Deciding Plug in and accuracy

[tanvisharma]

Hi,

Could you give an example on how to change the plug in to say CACTI instead of dummy for energy estimation? I tried to simply add plug_in and minimum_accuracy attributes to my memory level, but it runs into an error. Amusingly, in the log files I could see that it estimated the energy using CACTI for read, write and idle operations, but failed the fourth time for update(). I ran it again now and have copied the relevant log below.

2023-10-16 00:20:09 INFO        CACTI estimated 40960p with accuracy 95%. 
2023-10-16 00:20:09 INFO        
2023-10-16 00:20:09 INFO        ENERGY ESTIMATION for DRAM(type=LPDDR4, width=5120, datawidth=8, cluster_size=1, read_bandwidth=1300, write_bandwidth=1300, plug_in=CACTI, minimum_accuracy=90, technology=-1, n_banks=2, reduction_supported=True, multiple_buffering=1, min_utilization=0, allow_overbooking=False, meshX=1, meshY=1).write()
2023-10-16 00:20:09 INFO        CACTI estimated 40960p with accuracy 95%. 
2023-10-16 00:20:09 INFO        
2023-10-16 00:20:09 INFO        ENERGY ESTIMATION for DRAM(type=LPDDR4, width=5120, datawidth=8, cluster_size=1, read_bandwidth=1300, write_bandwidth=1300, plug_in=CACTI, minimum_accuracy=90, technology=-1, n_banks=2, reduction_supported=True, multiple_buffering=1, min_utilization=0, allow_overbooking=False, meshX=1, meshY=1).idle()
2023-10-16 00:20:09 INFO        CACTI estimated 0p with accuracy 95%. 
2023-10-16 00:20:09 INFO        
2023-10-16 00:20:09 INFO        ENERGY ESTIMATION for DRAM(type=LPDDR4, width=5120, datawidth=8, cluster_size=1, read_bandwidth=1300, write_bandwidth=1300, plug_in=CACTI, minimum_accuracy=90, technology=-1, n_banks=2, reduction_supported=True, multiple_buffering=1, min_utilization=0, allow_overbooking=False, meshX=1, meshY=1).update()
2023-10-16 00:20:09 ERROR       ================= An error has caused Accelergy to crash. Error below =================
2023-10-16 00:20:09 ERROR       Can not find an energy estimator for DRAM(type=LPDDR4, width=5120, datawidth=8, cluster_size=1, read_bandwidth=1300, write_bandwidth=1300, plug_in=CACTI, minimum_accuracy=90, technology=-1, n_banks=2, reduction_supported=True, multiple_buffering=1, min_utilization=0, allow_overbooking=False, meshX=1, meshY=1).update()
2023-10-16 00:20:09 ERROR       Logs for plug-ins that could estimate query:
2023-10-16 00:20:09 ERROR       | CACTI with accuracy 0% estimating accuracy:
2023-10-16 00:20:09 ERROR       |  | Accuracy is 0%. Not supported.
2023-10-16 00:20:09 ERROR       | Library with accuracy 0% estimating accuracy:
2023-10-16 00:20:09 ERROR       |  | Found 0 entries for dram.update.
2023-10-16 00:20:09 ERROR       |  | ValueError: Could not find energy for dram
2023-10-16 00:20:09 ERROR       | Aladdin_table with accuracy 0% estimating accuracy:
2023-10-16 00:20:09 ERROR       |  | Accuracy is 0%. Not supported.
2023-10-16 00:20:09 ERROR       | table-based-plug-ins with accuracy 0% estimating accuracy:
2023-10-16 00:20:09 ERROR       |  | Accuracy is 0%. Not supported.
2023-10-16 00:20:09 ERROR       | dummy_table with accuracy 0% estimating accuracy:
2023-10-16 00:20:09 ERROR       |  | Plug-in dummy_table was not selected for query.
2023-10-16 00:20:09 ERROR       Why plug-ins did not estimate:
2023-10-16 00:20:09 ERROR       | CACTI with accuracy 0% estimating accuracy: Accuracy is 0%. Not supported.
2023-10-16 00:20:09 ERROR       | Library with accuracy 0% estimating accuracy: ValueError: Could not find energy for dram
2023-10-16 00:20:09 ERROR       | Aladdin_table with accuracy 0% estimating accuracy: Accuracy is 0%. Not supported.
2023-10-16 00:20:09 ERROR       | table-based-plug-ins with accuracy 0% estimating accuracy: Accuracy is 0%. Not supported.
2023-10-16 00:20:09 ERROR       | dummy_table with accuracy 0% estimating accuracy: Plug-in dummy_table was not selected for qu```

Thanks!

[tanner-andrulis]

Fixed in the latest CACTI plug-in version. Can you update to the latest version and see if the bug persists? https://github.com/Accelergy-Project/accelergy-cacti-plug-in