Commit Graph

46 Commits

Author SHA1 Message Date
R. Ryantm
27a673ef3e llama-cpp: 2700 -> 2746 2024-04-26 18:14:07 +00:00
Roman Zakirzyanov
aedebb76de llm-ls: 0.4.0 -> 0.5.2 2024-04-25 17:11:07 +03:00
R. Ryantm
fdcc5233d8 llama-cpp: 2674 -> 2700 2024-04-21 10:39:42 +00:00
R. Ryantm
1897af2d37 llama-cpp: 2636 -> 2674 2024-04-14 23:18:05 +00:00
R. Ryantm
ef16276bb6 llama-cpp: 2589 -> 2636 2024-04-09 21:33:46 +00:00
R. Ryantm
90891cd009 llama-cpp: 2568 -> 2589 2024-04-04 15:32:57 +00:00
Jonathan Ringer
65c4c21a2b llama-cpp: use pkgs.autoAddDriverRunpath 2024-03-31 10:15:47 -07:00
Joseph Stahl
a06a03ed7c
llama-cpp: update from b2481 to b2568 2024-03-28 20:56:55 -04:00
Joseph Stahl
e1ef3aaacc
llama-cpp: embed (don't pre-compile) metal shaders
port of https://github.com/ggerganov/llama.cpp/pull/6118, although compiling shaders with XCode disabled as it requires disabling sandbox (and only works on MacOS anyways)
2024-03-26 14:01:29 -04:00
Joseph Stahl
7aa588cc96
llama-cpp: rename cuBLAS to CUDA
Matches change from upstream 280345968d
2024-03-26 13:54:30 -04:00
Christian Kögler
2af438f836
llama-cpp: fix blasSupport (#298567)
* llama-cpp: fix blasSupport

* llama-cpp: switch from openblas to blas
2024-03-25 18:55:45 +01:00
R. Ryantm
c70ff30bde llama-cpp: 2454 -> 2481 2024-03-21 17:34:35 +00:00
Someone
e7797267a2
Merge pull request #281576 from yannham/refactor/cuda-setup-hooks-refactor
cudaPackages: generalize and refactor setup hooks
2024-03-19 20:06:18 +00:00
R. Ryantm
1c2a0b6df9 llama-cpp: 2424 -> 2454 2024-03-18 12:50:17 +00:00
Yann Hamdaoui
63746cac08
cudaPackages: generalize and refactor setup hook
This PR refactor CUDA setup hooks, and in particular
autoAddOpenGLRunpath and autoAddCudaCompatRunpathHook, that were using a
lot of code in common (in fact, I introduced the latter by copy pasting
most of the bash script of the former). This is not satisfying for
maintenance, as a recent patch showed, because we need to duplicate
changes to both hooks.

This commit abstract the common part in a single shell script that
applies a generic patch action to every elf file in the output. For
autoAddOpenGLRunpath the action is just addOpenGLRunpath (now
addDriverRunpath), and is few line function for
autoAddCudaCompatRunpathHook.

Doing so, we also takes the occasion to use the newer addDriverRunpath
instead of the previous addOpenGLRunpath, and rename the CUDA hook to
reflect that as well.

