Kingston Fury Beast 2 x 32GB A-Die Timings

Soldato
Joined
27 Jan 2003
Posts
5,962
Location
Chesterfield, UK
I just picked up a pack of 2x32GB Fury Beast 6000 C40 and luckily I got A-Die. Just logging some of my testing hoping it might be use to someone.

CPU 7950X3D CCD0 CO -34, CCD1 -21, PPT 110 Watts

Label on the sticks
KF560C40-40BBK2-64
EOMH1682374
 
Last edited:
After many hours of testing with TestMem5 I have settled on these settings. I wanted to a low SOC and didn't want to actively cool the ram, its a silent system. Running PPT of 110 Watts which keeps the chip below 80'c under load.

Cinebench Scores
R23 34,473
R24 2058

Timings are based on a modified BuildZoid.

ZenTimings

Just to confirm, Samsung B-Die at the same speed but at C34 creates a slightly better score with tighter timings. Obviously A-Die will go faster which would mean better scores, if I was prepared to increase the voltage and heat, which I don't. For people who want to fit and forget, Samsung B-Die is better but not very future proof. Newer chips will run faster on the memory and this is where A-Die will still be useful.
 
Last edited:
Just the sort of info I've been looking for, couple of questions:

How do you know if you have Hynix A Die or something else?
Looking at the screenshot, you've kept the memory at 6000 but gone from CL40 to CL32?
HWinfo Memory Timings gives you the die info. Thaiphoon is another program.

Yes, the default profile is CL40. Most memory can run at other settings with more voltage or at the same voltage. A-Die can be tweaked on the timings and speed, M-Die is tighter but a bit more limited on the speed. Samsung B-Die doesn't go very high but can give you very tight timings.

A-Die can be found in some cheaper memory if your lucky but from what I have read, chips are being binned for the 7000 plus sticks. It depends if you are into tweaking, if not just get memory at the required speed and latency. Lower latency CL32/34 is better.
 
Last edited:
Thanks for that info.

Don't suppose you have a before and after of those cinebench scores at all?

Interested to see how much difference it makes over EXPO settings.
These sticks are XMP for Intel, so I can’t compare. My Corsair Vengence are EXPO and to be honest it will be only be a lite tweak of the settings. Any settings will be based on something that will work with all memory chips.

If you go for the X3D version tweaking timings nets you very little. These CPU’s need speed, but your very limited how far the memory can be pushed. 6400 max if your lucky. Intel platforms really benefit from tweaking which is why A-Die is so great for these platforms.
 
Last edited:
Out of interest with these timings I upped the Memory speed and FLK to see how R23 improved over 6000. I limited the PPT to 110 Watts

6000/2000 Scored 34,473
6200/2033 Scored 34,523
6400/2100 Scored 34,478

With the speed set back to 6000/2000 but upping the PPT to 120 watts gave a score of 35,508.

To compare with DDR5 tweaked Samsung B-Die at this speed and power gave a score of 35,988.
 
Last edited:
Back
Top Bottom