• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Share your AIDA 64 cache and memory benchmark here

My sub par DDR5.. should have some new sticks shortly.. nice ones :)

Screenshot 2025-07-14 185147.png
 
I think 1.7vdd and 1.64vddq but these were overshot on purpose. With gdm on, 6400mt, can do cl26 1.47v. This is the pinnacle of ddr5, just waiting on a proper bios to take on 8000MT CL30.
Thats quite a diffrence in voltage reqs for gdm off. Reaching stable cl26 was quite easy in my case. So far i always went with GDMOFF (along with lowest SCLs) but lately i see more opionions that ultimately gdmoff /scls is not that big latency reducer as was said erly on. Since I target best compromise between performance with low idle power and cooling requiments , I should check if hat voltage overhead might be better spent for better speed/cl instead,
 
Thats quite a diffrence in voltage reqs for gdm off. Reaching stable cl26 was quite easy in my case. So far i always went with GDMOFF (along with lowest SCLs) but lately i see more opionions that ultimately gdmoff /scls is not that big latency reducer as was said erly on. Since I target best compromise between performance with low idle power and cooling requiments , I should check if hat voltage overhead might be better spent for better speed/cl instead,

Well 1.47 is for cl26, not cl24. Cl24 is around 1.59 to 1.6x. Gdm off bumps that requirement by maybe 0.4-0.5, so maybe 1.66vdd would be enough.

And Aida 64 latency and clam with large pages are correlated but seem to be affected by different timings/settings. Aida is very "superficial". So you should retest your understanding and stick to clam as the final determiner of settings.

 
Last edited:
Back
Top