Adc changes, csv utility and other changes #30
Conversation
| @@ -381,21 +392,21 @@ | |||
| xbar_rd_pow_dyn = xbar_rd_pow | |||
| xbar_wr_pow_dyn = xbar_wr_pow | |||
| dac_pow_dyn = dac_pow_dyn_dict [str(cfg.dac_res)] | |||
| adc_pow_dyn = adc_pow_dyn_dict [str(cfg.adc_res)] | |||
| adc_pow_dyn = 0 #adc_pow_dyn_dict [str(cfg.adc_res)] | |||
There was a problem hiding this comment.
Why make adc_pow_dyn 0 here, when the param.adc_pow_dyn is not used in hw_stats.py (hw_stats.py now is directly using entries from param.adc_pow_dyn_dict) ?
There was a problem hiding this comment.
Yes, it will access the adc_pow_dyn_dict dictionary based on adc resolution.
param.adc_pow_dyn_dict[str(cfg.adc_res_new[str('matrix_adc_'+str(k))])]
Are you ok with that?
| adc_res_new = { | ||
| 'matrix_adc_0' : 8, | ||
| 'matrix_adc_1' : 4, | ||
| 'matrix_adc_1' : 8, | ||
| 'matrix_adc_2' : 8, | ||
| 'matrix_adc_3' : 4 | ||
| 'matrix_adc_3' : 8 |
There was a problem hiding this comment.
For lines 49-53, I understand this would only work for configuration with num_matrix=2. Can you either put this within a new flag (maybe if hetero-adc), so that the adc_res_new dictionary is only created when flag==true, otherwise it just creates an empty adc_res_new dictionary. This will ensure that if num_matrix>2, ima.py::init will just fall back to using old adc_res for all ADCs.
There was a problem hiding this comment.
As I changed the way that other dictionaries will be access because it will depend on the value of adc_res_new (we can change that for a better name) I think what we could do is is create a new flag as you suggested, hetero-adc, but uses the old adc_res as default value for adc_res_new to avoid change the other dictionaries again. Something like:
hetero-adc = false
if not(hetero-adc):
adc_res_new = {
'matrix_adc_0' : adc_res,
'matrix_adc_1' : adc_res,
'matrix_adc_2' : adc_res,
'matrix_adc_3' : adc_res
}
else:
adc_res_new = {
'matrix_adc_0' : 4,
'matrix_adc_1' : 6
'matrix_adc_2' : 4
'matrix_adc_3' : 6
}
What do you think?
| for k in range (cfg.num_adc): | ||
| hw_comp_energy['adc_'+str(k)] = param.adc_pow_dyn_dict[str(cfg.adc_res_new[str('matrix_adc_'+str(k))])] |
There was a problem hiding this comment.
Can str(cfg.adc_res_new[str('matrix_adc_'+str(k))]) also be put within hetero-adc flag, so that in cases where num_matrix>2, this can fall back to cfg.adc_res ?
There was a problem hiding this comment.
Please look my suggestion above
| for k in range (cfg.num_adc): | ||
| area += param.adc_area_dict[str( cfg.adc_res_new[str('matrix_adc_'+str(k))] )] |
There was a problem hiding this comment.
Can str(cfg.adc_res_new[str('matrix_adc_'+str(k))]) also be put within hetero-adc flag, so that in cases where num_matrix>2, this can fall back to cfg.adc_res ?
There was a problem hiding this comment.
Please look my suggestion above
| for k in range (cfg.num_adc): | ||
| leak_pow += param.adc_pow_leak_dict[str( cfg.adc_res_new[str('matrix_adc_'+str(k))] )] |
There was a problem hiding this comment.
Please look my suggestion above
| for k in range (cfg.num_adc): | ||
| dyn_pow += param.adc_pow_dyn_dict[str( cfg.adc_res_new[str('matrix_adc_'+str(k))] )] |
There was a problem hiding this comment.
Please look my suggestion above
This branch contains: