View Single Post
Old 01-01-2020, 02:19 AM
A Computer Guy A Computer Guy is offline
Registered User
A Computer Guy's PC Specs
Join Date: Jan 2019
Posts: 209
POST ID # = 1029346
A Computer Guy Reputation: 10

Originally Posted by BleeBleeBlaBla View Post
Here is a screen shot of someone elses system....same memory.....

Can the CPU make or break RAM Memory performance ?
First - Happy New Year!

Generally better CPU better ram performance potential. (grain of salt - this is an oversimplified statement)

( I don't know why but I felt like writing on this first day of the new year. Hopefully helpful and not too much below. )

What is your ram version? (should be printed on the back of your ram with your model number)

Your performance will be based on the combination of motherboard + cpu + ram among other factors such as UEFI/BIOS version, what channel (single/dual/quad), the frequency, and timings, and the quality of the memory controller in the CPU.

For example you can have a BIOS update that totally screws your ram performance potential - doesn't even have to be related to overclocking.

In that new screen shot of another persons system you see it says 2 of 4 slots are used so in that case it's likely running in dual channel mode where as your original screen shot says 1 of 4 and in that case only single channel mode was possible during that test. I suspect something is wrong with your installed ram. In dual channel mode you will most certainly have better performance so if you have two sticks of memory and your not running dual channel mode (install in slots A2 + B2) you want to fix that first. Ensure both ram sticks are detected in UEFI/BIOS. They should both be reporting their proper capacities in UEFI/BIOS. If not something is wrong - like not seated properly, maybe bad stick, broken board, incompatible issue, need bios, etc...troubleshooting needed.

Something I noticed on my ASRock boards is the DDR4 slots have one fixed end and one latch end for each ram slot instead of latches being at both ends. Not sure if that's a new fad, or DDR4 specific thing, but make sure both of your ram modules are properly seated on both sides.

In one situation I had a ram stick not properly seated (just by a bit) and didn't notice. Interestingly BIOS reported it correctly but it kept failing to boot windows. Once I checked it I was surprised it clicked into place and well everything was ok after that.

One thing to keep in mind when comparing userbenchmark results is some people use the system as a feedback tool to initially rank then improve their benchmark scores as a means of tuning their computer. Simply they may have overclocks or other adjustments in place that end up effecting a particular benchmark, or skewing the results, which may not be apparent when you view other peoples results. This can make comparisons difficult.

I don't have my userbenchmark stats for my CMK16GX4M2A2400C16 (ver 5.30) kits anymore but in my examples below you can see that even having a better CPU doesn't necessarily mean you will have better ram performance as the integrated memory controller (IMC) is going to play a big part in what you can do. Due to different IMC and BIOS my ram timing and compatibility had changed so much I couldn't get nearly the same results after I upgraded my CPU with the same RAM.

Ryzen 5 2600, Corsair CMW32GX4M2C3200C16 (ver 4.32, dual rank, 2 sticks, 32GB, 18-17-17-36-58-tRFC=512)
running at DDR4-3200 -> Bench = 116%

Ryzen 7 3800x, Corsair CMW32GX4M2C3200C16 (ver 4.32, dual rank, 2 sticks, 32GB, 18-17-17-36-58-tRFC=512)
running at DDR4-3200 -> Bench = 93.7%

Ryzen 7 3800x, Corsair CMW32GX4M2C3200C16 (ver 4.32, dual rank, 2 sticks, 32GB, 18-19-19-42-68-tRFC=560)
running at DDR4-3600 -> Bench = 104%

Ultimately it didn't matter anyway as the only way humanly possible in my use case to notice the difference was with synthetic benchmarks anyway.

For example consider this result in contrast.

Ryzen 7 3800x, Micron 18ASF2G72AZ-2G6D1 4x16GB (dual rank, 4 sticks, 64GB, 20-20-20-44-66-tRFC=480)
running at DDR4-3200 -> Bench = 91%

My point is don't worry so much about chasing other peoples numbers (unless that is what you want to do) but get the most realistic value out of the parts you have and can afford. First fix your dual channel problem. That may likely give you a boost you can feel when using your computer and maybe put your ram benchmark numbers in the average range for your RAM benchmark.

If you still feel your computer is slow for what you need it for - I think if you pop a Ryzen 5 2600 (it's pretty cost effective) in your board you'll get a nice bump in performance and user experience over an A10-9700 with an opportunity to OC your ram for additional performance if your comfortable with doing that kind of thing. Just be sure to follow ASRocks instructions on their website first regarding BIOS updates!

Also keep in mind what your workload will be. If your use cases for your computer doesn't utilize high memory bandwidth then not having high memory bandwidth won't matter that much compared to having a good enough CPU and GPU to make your computing experience enjoyable.
I seem to hang out here too much. I blame the RGB.
Reply With Quote