Comment by yaoke259
17 hours ago
Supporting Huawei Ascend 950 NPUs with open source inference code has huge implications in the global AI hardware market. Imagine if all smaller labs now can build on top of this stack and add increasing support for more compatible hardware , software and models. This could seriously challenge NVidias death grip. Although inference is still only a part of all AI workloads and open sourcing training on Huawei chips would be even bigger.
I thought that inference was now the majority of AI workload, not training.
in terms of raw token throughput and serving users it is, but in terms of R&D and long term lab capabilities, training is still the goose that lays the golden egg
too bad two golden eggs are required to put the goose in a position to lay a golden egg
1 reply →
Yeah, the first place my mind went was that this is an incredible on-boarding / marketing effort for the Huawei hardware.