llama.cpp
ggml: backend-agnostic tensor parallelism
#19378
Open
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Overview
Commits
15
Changes
View On
GitHub
ggml: backend-agnostic tensor parallelism
#19378
JohannesGaessler
wants to merge 15 commits into
ggml-org:master
from
JohannesGaessler:ggml-meta-backend-8
JohannesGaessler
requested a review
from
CISC
6 days ago
JohannesGaessler
requested a review
from
ggerganov
6 days ago
JohannesGaessler
requested a review
from
taronaeo
6 days ago
JohannesGaessler
requested a review
from
reeselevine
6 days ago
JohannesGaessler
requested a review
from
0cc4m
6 days ago
JohannesGaessler
requested a review
from
rgerganov
6 days ago
JohannesGaessler
requested a review
from
max-krasnyansky
6 days ago
JohannesGaessler
requested a review
from
lhez
6 days ago
github-actions
added
Nvidia GPU
github-actions
added
Vulkan
github-actions
added
examples
github-actions
added
ggml
github-actions
added
SYCL
github-actions
added
Apple Metal
github-actions
added
Ascend NPU
github-actions
added
OpenCL
github-actions
added
IBM zDNN
jeffbolznv
commented on 2026-02-06
ggml: backend-agnostic tensor parallelism
a0d9dd20
support for GPT-OSS, Qwen 3 MoE
ab69c58a
partial Vulkan fix
4b8aa266
add support for 4/8 GPUs
2ffa49de
unconditional peer access
02325685
re-use buffers + ggml contexts
c9255634
fix output pattern
c5314444
NCCL support
8de41b5b
JohannesGaessler
force pushed
from
99928361
to
fca3954a
1 day ago
IMbackK
requested changes on 2026-02-11
GGML: HIP: add RCCL support
29c5327d
Remove shfl and AllReduce from backend interface
4dc3d10e
move allocation workaround out of ggml-alloc.c
76d94392
2d tensor set/get support
3fdd0b7a
JohannesGaessler
force pushed
from
fca3954a
to
3fdd0b7a
21 hours ago
Fix the seg fault without NCCL
10385e8f
Apply suggestion from @JohannesGaessler
9bb9d783
Merge pull request #4 from gaugarg-nv/minor_fixes
b12a5635
Login to write a write a comment.
Login via GitHub
Reviewers
IMbackK
jeffbolznv
CISC
ggerganov
taronaeo
reeselevine
0cc4m
rgerganov
max-krasnyansky
lhez
Assignees
No one assigned
Labels
Nvidia GPU
Vulkan
examples
ggml
SYCL
Apple Metal
Ascend NPU
OpenCL
IBM zDNN
Milestone
No milestone
Login to write a write a comment.
Login via GitHub