Co-Authored-By: Connor Baker <connor.baker@tweag.io>
2024-03-15 15:54:21 +01:00
R. Ryantm
4744ccc2db llama-cpp: 2382 -> 2424 2024-03-14 20:36:15 +00:00
R. Ryantm
bab5a87ffc llama-cpp: 2346 -> 2382 2024-03-10 12:09:01 +00:00
R. Ryantm
821ea0b581 llama-cpp: 2294 -> 2346 2024-03-05 19:34:21 +00:00
happysalada
03173009d5 llama-cpp: 2249 -> 2294; bring upstream flake 2024-02-28 19:44:34 -05:00
R. Ryantm
33294caa77 llama-cpp: 2212 -> 2249 2024-02-23 15:04:25 +00:00
R. Ryantm
3e909c9e10 llama-cpp: 2167 -> 2212 2024-02-20 05:27:18 +00:00
R. Ryantm
efa55a0426 llama-cpp: 2135 -> 2167 2024-02-16 21:01:12 +00:00
R. Ryantm
aee2928614 llama-cpp: 2105 -> 2135 2024-02-13 10:34:27 +00:00
R. Ryantm
e5a9f1c720 llama-cpp: 2074 -> 2105 2024-02-09 03:07:40 +00:00
R. Ryantm
27398a3fe2 llama-cpp: 2050 -> 2074 2024-02-06 00:29:01 +00:00
R. Ryantm
3a67f01b7f llama-cpp: 1892 -> 2050 2024-02-02 19:34:54 +00:00
happysalada
aaa2d4b738 llama-cpp: 1848->1892; add static build mode 2024-01-19 08:42:56 -05:00
Alex Martens
49309c0d27 llama-cpp: 1742 -> 1848 2024-01-12 19:01:51 -08:00
Weijia Wang
1f54e5e2a6
Merge pull request #278120 from r-ryantm/auto-update/llama-cpp
llama-cpp: 1710 -> 1742
2024-01-13 02:57:28 +01:00
R. Ryantm
b63bdd46cd llama-cpp: 1710 -> 1742 2024-01-01 18:52:11 +00:00
happysalada
47fc482e58 llama-cpp: fix cuda support; integrate upstream 2023-12-31 16:57:28 +01:00
Nick Cao
e51a04fa37
Merge pull request #277451 from accelbread/llama-cpp-update
llama-cpp: 1671 -> 1710
2023-12-29 10:42:00 -05:00
Archit Gupta
ab64ae8fdd llama-cpp: 1671 -> 1710 2023-12-28 18:19:55 -08:00
Archit Gupta
6cf4c910f9 llama-cpp: change default value of openblasSupport
The previous default caused build failures when `config.rocmSupport` was
enabled, since rocmSupport conflicts with openblasSupport.
2023-12-28 18:03:30 -08:00
Nick Cao
51b96e8410
Merge pull request #275807 from KaiHa/pr-llama-cpp-assert
llama-cpp: assert that only one of the *Support arguments is true
2023-12-22 08:36:34 -05:00
Kai Harries
ded8563fb3 llama-cpp: assert that only one of the *Support arguments is true
When experimenting with llama-cpp I experienced a mysterious build
error that in the end I tracked down to me setting openclSupport and
openblasSupport to true.  To prevent others from this mistake add an
assert that complains if more than one of the arguments openclSupport,
openblasSupport and rocmSupport is set to true.
2023-12-22 09:00:32 +01:00
Сухарик
6141d1bdd5 llama-cpp: 1573 -> 1671 2023-12-21 21:37:24 +03:00
R. Ryantm
72268042cf llama-cpp: 1538 -> 1573 2023-11-28 07:26:02 +00:00
annalee
53aeb5c67b llama-cpp: 1483 -> 1538 2023-11-19 10:25:46 +01:00
annalee
05967f9a2c llama-cpp: fix build due to openblas update
add patch so build checks for openblas64 module

https://github.com/NixOS/nixpkgs/pull/255443 updated from openblas
0.3.21 -> 0.3.24.  openblas 0.3.22 moved openblas.pc -> openblas64.pc

https://github.com/OpenMathLib/OpenBLAS/issues/3790
2023-11-19 10:25:46 +01:00
Martin Weinelt
ea1cd4ef8e
treewide: use config.rocmSupport
Makes use of the per nixpkgs config flag to enable rocmSupport in
packages that support it.
2023-11-14 01:51:57 +01:00
OTABI Tomoya
5a36456ad6
Merge pull request #265462 from r-ryantm/auto-update/llama-cpp
llama-cpp: 1469 -> 1483
2023-11-10 18:22:02 +09:00
Matthieu Coudron
ba774d337e
Merge pull request #263320 from jfvillablanca/llm-ls
llm-ls: init at 0.4.0
2023-11-06 11:09:01 +01:00
jfvillablanca
ab1da43942 llm-ls: init at 0.4.0 2023-11-06 08:38:31 +08:00
R. Ryantm
575af23354 llama-cpp: 1469 -> 1483 2023-11-04 13:58:41 +00:00
Enno Richter
ff77d3c409 llama-cpp: init at 1469 2023-11-02 09:31:40 +01:00