Topping up RAM, does the MHz affects?

[AlchemistJ19]

HI all pros,
want to check whether for the RAM, if i want to top up some RAM.
Does it matter if the MHz is different?
my initial was 3200MHz, now in amazon the 3600MHz is cheaper than 3200.

 

[Koenig168]

Doesn't matter. All sticks will run at the lower frequency.

 

[Entry Level PC]

HI all pros,
want to check whether for the RAM, if i want to top up some RAM.
Does it matter if the MHz is different?
my initial was 3200MHz, now in amazon the 3600MHz is cheaper than 3200.

the CL matters as well
best is you go youtube watch benchmark.
even if same CL you may not get 5% gains

 

[AlchemistJ19]

My current ram that i wanted to top up:​

TEAMGROUP T-Force Delta RGB DDR4 16GB (2x8GB) 3200MHz (PC4-25600) CL16 Desktop Memory Module ram TF4D416G3200HC16CDC01 - White​

Price:

S$146.00

kns this one cheaper

TEAMGROUP T-Force Night Hawk RGB 3600MHz 16GB Kit (2x8GB) CL18 DDR4 SDRAM (PC4-28800) Desktop Memory Module Ram TF2D416G3600HC18JDC01 - White​

Price:

S$112.22

 

[AlchemistJ19]

one is CL16 another is CL18???

 

[wwenze]

Buy the cheaper one and run it at 3200CL16

 

[NightRaven49]

one is CL16 another is CL18???

3200cl16 and 3600cl18 is the same, 10ns latency (at least for primary timings).

 

[Mach3.2]

Make sure the ICs on all sticks are the same, if not you'll probably run into problems.

I tried mixing Samsung B dies and Micron Rev. B ICs and it's a **** show.

 

[wwenze]

Yea, haven't upgraded computer for half a decade and I don't know how the newer motherboards do their auto ram timings now. When I put timing auto and increased clockspeed all my old motherboards would loosen the timings proportionally yet the B550 used the same default timings as according to XMP. Never a smart decision.

Samsung and Micron have very different subtiming limits so who knows what the motherboard will set.

 

[Yongkit]

This will interest many for mixing RAM setup

 

[Mach3.2]

Yea, haven't upgraded computer for half a decade and I don't know how the newer motherboards do their auto ram timings now. When I put timing auto and increased clockspeed all my old motherboards would loosen the timings proportionally yet the B550 used the same default timings as according to XMP. Never a smart decision.

Samsung and Micron have very different subtiming limits so who knows what the motherboard will set.

I'm running custom timings, didn't use XMP.


Point being, mixing ICs can give you unexpected results and will probably not work unless you run JEDEC speed and timing.

 

[Yongkit]

Based on what I see the price in Amazon offer right now for this set of night hawk white series kit RAM I would sell the 16GB RAM and buy the 32gb directly instead of mixing the ram kit.

 

[Koenig168]

Night Hawk can be a bit hit or miss. I'm currently running a pair of DDR4-3200 CL16 Night Hawks at DDR4-3600 CL16 1.36V. They can OC to DDR4-3733 CL 17 at the same 1.36V. However, some reviewers did not do so well on the OC.

I agree with Yongkit. Might as well go for the 32GB kit rather than mixing sticks.

 

[Yongkit]

Yea, haven't upgraded computer for half a decade and I don't know how the newer motherboards do their auto ram timings now. When I put timing auto and increased clockspeed all my old motherboards would loosen the timings proportionally yet the B550 used the same default timings as according to XMP. Never a smart decision.

Samsung and Micron have very different subtiming limits so who knows what the motherboard will set.

Based on my testing if I only increased the clockspeed of my RAM (advertised 3200MHz CL16 memory kit) the only changes show in CPU-Z was the Bank Cycle Time (tRC) as following:

(tRC) 2666MHz = 61 Clocks
(tRC) 3200MHz = 74 Clocks
(tRC) 3466MHz = 80 Clocks
(tRC) 3600MHz = 83 Clocks



I also done TimeSpy custom CPU score as following setting & results:

Windows 11
3800x (8 core 16 threads) CPU Frequency all clock locked = 4,175MHz
CPU Voltage = 1175mV (undervolted)

Memory XMP clock @ CL16 = CPU score (% round up):

2666MHz = 9,672 (stable)
3200MHz = 10,169 (5% improvement over stock) (stable)
3466MHz = 10,315 (7% improvement over stock) (stable)
3600MHz = 10,536 (9% improvement over stock) (not stable for overall system)

my motherboard bios can only support my memory kit with stability @ 3466MHz CL16 XMP OC for both gaming & application so basically I am already exceed the advertised speed of 3200MHz CL16.

I can only rely on XMP really cannot compete with all Guru who got their RAM manually OC

3600MHz for my memory kits have no issue running TimeSpy benchmarks, but very unstable for games and crash easily once in gaming environment.