[6nf]

Faster processing alone won't make training 1000x faster, the bottleneck is more on the memory size / bandwidth side

[anigbrowl]

I feel people are overlooking the OP's mention of parallel improvements in storage and speed of access. While there are physical limits to this, I feel like capabilities will continue to expand not so much in terms of pure speed as in better automation of parallelization and resource allocation.

[rowanG077]

The thread is about the whole stack 1000x'ing. Not just processing speed.