Dev/cuda - #5

Open

Ra4ster wants to merge 32 commits into

mainfrom

Commits on Sep 5, 2026

Now implementing CUDA backend. WIP.
Ra4ster
committed
4afc0a0
See last commit.
Ra4ster
committed
3b03e29
Implemented CUDABackend.cpp
Ra4ster
committed
11a6f8d

Commits on Sep 6, 2026

Commits on Sep 7, 2026

Now moving on to DKP-PCN CUDA
Ra4ster
committed
fcfb71b
Now moved DKP-PCN to CUDA. Next up, training on ImageNet!
Ra4ster
committed
2d56025

Commits on Sep 11, 2026

Added GELU.
Ra4ster
committed
582cf57
Added CUDA GELU
Ra4ster
committed
c31121d
Fixed switch-break
Ra4ster
committed
8fa31c5

Commits on Sep 12, 2026

Fixed a Gelu bug and updated Relu! The new results are superior to the standard library.
Ra4ster
committed
937246e

Commits on Sep 13, 2026

Stop tracking generated cache/log files
RA4-JACK
committed
144f227
Removed pycaches.
RA4-JACK
committed
af7a512
Fixed some typing issues.
RA4-JACK
committed
66b8dce
Removed venv fragments.
RA4-JACK
committed
68f02c1
Fixed CMakePresets version.
Ra4ster
committed
063caae
Added test case.
Ra4ster
committed
8ff1063

Commits on Sep 14, 2026

Commits on Sep 15, 2026

Added convolutional backend structures. TODO: Work on CUDA equivalent.
Ra4ster
committed
1e07252
Added more Backend utils for Convolution.
Ra4ster
committed
900d012

Commits on Sep 16, 2026

Preparing to test CUDA Convolution.
Ra4ster
committed
c99944d
Added pybindings and verified CUDA.
Ra4ster
committed
f7e2219
Fixed SimplePCN
Ra4ster
committed
ab320bf
Fixed SimplePCN wrapper bug.
Ra4ster
committed
1889254