-
Notifications
You must be signed in to change notification settings - Fork 460
Pull requests: google/highway
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Faster NEON CompressStore: use fused BitsFromMask and PopCount
ready to pull
#3277
opened Aug 12, 2026 by
a-eridani
Loading…
2x unroll IndexOfMin/IndexOfMax, plus MaskedEq and ReduceMax suggestions
ready to pull
#3275
opened Aug 11, 2026 by
Wint3rNight
Contributor
Loading…
hwy btree: Reduce internal_pool_ chunk size to 4 KB to eliminate small-tree tail slack.
#3274
opened Aug 11, 2026 by
copybara-service
Bot
Loading…
contrib/math: preserve negative zero in Expm1
#3273
opened Aug 11, 2026 by
EylonKrause
Contributor
Loading…
Add float16 exp/log math functions via promotion to float32
ready to pull
#3228
opened Jul 22, 2026 by
dcsid
Contributor
Loading…
Add attention kernels optimized for arm's i8mm instruction.
#3168
opened Jun 30, 2026 by
copybara-service
Bot
Loading…
Updated implementation of Compress ops
ready to pull
#3166
opened Jun 30, 2026 by
johnplatts
Contributor
Loading…
Add implementations of Reduce2/4/8 to quickly compute simultaneous sum/min/max of N registers + reusable template for those functions.
#3115
opened Jun 12, 2026 by
copybara-service
Bot
Loading…
Helper functions to use arm matrix multiplication instructions in a way that is testable on other platofrms
#3101
opened Jun 10, 2026 by
copybara-service
Bot
Loading…
Create executable notebook for simple vector add example
#3047
opened May 11, 2026 by
bkmgit
Contributor
Loading…
Replace Tanh with FastTanh in flash_attention
#3006
opened Apr 23, 2026 by
copybara-service
Bot
Loading…
Use CappedTag to prevent potential out of bound reads.
#2895
opened Mar 5, 2026 by
copybara-service
Bot
Loading…
Add FastGelu activation function in a newly created created fast_ops-inl.h files.
#2881
opened Feb 24, 2026 by
copybara-service
Bot
Loading…
[WIP] Try to support header only mode for core library.
#2832
opened Dec 24, 2025 by
LanderlYoung
Contributor
Loading…
contrib/sort: enable f16/f64 sorting on no native FP
ready to pull
#2737
opened Sep 28, 2025 by
seiko2plus
Contributor
Loading…
Use
LoadN for short BF16 dot product inputs instead of scalar conversion
ready to pull
#2703
opened Sep 10, 2025 by
vekterli
Contributor
Loading…
Previous Next
ProTip!
Type g i on any issue or pull request to go back to the issue listing page.