"Hey comma" to bookmark segments #35732
Conversation
|
things im a little iffy on
|
|
I trolled and I thought the bad cores were the high index ones... quantized the embedding model to fp16, tinygrad LLVM has worse performance than fp32 on the bad core, so switched backend to CPU=1 and gets 42% CPU usage on bad cores (however on the good core, LLVM seems to do better than CPU with the quantized model???) there also seems to be possibly be some performance gap depending on if the net was compiled on the same type of core it is run on |
|
17-18% little core CPU usage |
|
IMAGE=2 compiles the model to be ~ same time as BEAM=1 so this the previous cpu/gpu usage is the same, and compile times are normal
|
cereal/log.capnp
Outdated
| wakewordExecutionTime @3 :Float32; | ||
| totalExecutionTime @4 :Float32; | ||
|
|
||
| wakewordProb @5 :Float32; |
There was a problem hiding this comment.
float16 isn't natively supported in capnp
There was a problem hiding this comment.
how much do we care about making it small? I could scale and cast to UInt16 and probably UInt8 for wakewordProb
selfdrive/feedback/feedbackd.py
Outdated
| if sm.updated['bookmarkButton'] and (current_time - last_user_flag_time) > debounce_period: | ||
| cloudlog.info("Bookmark button pressed!") | ||
| should_send_flag = True |
There was a problem hiding this comment.
I wouldn't read this here. the backend should just handle both
There was a problem hiding this comment.
Not sure how you wanted to handle the bookmark button, but if we wanted to have the bookmark button also record audio feedback in the future, I think it would still make sense to handle both in feedbackd.
Saying "Hey comma" will let you bookmark a segment, in the same way that clicking the bookmark button in the sidebar currently does.