Overclocking multiple seperate Crucial RAM kits

Associate
Joined
27 Jul 2012
Posts
1,705
I received my original RMA RAM back and kept the RAM I bought to replace it In the mean time
  • Crucial Ballistix Sport AT BLS16G4D32AEST 3200MHz CL16 E-die 16gb x 2
  • Crucial Ballistix RGB BL2K8G30C15U4BL 3000MHz (E-die I think?) 8gb x 2
  • Motherboard - B450 Tomahawk Max
Should the 2nd and 4th be my 16gbx2 kit and the 1st and 3rd be my 8gb x 2?
Will I be OC the 2 Dimm Pairs separately or lower the OC overall to allow for the mix match of speeds?

I am hoping that 3000mhz CL15 equates to 3200mhz CL16

any tips or guides would be really appreciated, running Ryzen 3600
 
Soldato
Joined
20 Oct 2008
Posts
12,096
They'll all need to run at settings the slowest sticks will handle.

Do you actually need more than 32GB? If not, running with all of the slots full probably limit what you can achieve for no benefit.
 
Associate
OP
Joined
27 Jul 2012
Posts
1,705
They'll all need to run at settings the slowest sticks will handle.

Do you actually need more than 32GB? If not, running with all of the slots full probably limit what you can achieve for no benefit.

No, but selling the 16gb is more hassle than its worth.
The slowest stick could easily be in the 32gb kit as well, so I mean it wouldn't be that much slower.
Do you have any good guides for this Mobo?
 
Soldato
Joined
20 Oct 2008
Posts
12,096
I was referring to the load on the CPU's memory controller.

For best results, you want two single ranked sticks. With your 32GB, you already have two double ranked sticks (I assume), and then you're adding another two single ranked sticks on top.

If you aren't going to be doing anything that'll use the addition 16GB you're just making life difficult for no good reason.
 
Associate
OP
Joined
27 Jul 2012
Posts
1,705
I was referring to the load on the CPU's memory controller.

For best results, you want two single ranked sticks. With your 32GB, you already have two double ranked sticks (I assume), and then you're adding another two single ranked sticks on top.

If you aren't going to be doing anything that'll use the addition 16GB you're just making life difficult for no good reason.

Do you know of any good ways to stress test the RAM or to see if the OC has improved any real-world usage scenarios? (Excl. Memtest 86)
 
Soldato
Joined
12 Feb 2014
Posts
2,826
Location
Somewhere Only We Know
Do you know of any good ways to stress test the RAM or to see if the OC has improved any real-world usage scenarios? (Excl. Memtest 86)

Ryzen DRAM calc has built in tests, I find the default ram test picks up on errors pretty quickly and if you pass that and are happy with your overclock, then just run the full memtest overnight.

I just sold my G.Skill ram and brought 2 kits of Crucial E-Die ram, 4 x 8gb (32gb) I had great results with overclocking them, they are currently running at 3733mhz CL16 1.45v, but just to have a mess about with them, I passed memtest with them running 4000mhz CL18 1.45v, the latency was terrible though as I had to run the fabric at 2:1
 
Associate
OP
Joined
27 Jul 2012
Posts
1,705
Ryzen DRAM calc has built in tests, I find the default ram test picks up on errors pretty quickly and if you pass that and are happy with your overclock, then just run the full memtest overnight.

I just sold my G.Skill ram and brought 2 kits of Crucial E-Die ram, 4 x 8gb (32gb) I had great results with overclocking them, they are currently running at 3733mhz CL16 1.45v, but just to have a mess about with them, I passed memtest with them running 4000mhz CL18 1.45v, the latency was terrible though as I had to run the fabric at 2:1

Yo how do you know if the latency is shot or the fabric is running on 2:1? How do you benchmark the perf.
 
Soldato
Joined
12 Feb 2014
Posts
2,826
Location
Somewhere Only We Know
Yo how do you know if the latency is shot or the fabric is running on 2:1? How do you benchmark the perf.

CPU-z will show you what your fabric is running at, it should be half of your effective ram speed, so if your running 3000mhz RAM, fabric should be 1500mhz, thats 1:1, if its running at half that then its 2:1

to test latency, run AIDA64 memory benchmark, or DRAM calc's Default memory test.

SJ5mXeF.jpg
 
Associate
OP
Joined
27 Jul 2012
Posts
1,705
CPU-z will show you what your fabric is running at, it should be half of your effective ram speed, so if your running 3000mhz RAM, fabric should be 1500mhz, thats 1:1, if its running at half that then its 2:1

to test latency, run AIDA64 memory benchmark, or DRAM calc's Default memory test.

OK great,
so I am running FCLK @ 1900 and 3800 CL18 - I read that the improvement in FCLK to 1900 far outweighs the minor memory latency improvement in going 3600/cl14 vs 3800/cl18 - thoughts?

Karhu's RAM Test is meant to be much better... wish I could find a free download lol

gwKRMiF.png
 
Back
Top Bottom