search

neurosim / DNN_NeuroSim_V1.3

Benchmark framework of compute-in-memory based accelerators for deep neural network (inference engine focused)
61 stars 36 forks source link

the mapping method from FP32 weight to synaptic condutance #29

Closed JoyKwan closed 2 years ago

JoyKwan commented 2 years ago

Hello, neurosim, I've learned a lot from Neurosim. But I have some quick questions about the mapping method from FP32 weight to synaptic condutance, Firstly, why don't you use the more common differential method(A signed kernel weight is mapped to the differential conductance of a pair of memristors used in the paper Fully hardware-implemented memristor convolutional neural network(https://www.nature.com/articles/s41586-020-1942-4)? Secondly, The fig.18 in DNN_Neurosim_v1.3_manual.pdf is a little bit strange, why 0.8906 mapped to 15 while 0.9187(>0.8906) mapped to 14(<15)? Is there a formula for this mapping method?

Kind Regards,

Joy Kwan

neurosim commented 2 years ago

Hi Joy, thanks for your interest in NeuroSim. Yes, the differential mapping is common, but it requires twice devices. The problem about fig.18 is a typo I think. There is no reason to map 0.8906 to 15 while 0.9187 to 14. It's just a schematic and sorry that we didn't check very carefully.