llama.cpp
SYCL: Add fp16 type support to unary op kernels
#12788

Merged

SYCL: Add fp16 type support to unary op kernels #12788

NeoZhangJianyu merged 7 commits into master from sycl/fp16elewisesupport

github-actions added ggml

github-actions added SYCL

qnixsynapse requested a review from

Rbiessy 1 year ago

qnixsynapse requested a review from

Alcpz 1 year ago

qnixsynapse requested a review from

NeoZhangJianyu 1 year ago

qnixsynapse marked this pull request as draft 1 year ago

qnixsynapse marked this pull request as ready for review 1 year ago

AD2605 commented on 2025-04-08

NeoZhangJianyu commented on 2025-04-09

NeoZhangJianyu approved these changes on 2025-04-09

qnixsynapse force pushed to fc8d0a6b 1 year ago

Rbiessy commented on 2025-04-09

Rbiessy approved these changes on 2025-04-10

SYCL: Add fp16 support to some elementwise OP kernels

4c61c275

qnixsynapse force pushed from eed23cdf to 83980606 1 year ago

remove comment

8f89e132

Use static_cast directly

0219e20d

remove not needed cast from tanh

ecc5eaf9

Use static cast and remove unneeded castings

12914658

Adjust device_support_op for unary OPs

f6a7c5f7

Use cast_data and typed_data struct to deduplicate casting code

83980606

NeoZhangJianyu merged fccf9cae into master 1 year ago

qnixsynapse deleted the sycl/fp16elewisesupport branch 1 year ago

Reviewers

Rbiessy

NeoZhangJianyu

AD2605

Alcpz

Assignees

No one assigned

Labels

ggml SYCL

Milestone

No milestone

llama.cpp SYCL: Add fp16 type support to unary op kernels #12788 Merged

SYCL: Add fp16 type support to unary op kernels #12788

llama.cpp
SYCL: Add fp16 type support to unary op kernels
#12788

Merged