What You Need to Know about RAM Speeds – Does RAM Speed Matter?

A more comprehensive look at RAM speeds and how they affect gaming, workstation, and every day

Shaun Conroy

[lmt-post-modified-info]

WePC’s mission is to be the most trusted site in tech. Our editorial content is 100% independent and we put every product we review through a rigorous testing process before telling you exactly what we think. We won’t recommend anything we wouldn’t use ourselves. Read more

K-Black

Nov 23, 2021 at 6:16 am

Well explained.
Reply
puzzled

Jan 16, 2020 at 8:40 pm

I already tried to post this but it didn’t work. If that’s regular time spy your scores are way low. Like 50% of what I get with the same CPU, GPU and 3600 CL16 ram. Both the GPU and CPU are capable of scoring 10k in time spy and that’s thousands away from record high for that hardware.
Reply
puzzled

Jan 16, 2020 at 8:36 pm

What time spy are you running to test this? Is it time spy extreme or something? The reason I ask is because I have a ryzen 7 3800x and a 5700xt with 2 x 8 3600 CL16 ram and I’m scoring 9800 without even trying or just over 10000 with a little effort. Both my CPU and GPU are capable of scoring 10k on their own and I’m definitely not alone there. Just comparing the results online I see people doing way better on the same hardware too. I have a better motherboard but it shouldn’t make double the performance. Just saying it was a great article but if that’s all you’re getting out of that hardware then you should probably be putting in an RMA. You can look on the 3DMark sore and search results by hardware to check it out yourself.
Reply
Peter

Nov 19, 2019 at 3:21 am

I think RAM speed does matter more than the article states. Your testing method could be improved a bit. The fact of the matter is that we need a bottleneck hierarchy to be stated. You are correct that the GPU is the one doing most of the work in these graphically intensive games, and this is the point I want to bring up. When the GPU is able to run the settings you are trying to play with without breaking a sweat anymore, then it can go ahead and use the rest of the GPU resources to try and pump in a few more FPS. (I think in your test, the RX 580’s resources were already being fully utilized and it makes sense that having faster RAM or more ram wouldn’t make a difference.) In order to do this, it seeks help from the CPU. Since there are GPU resources available, the CPU (which is faster than the RAM) assists in the process of pumping in more FPS. The thing is that RAM needs to send this data to the CPU. If the CPU and GPU are “waiting” too long and spending several clock cycles not doing anything productive to increase the FPS, it is because they are “waiting” for the RAM to bring the resources to the CPU. In your case, the GPU’s resources were already being fully taken advantage of. So no matter what, the different RAM kits you used did not make any impact on the FPS because there was no “juice” available left to draw from the GPU. Had you used a more superior GPU and CPU (I say CPU because the single core / single thread score is not very good on Ryzen’s which is required for a higher FPS output), then yes, faster memory would have allowed for a higher FPS output. However, I will mention that your system is more or less a perfect case of no bottleneck. The only disadvantage is that in terms of gaming, upgrading anyone of the components to increase your gaming FPS will provide diminishing returns. If you upgrade the GPU, your CPU will be too slow to take advantage of it and will barely give better FPS and actually lead to stuttering. If you upgrade the CPU, you won’t get better FPS because your RX 580 was already being used to its full potential. If you upgrade the memory, same scenario, you won’t get better FPS because your RX 580 is already being used to its full potential.
Reply
- Replier
  
  Jun 22, 2020 at 1:34 am
  
  RX 580? Blind or some?
  Reply
Kevin Christian Dies

Aug 1, 2019 at 4:16 pm

WePc, I’m building my own pc right now this is a really well define article love it.

