Baal Netbeck
Fleet Admiral
- Registriert
- Apr. 2008
- Beiträge
- 11.936
I agree, that it is simplifying a very complex topic.
But 99% of this paper is useless for our discussion....and I did read the 28 pages(not too thourough so)
I don't care how much power it takes to perform each operation when using jedec 2666.
And yes the paper explains how to add this up to get a theoretical power consumption, but this will not answer, if a system with DDR4 or 5 is performing with a better efficiency.
The controller and the VRM are part of the system and need to be considered.
The voltage scaling is useless too!
Different chips... different VRM...different starting point.
But the question arised and I think I answered it with a good approximation.
The only way would be a real test....like buildzoid did by modifying the hardware for specific power measurements.
Or having "identical" Mainboards for ddr4/ddr5....and then comparing external power measurements.
But 99% of this paper is useless for our discussion....and I did read the 28 pages(not too thourough so)
I don't care how much power it takes to perform each operation when using jedec 2666.
And yes the paper explains how to add this up to get a theoretical power consumption, but this will not answer, if a system with DDR4 or 5 is performing with a better efficiency.
The controller and the VRM are part of the system and need to be considered.
The voltage scaling is useless too!
Different chips... different VRM...different starting point.
But the question arised and I think I answered it with a good approximation.
Ergänzung ()
The only way would be a real test....like buildzoid did by modifying the hardware for specific power measurements.
Or having "identical" Mainboards for ddr4/ddr5....and then comparing external power measurements.
Zuletzt bearbeitet:
