novarobot
/
rllama
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
68 Commits
3 Branches
0 Tags
1.4 MiB
k4bit
master
broken-opencl-code
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from '40121e1c82'
${ noResults }
Commit Graph

3 Commits (40121e1c82e00a3f9567c7af66d632b3d41fca22)

Author SHA1 Message Date
Mikko Juola 40121e1c82 Multithread the k4 * f32 matrix multiplication. 3 years ago
Mikko Juola b8946da2d8 Implement matrix multiplication for 4-bit * 32-bit floats. 
As of this commit, test works. But I want to optimize this a bit, seeing
if increasing load instruction : arithmetic instruction ratio will make
single-threaded performance a bit speedier.
 3 years ago
Mikko Juola 9c86c17318 Refactor all SIMD to one file, simd_support.rs 
This should make it a bit easier to port to other SIMD instruction sets
when the SIMD instructions are not littered randomly around the
tensor.rs file.
 3 years ago