cloud11665

vo_tct: rewrite using the newer vo api, generalize and optimize code

2

Benchmarked: ###### 1080p 60fps 41s video, played on a 384x102 terminal (alacritty) | frames dropped | stock | term256 | ALGO_PLAIN | ALGO_PLAIN + term256 | | -------------- | -----...

Maybe we should just write the ansi escape codes by hand?

fix inf/nan codegen

3

We were appending f to NAN and INFINITY constatns, and now we don't! Constants taken from https://www.gnu.org/software/libc/manual/html_node/Infinity-and-NaN.html

Cuda on cpu

34

tested singular ptx kernels. Setting up the env for this is a pain

cuda: fix fp16, uint8, int64, half4 codegen

7

`test/test_dtype.py` -> tests failed went down to 3 from 20. only ones left are f16, but I haven't looked into them yet. PS: this change also works on CUDACPU

opencl test_ceil / floor failing

4

Failing with OPT=3 ``` PYOPENCL_COMPILER_OUTPUT=1 DEBUG=3 CPU=0 OPT=3 python -m pytest -s -v -n=auto test/test_ops.py::TestOps::test_ceil -n=1 test/test_ops.py:39: in helper_test_op compare("forward pass", ret.numpy(), out.detach().numpy(), atol=atol, rtol=rtol) _ _ _ _ _...

cuda ci on cpu

1

Opening a new pr, as I couldn't reopen the old one, sorry. - pycuda.driver.Context - mocked - cuda - mocked - no mentions of CUDACPU in the implementation code -...

(WIP) ptx in ci

2

All testing was done on a 4070, cpu tests will come after all pass on device. test_ops: - 3 failed, 161 passed, 11 skipped, 1 warning in 32.7 fails on...