Closed ldrame21 closed 3 months ago
reyery
commented
4 months ago Hi @ldrame21, thank you for reporting this issue.
Can I confirm if you installed it using the steps found in the docs here (https://city-energy-analyst.readthedocs.io/en/latest/installation/installation-on-macos.html#developer-version)?
ldrame21
commented
4 months ago Hi @reyery, yes I did. Looking at this issue on the wrnt github, it seems that they have maybe not released a version for arm64 processor yet: https://github.com/USEPA/WNTR/issues/331
I've tried to create a new separate environment with this wntr library and then import it in a python script, the same error occurred. However, when done with a Rosetta terminal the error disappears.
I therefore reinstalled everything (brew, micromamba, cea etc.) via a Rosetta terminal and it fixed the issue. Even brew needs to be reinstalled otherwise when doing brew install micromamba, an arm64 version of micromamba will be downloaded. If someone needs to do the same, be careful with the path settings for micromamba and brew as they are different installations than the one that are used to download arm64 software on your mac. I would recommend creating a separate .bash file specifically for this rosetta installation.
I however now have an unrelated error when running this thermal network simulation and maybe you can help me with that. It says that Ths_sys_sup_ahu_C is not in the DataFrame and I see that there's a FIXME comment in the code. Do you have any idea what could cause this error ?
Error
Job Output for 67 - thermal-network
City Energy Analyst version 3.37.1
Running `cea thermal-network` with the following parameters:
- general:scenario = /Users/louis/Desktop/EPFL/Thesis/Github/CEA/cea/AMS_OOST/gva
(default: {general:project}/{general:scenario-name})
- general:multiprocessing = True
(default: True)
- general:number-of-cpus-to-keep-free = 1
(default: 1)
- thermal-network:network-type = DH
(default: DH)
- thermal-network:network-model = simplified
(default: simplified)
- thermal-network:min-head-substation = 20.0
(default: 20.0)
- thermal-network:hw-friction-coefficient = 100
(default: 100)
- thermal-network:peak-load-velocity = 2.0
(default: 2.0)
- thermal-network:equivalent-length-factor = 0.2
(default: 0.2)
- thermal-network:peak-load-percentage = 100.0
(default: 100.0)
- thermal-network:network-names = []
(default: [])
- thermal-network:set-diameter = True
(default: True)
- thermal-network:load-max-edge-flowrate-from-previous-run = False
(default: False)
- thermal-network:start-t = 0
(default: 0)
- thermal-network:stop-t = 8760
(default: 8760)
- thermal-network:use-representative-week-per-month = False
(default: False)
- thermal-network:minimum-mass-flow-iteration-limit = 30
(default: 30)
- thermal-network:minimum-edge-mass-flow = 0.1
(default: 0.1)
- thermal-network:diameter-iteration-limit = 10
(default: 10)
- thermal-network:substation-cooling-systems = ['ahu', 'aru', 'scu']
(default: ['ahu', 'aru', 'scu'])
- thermal-network:substation-heating-systems = ['ahu', 'aru', 'shu', 'ww']
(default: ['ahu', 'aru', 'shu', 'ww'])
- thermal-network:temperature-control = VT
(default: VT)
- thermal-network:plant-supply-temperature = 80.0
(default: 80.0)
- thermal-network-optimization:use-representative-week-per-month = False
(default: False)
Traceback (most recent call last):
File "/Users/louis/Desktop/EPFL/Thesis/Github/CEA/CityEnergyAnalyst/cea/worker.py", line 146, in worker
run_job(config, job, server)
File "/Users/louis/Desktop/EPFL/Thesis/Github/CEA/CityEnergyAnalyst/cea/worker.py", line 108, in run_job
script(config=config, **parameters)
File "/Users/louis/Desktop/EPFL/Thesis/Github/CEA/CityEnergyAnalyst/cea/api.py", line 60, in __call__
self._runner.__call__(*args, **kwargs)
File "/Users/louis/Desktop/EPFL/Thesis/Github/CEA/CityEnergyAnalyst/cea/api.py", line 38, in script_runner
script_module.main(config)
File "/Users/louis/Desktop/EPFL/Thesis/Github/CEA/CityEnergyAnalyst/cea/technologies/thermal_network/thermal_network.py", line 3443, in main
thermal_network_simplified(locator, config, network_name)
File "/Users/louis/Desktop/EPFL/Thesis/Github/CEA/CityEnergyAnalyst/cea/technologies/thermal_network/simplified_thermal_network.py", line 195, in thermal_network_simplified
substation.substation_main_heating(locator, total_demand, building_names, DHN_barcode=DHN_barcode)
