Repository navigation

All pull requests

Labels (9)Milestones (0)New pull request

Open

(26)

Closed

(25)

Filtering: keep the MVE FIR q7/q15 fixed-tap kernels and the interpolators within the state
#51danjacobellis opened Oct 14, 2026
Filtering: keep the MVE df1 biquad reads within the source and destination
#50danjacobellis opened Oct 14, 2026
ML: load the MVE SVM and naive Bayes tail blocks with the tail predicate
#49danjacobellis opened Oct 14, 2026
Transform: keep the MVE Q15/Q31 RFFT split within the forward input
#48danjacobellis opened Oct 14, 2026
BasicMath: load the MVE clip tail blocks with the tail predicate
#47danjacobellis opened Oct 14, 2026
Include: stop float16 builds with GCC before 15.3 on M-profile cores without MVE
#46danjacobellis opened Oct 14, 2026
Matrix: build the QR Q of an m x n input from n reflectors
#45danjacobellis opened Oct 14, 2026
Filtering: keep the MVE FIR f16 within its documented state and input
#44danjacobellis opened Oct 14, 2026
Matrix: keep the last pivot of a rank-deficient LDL^t in D
#43danjacobellis opened Oct 14, 2026
Matrix: load the MVE transpose, product and QR tail blocks with the tail predicate
#42danjacobellis opened Oct 14, 2026
Filtering: keep the MVE q31 FIR at 29-31 taps within the documented state
#41danjacobellis opened Oct 14, 2026
Filtering: load the MVE convolution and correlation tail blocks with the tail predicate
#40danjacobellis opened Oct 14, 2026
Filtering: keep the MVE FIR q15/q7 general kernels within the state
#39danjacobellis opened Oct 14, 2026
Filtering: compute the remainder outputs of the MVE q31 FIR at 29-32 taps
#38danjacobellis opened Oct 14, 2026
Filtering: keep the MVE Q31 FIR within Arm's documented state length
#37danjacobellis opened Oct 14, 2026
Filtering: correct the MVE FIR q15/q7 general-kernel remainder outputs
#36danjacobellis opened Oct 14, 2026
Matrix: keep the MVE LDL^t f32 update within its operands
#35danjacobellis opened Oct 14, 2026
Filtering: run the MVE df1 Q15/Q31 biquads at small block sizes
#34danjacobellis opened Oct 14, 2026
CommonTables: compile the MVE tables without the HELIUM option
#33danjacobellis opened Oct 14, 2026
Statistics: load the MVE tail block of 18 kernels with the tail predicate
#32danjacobellis opened Oct 14, 2026
Statistics: saturate the absolute value in the MVE absmin Q7/Q15/Q31
#31danjacobellis opened Oct 14, 2026
Filtering: keep the MVE f32 FIR within Arm's documented state length
#30danjacobellis opened Oct 14, 2026
Load B contiguously in the MVE f16 4x4 matrix product
#27danjacobellis opened Oct 9, 2026
Reduce the MVE f32 matrix-vector accumulators by a transpose under GCC
#26danjacobellis opened Oct 9, 2026
Reduce the MVE f32 decimator accumulators by a transpose under GCC
#25danjacobellis opened Oct 9, 2026

Pull requests