No description
modeco80
e0e78a92e9
a frame is now [N tiles] then the whole framebuffer not great for bandwidth, but it's tolerable enough and the performance of it is vastly better. at some point we can use ivshmem to sidestep that and just make it a cpu->cpu copy I guess |
||
---|---|---|
agent | ||
src | ||
.clang-format | ||
.gitignore | ||
Cargo.lock | ||
Cargo.toml |