George HotzandGitHub 9cc2577a08 use hip events (#3157) ...

* use hip events

* cleanup

2024-01-17 10:39:57 -08:00

..

codegen

wmma: enable METAL half tensor cores and clean up cstyle (#3095)

2024-01-12 16:25:28 -05:00

features

sharding for llama (#3151)

2024-01-16 19:28:00 -08:00

nn

sharding for llama (#3151)

2024-01-16 19:28:00 -08:00

renderer

wmma: enable METAL half tensor cores and clean up cstyle (#3095)

2024-01-12 16:25:28 -05:00

runtime

use hip events (#3157)

2024-01-17 10:39:57 -08:00

shape

support for non-uniform sharding (#3154)

2024-01-16 20:33:32 -05:00

__init__.py

move dtypes to dtype.py (#2964)

2024-01-01 14:58:48 -08:00

device.py

hotfix: copy size is in bytes

2024-01-17 16:44:15 +00:00

dtype.py

simplify dtype (#3137)

2024-01-15 16:27:43 -08:00

graph.py

graph cleanup (#3155)

2024-01-16 20:57:31 -08:00

helpers.py

fast resnet eval (#3135)

2024-01-15 14:15:18 -08:00

jit.py

multitensor jit (#3149)

2024-01-16 09:09:15 -08:00

lazy.py

move fromcpu out of lazy.py (#3122)

2024-01-14 18:21:08 -08:00

mlops.py

Fix backward fn for < and == (#3037)

2024-01-14 20:39:52 -08:00

ops.py

remove unused reciprocal (#3053)

2024-01-09 08:59:04 -08:00

realize.py

mem_estimate is always int, not symbolic (#3083)

2024-01-10 23:39:51 -05:00

tensor.py

sharding for llama (#3151)

2024-01-16 19:28:00 -08:00