Michael Bussmann
Michael Bussmann
To me this looks like a case of 'highest ionization rate possible' achieved. I am not sure PIConGPU can (and should) get in such a state. But if you assume...
I may remember an early issue on AMD which looked like only 50% of GPU memory could be allocated. Was it this one you remembered @sbastrakov ?
Surely, memory access and allocation need better care in Alpaka.
Great idea!
I'm sorry, but what is HEP? What is dimuon? You should strongly reconsider your strategy on issue documentation. In the long run, your strategy of minimal doc + assumed self...
Can you put these numbers in perspective?
Of course, this goes in the same direction as ranges for arrays, but you know this already ;-)
That's a bit tight. We're all at GSI and discuss this later and then come back to you.
Your last comment is extremely relevant. HPC is often bleeding edge and certain hardware 2-3 years behind in supporting standards, but these years make all the difference in real world...
@bernhardmgruber any ideas on this?