Tomeu Vizoso 0c0d62ba70 etnaviv/nn: Implement zero run length encoding of weights ... Check how much smaller can the weight+bias buffers be with different amount of bits to encode runs of zeroes and choose the smallest one. This reduces the bandwidth considerably, which is at present the bottleneck with useful models. On a Libre Computer Alta AML-A311D-CC, I see these improvements: MobileNetV1: 15.650ms -> 9.991ms SSDLite MobileDet: 56.149ms -> 32.692ms Acked-by: Christian Gmeiner <cgmeiner@igalia.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27513>		2024-04-23 10:55:24 +02:00
..
ci	etnaviv: flip the switch on MSAA support	2024-04-16 10:35:41 +00:00
common	etnaviv/nn: Implement zero run length encoding of weights	2024-04-23 10:55:24 +02:00
drm	etnaviv: drm: Drop NPU-related params	2024-04-23 05:39:57 +00:00
drm-shim	etnaviv: decouple drm from gallium	2023-08-03 19:38:56 +00:00
hw	etnaviv: Move hw header to common place	2024-04-13 12:43:52 +00:00
hwdb	etnaviv/nn: Implement zero run length encoding of weights	2024-04-23 10:55:24 +02:00
isa	etnaviv: Move swizzle related macros to scr/etnaviv	2024-03-25 10:27:50 +00:00
meson.build	etnaviv: hwdb: Add etna_query_feature_db(..)	2024-04-13 12:43:52 +00:00