nimlgen bd580d8ea4 hcq: use mmio interface in nv (#9986) ... * hcq: start mmio interface * allow double cast * revert * faster? * simpler, not needed more now * dd * types * fix		2025-04-22 21:58:12 +03:00
..
codegen	faster block reorder (#9990)	2025-04-22 19:18:57 +01:00
engine	refactor schedule linearize to defaultdict [pr] (#9984)	2025-04-23 00:00:23 +08:00
frontend	add onnx frontend stub [pr] (#9558)	2025-03-24 12:24:34 +08:00
nn	nn.state docs cleanup (#8332)	2025-03-18 17:16:40 -04:00
renderer	don't use ops name, just keep sink (#9922)	2025-04-18 08:59:18 +01:00
runtime	hcq: use mmio interface in nv (#9986)	2025-04-22 21:58:12 +03:00
shape	cache folded upcast [pr] (#9733)	2025-04-04 11:23:19 +08:00
viz	don't use ops name, just keep sink (#9922)	2025-04-18 08:59:18 +01:00
__init__.py	UOp.variable (#7010)	2024-10-12 18:20:44 +08:00
device.py	make DISABLE_COMPILER_CACHE a ContextVar [pr] (#9983)	2025-04-22 10:32:54 -04:00
dtype.py	FP8s part 1 (#9887)	2025-04-15 11:20:02 -04:00
gradient.py	fix kernelize usage with pm_gradient (#9953)	2025-04-22 17:26:05 +08:00
helpers.py	make DISABLE_COMPILER_CACHE a ContextVar [pr] (#9983)	2025-04-22 10:32:54 -04:00
ops.py	non recursive toposort is 2x+ faster (#9979)	2025-04-22 13:59:38 +01:00
py.typed	add a single py.typed (#6083)	2024-08-14 17:31:46 -07:00
spec.py	microbenchmarks + micro speed ups (#9972)	2025-04-22 11:30:46 +01:00
tensor.py	move metadata set to tensor [pr] (#9976)	2025-04-22 12:30:35 +01:00
upat.py	don't use ops name, just keep sink (#9922)	2025-04-18 08:59:18 +01:00