Skip to content

vulkan: optimize Q2_K and Q3_K mul_mat_vec #10459

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
Nov 27, 2024
Merged

vulkan: optimize Q2_K and Q3_K mul_mat_vec #10459

merged 1 commit into from
Nov 27, 2024

Conversation

jeffbolznv
Copy link
Collaborator

Similar to optimizations I did recently for Q{4-6}_K.

Perf results on RTX 4070:

before
| llama 8B Q2_K - Medium         |   2.95 GiB |     8.03 B | Vulkan     | 1000 |         tg128 |         59.12 ± 0.66 |
| deepseek2 16B Q2_K - Medium    |   5.99 GiB |    15.71 B | Vulkan     | 1000 |         tg128 |        102.23 ± 0.43 |

after
| llama 8B Q2_K - Medium         |   2.95 GiB |     8.03 B | Vulkan     | 1000 |         tg128 |         72.17 ± 0.51 |
| deepseek2 16B Q2_K - Medium    |   5.99 GiB |    15.71 B | Vulkan     | 1000 |         tg128 |        109.92 ± 0.62 |

@jeffbolznv jeffbolznv requested a review from 0cc4m November 23, 2024 02:14
0cc4m
0cc4m approved these changes Nov 27, 2024
View reviewed changes
Copy link
Collaborator

@0cc4m 0cc4m left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good on Nvidia, AMD and Intel. Thank you.

@0cc4m 0cc4m merged commit 4a57d36 into ggml-org:master Nov 27, 2024
7 checks passed
@stduhpf stduhpf mentioned this pull request Dec 7, 2024
Closed
arthw pushed a commit to arthw/llama.cpp that referenced this pull request Dec 20, 2024
@jeffbolznv @arthw
vulkan: optimize Q2_K and Q3_K mul_mat_vec (ggml-org#10459)
9bb00fe
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@0cc4m 0cc4m 0cc4m approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@jeffbolznv @0cc4m