llama.cpp

Commit Graph

Author	SHA1	Message	Date
Aaron Teo	adbfbf9086	ggml-blas: refactor backend Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-20 15:11:45 +08:00
Aaron Teo	2ee4d5fe2f	ggml-blas: fix graph realloc Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-20 14:03:24 +08:00
Aaron Teo	623e7135c2	ggml-blas: fix memleak Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-20 13:37:26 +08:00
Aaron Teo	04ed19bbc0	ggml-blas: further cleanup Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-14 23:37:56 +08:00
Aaron Teo	10ce5e056d	ggml-blas: more code formatting Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-14 23:20:28 +08:00
Aaron Teo	75e506ff22	ggml-blas: clean up code Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-14 23:19:22 +08:00
Aaron Teo	7998d08b29	ggml-blas: bring back openmp Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-14 23:07:54 +08:00
Aaron Teo	e481be6da6	ggml-blas: move global blas n threads to set_n_threads Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-14 22:19:12 +08:00
Aaron Teo	6dff031caa	ggml-blas: force dequant routine to use max logical cores Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-14 21:57:09 +08:00
Aaron Teo	447057973c	ggml-blas: fix ne Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-14 19:27:42 +08:00
Aaron Teo	717531b1a7	ggml-blas: add note Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-14 19:22:14 +08:00
Aaron Teo	aae6d1e9b0	ggml-blas: fix invalid data access Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-14 19:15:34 +08:00
Aaron Teo	9a14a094ac	ggml: rewrite ggml-blas Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-14 18:06:31 +08:00
Aaron Teo	1926e07e1a	ggml-blas: code clean up Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-11 21:27:13 +08:00
Aaron Teo	19c8ec9964	ggml-blas: fully working mmid Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-11 21:10:25 +08:00
Aaron Teo	f682374613	ggml-blas: initial mmid impl Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>	2025-12-11 20:51:02 +08:00
Jeff Bolz	c0b45097c3	rename optimize_graph to graph_optimize (#16082 )	2025-09-18 13:46:17 -05:00
Jeff Bolz	e68aa10d8f	vulkan: sort graph to allow more parallel execution (#15850 ) * vulkan: sort graph to allow more parallel execution Add a backend proc to allow the backend to modify the graph. The vulkan implementation looks at which nodes depend on each other and greedily reorders them to group together nodes that don't depend on each other. It only reorders the nodes, doesn't change the contents of any of them. With #15489, this reduces the number of synchronizations needed. * call optimize_graph per-split	2025-09-09 02:10:07 +08:00
AN Long	cd6983d56d	ggml : fix field name when new ggml_backend (#14944 )	2025-08-08 14:37:22 +02:00
Diego Devesa	5931c1f233	ggml : add support for dynamic loading of backends (#10469 ) * ggml : add support for dynamic loading of backends --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>	2024-11-25 15:13:39 +01:00
Diego Devesa	ae8de6d50a	ggml : build backends as libraries (#10256 ) * ggml : build backends as libraries --------- Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> Co-authored-by: R0CKSTAR <xiaodong.ye@mthreads.com>	2024-11-14 18:04:35 +01:00

21 Commits