Thanks for that info hope to find the best ram suitable for my needs.
Reply
Simon

Jul 24, 2019 at 6:49 pm

Do some testing with ACTUAL GAMES. 3dmark is basically perfectly optimized, something that you’ll never encounter in the real world. In some games you’ll see up to a 10% increase in avg fps going from 2133MHz to 4000MHz and even in games that don’t gain much avg fps the frametime stability improves.
Reply
Seger van Wijk

Jul 21, 2019 at 12:14 am

Thanks very much. Valuable information. I made a mistake buying a double processor motherboard a few years back. I need faster but fewer cores I’ve come to experience. Wanted to check what the effect and importance is of faster memory. It seems that it’s not that crazily important. That’s then saving my a few pop. Let me know where to donate 🙂
Reply
Claire

Jul 7, 2019 at 10:36 pm

Thanks for explaining what memory bandwidth means better than any other article I could find. I’m an artist/filmmaker building a workstation for animation in both 3D (Maya) and 2D (After Effects) as well as editing (4k or less footage).

Today I was trying to decide whether or not to buy one of the latest gen Ryzen processors vs. the 2nd gen version of the same processor, and it concerns me a lot that they seem to have decreased the memory bandwidth for writes by half on the new 3rd gen Ryzens. (source: https://www.techspot.com/review/1869-amd-ryzen-3900x-ryzen-3700x/#add-new-comment) Do you have any thoughts on how this might impact an animation workflow, where new versions of clip are being constantly cached/reloaded into RAM as the artist works? It’s possible I’m not completely understanding how animation utilizes RAM, as all this is really over my head as an artist trying to build a PC. I’m just trying to figure out whether these new processors would actually slow down my workflow as opposed to speed it up?
Reply
Dave

Jun 19, 2019 at 4:22 am

You have a typo in your “Clock cycles” section: “if RAM is rated at 3200 MHz, it performs 3200 cycles per second.”. You’re leaving out the “MHz” of the second part. 3200 cycles per second is 3.2 kHz, or 0.0032 MHz.
Reply
Mingle

Apr 24, 2019 at 9:39 pm

Please take a look not only at FPS but 1% and 0.1% frametime lows. Please use more than Time Spy, such as a CPU-heavy game title (heck, even OG Crysis can batter a modern CPU, although GPU not so much).

I would love to see a memory bandwidth/latency comparison with regards to stutters and framepacing in CPU heavy games.
Reply
Dylan Berry

Feb 6, 2019 at 9:59 pm

Isn’t 1 Hertz (1 – 1 millionth of 1000 Mhz) defined as 1 clock per second and by that default making 3200MHz (effective speed of 1700Mhz) 1.7 million cycles per second not 3200?
Reply
- Richard Jiles
  
  Nov 25, 2019 at 5:26 pm
  
  DDR stands for double data rate, meaning you clock data on the rising edge and falling edge of the clock signal. I haven’t looked into your question much but if you are only thinking that 1 clock cycle is 1 data out then you need to double your results to account for DDR mode
  Reply
Danielski

Jan 14, 2019 at 7:48 pm

There is sometimes big difference betwen for example 1×8 vs 2×4 gb because second option uses dual channel
Reply
Souljazk

Dec 12, 2018 at 9:21 pm

Nice article, thanx for the read.

You should test the various RAM configs, going from 2Ghz-3.6Ghz with games. IE BF, GTA5, Doom, Witcher, Mordor etc. This will help those with newer systems make a good choice.

Also older systems, IE 2nd-4th gen i7’s befit HUGELY from 2133-2400Mhz RAM VS 1600-1866Mhz RAM in games, esp when a 1060-1080Ti is used. As GPU efficiency increases, so the system needs faster RAM to be able to middleman the FPS to the CPU.
Reply

What You Need to Know about RAM Speeds – Does RAM Speed Matter?

The Author Who Worked On This Article

Shaun Conroy

You Might Be Interested In

DDR4 vs DDR5 – will it be worth the upgrade?

How Much RAM Do You Really Need for Gaming in 2022?

15 thoughts on “What You Need to Know about RAM Speeds – Does RAM Speed Matter?”

Leave a Comment Cancel reply