vulkan.cpp Port of Qwen3-VL support from mainline (#883) 2025-11-04 19:20:54 +02:00
ggml.c Fuse the attention gate in Step-3.5-Flash (#1244) 2026-02-07 07:56:58 +02:00