Function num_sms_cached

src/fused_mm_sampling/core.py:211–212 · view source on GitHub ↗

(device_index: int)

Source from the content-addressed store, hash-verified

209
210	@lru_cache(maxsize=1)
211	def num_sms_cached(device_index: int) -> int:
212	return torch.cuda.get_device_properties(device_index).multi_processor_count
213
214
215	MIN_BLOCK_SIZE_V = 128

fused_mm_sample_tritonFunction · 0.85

no outgoing calls

no test coverage detected