George Hotz 5472a14544 openpilot compile2 (#1977) ...

* start compile2

* tweak

* why are there two more kernels?

* minor cleanups

* don't break onnx tests

* add __metadata__ support to safetensors

* no early realize in onnx

* cleanups

* bugfix

* clean up image type, add optimize

* opt to match old

* try that

* opt work

* run compile2

* optimizer

* prt more

* prerealize

* imp

* NOLOCALS works

* no locals means no locals

* support fractional globals

* all locals welcome

* int that

* cleanups

* show gemv regression

* clean up diff

* use idx for the cond

* nolocals

---------

Co-authored-by: Comma Device <device@comma.ai>

2023-10-15 20:39:46 -07:00

..

.gitignore

updates from the chonker branch

2022-11-07 21:12:08 -08:00

amx.py

fixes (#1893)

2023-09-22 07:20:27 +08:00

cuda_matmul.py

FLOAT16 off works

2023-04-19 15:34:56 -07:00

gemm.c

Revert "update editorconfig, enforce via CI (#1343)" (#1380)

2023-07-31 10:35:50 -07:00

gemm.py

fixes (#1893)

2023-09-22 07:20:27 +08:00

gemv_845.py

openpilot compile2 (#1977)

2023-10-15 20:39:46 -07:00

hip_matmul.py

fast HIP gemm -> 100 TFLOPS (#1476)

2023-08-09 06:54:15 -07:00

metal_conv.py

move device to ops (#1646)

2023-08-23 08:30:17 -07:00

metal_matmul.py

good stuff from tensor cores branch (#1199)

2023-07-08 16:58:26 -07:00

metal_matvec.py

optimizer: add matvec optimizations (#1972)

2023-10-04 14:16:27 -07:00

simple_matmul.py

wmma: clean up to make WMMA arg order consistent (#2014)

2023-10-07 17:45:40 -07:00

tf_gemm.py

Add tensorflow GEMM benchmark script (#1000)

2023-06-18 10:57:45 -07:00

torch_gemm.py

faster RDNA assembly backend (#990)

2023-06-16 12:06:38 -07:00

tvm_gemm.py

fix tvm gemm example

2023-10-08 05:57:41 -07:00