File "/Users/louis/Desktop/EPFL/Thesis/Github/CEA/CityEnergyAnalyst/cea/technologies/substation.py", line 64, in substation_main_heating
Ths_supply_C, Ths_return_C = calc_temp_hex_building_side_heating(substation_demand, heating_configuration)
File "/Users/louis/Desktop/EPFL/Thesis/Github/CEA/CityEnergyAnalyst/cea/technologies/substation.py", line 85, in calc_temp_hex_building_side_heating
Ths_return, Ths_supply = calc_compound_Ths(building_demand_df, heating_configuration)
File "/Users/louis/Desktop/EPFL/Thesis/Github/CEA/CityEnergyAnalyst/cea/technologies/substation.py", line 380, in calc_compound_Ths
Ths_ahu_supply = building_demand_df.Ths_sys_sup_ahu_C.values # FIXME: it's the best to have nan when there is no demand
File "/Users/louis/micromamba/envs/cea/lib/python3.8/site-packages/pandas/core/generic.py", line 5989, in __getattr__
return object.__getattribute__(self, name)
AttributeError: 'DataFrame' object has no attribute 'Ths_sys_sup_ahu_C'I found the error, here's the solution:
When the Demand Forecasting/Building Energy Demand job is done monthly, Temparature values are not stored in the output dataframe. Therefore, if one wants to do Thermal Network or Supply System optimization, demand must be computed hourly.
reyery
commented
4 months ago Looking at this issue on the wrnt github, it seems that they have maybe not released a version for arm64 processor yet: https://github.com/USEPA/WNTR/issues/331
We actually built and included the arm64 binaries required for wntr to work, within the source code of CEA (https://github.com/architecture-building-systems/CityEnergyAnalyst/tree/master/cea/lib). It is automatically loaded when it detects that the machine is using the M series processors for Mac. I personally use a M1 Macbook to develop CEA as well. I tried reinstalling it again on my machine and was still able to run thermal-network without any issues.
CEA uses this logic to decide when to apply this "fix"
if sys.platform == "darwin" and platform.processor() == "arm":My guess is that maybe there was some PATH variable issue or something similar, where the python interpreter that you were using was the x86 version. Would it be possible for you to check by running the following command in the activated cea environment, to see if it is indeed "arm" or something similar (mine shows "arm"). Maybe for yours it gave "arm64", and that could be something we can change.
python -c "import platform; print(platform.processor())"The newer wntr versions do include binaries for arm64 in their source code. But after testing, the results were wildly different than the version we are using (0.2.3) for CEA. We do not have the resources to investigate the differences for now so this separate loading of binaries is a workaround.
ldrame21
commented
4 months ago Thank you for your answer @reyery. Indeed, I just checked even in my base environment, my python command is using x86 processor instead of arm. It's because my Homebrew ARM64 path variable was overridden by an Anaconda python executable using x86... As the thing is working now with Rosetta, I will continue like that. I will try again when I have time and update you on this post.
Best, Louis
ldrame21
commented
4 months ago Update The problem is fixed but it doesn't work on my pycharm terminal as it is not using the right processor for some reason:
` ~/Desktop/EPFL/Thesis/Github/CEA/CityEnergyAnalyst master !3 ?9 ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- 02:45:56 PM
python -c "import sys; print(sys.platform); import platform; print(platform.processor())" darwin i386 ` while it is when I'm using my normal terminal. It must be a problem with the pycharm interpreter but I will just use my normal terminal.
Thanks for your help!
reyery
commented
3 months ago Closing since it seems that the issue has been resolved. Let me know if there are any more issue.
Describe the bug I encountered an architecture incompatibility error while trying to run the Thermal Network Simulation (Part II) as well as the optimizations modules on an Apple Silicon (M1) Mac. The error indicates that the _evaluator.cpython-38-darwin.so binary is compiled for x86_64, but my system requires arm64 or arm64e. This issue seems to have been discussed between @jimenofonseca and @reyery here: #3120. I however do not see a clear resolution. To Reproduce Steps to reproduce the behavior:
Logs
Hardware (please complete the following information):
CEA version [3.37.0]
Thank you for your time.
Best, Louis