I have seen a lot of misleading information out there that upgrading from 8gb RAM to 16gb RAM can help improve framerate and remove stuttering. For example, this youtube video
misleads people into thinking that you get more FPS with 16gb of ram.
The actual reason is single vs. dual channel memory. When RAM is in dual channel mode, the CPU has access to twice the amount of RAM bandwidth. People who are using 8gb of RAM in single channel configuration will have half of the RAM bandwidth of people who are using 8gb of RAM in dual channel configuration. In the past, games were typically not CPU-bound, so the extra bandwidth didn’t really matter. However, Battlefield 1 is unique in that it is CPU intensive, so differences like dual and single channel RAM actually matter.
To test this out, I got my hands on two 4gb and two 8gb sticks of DDR3-1600 RAM. Using these, I was able to test a number of different configurations of RAM capacity and channel configuration.
Base computer specs: i5-3570k @ 4.2ghz, AMD 7850 @ 1000mhz
To test different configurations, I went to an empty server on team deathmatch, Argonne Forest. I stood in the same spot and looked at the same location for each of these tests. I used MSI Afterburner to measure FPS. I used MAXMEM to perform RAM benchmarks. In game settings are all low, set to 50% resolution scale, in order to force a stronger CPU bottleneck.
As you can see, the amount of RAM (8gb vs 16gb) does not matter. The aspect that matters is whether or not the RAM is running in single or dual channel mode.
I hope this information is useful to everyone, thanks.
Thanks for the awesome post and lol at people that think system RAM size matters for gaming. Most of the memory usage for games nowadays is GDDR5 on your graphics card.
Running low on ram results in page file being accessed and can cause frame drop stutter. Regardless what you think you don’t know people’s systems without at least screens shots of ram usage between the game and windows. 8gb is cutting very close no doubt lot of people who have no clue are running low on resources.
And who now a day not using dual channel. I was running triple like 5 years ago and now quad.
I get frame drop and stutter with ddr4 quad channel the problem is CPU load.
They really need to fix RAM Usage if its gonna be like that.
The game shouldn’t affect your performance dramatically if your ram slots are either single channel or dual channel.
It’s kind of misleading because to truly test this, you actually need to do a frame average while actually playing. And the amount of ram does matter if you’re running 1440p or higher on ultra. I’ve seen some at 1440 using 12gb ram. For me, at medium on 1080, I’m using just under 6gb.
Its actually 100% misleading because the argument being made is that the amount of ram doesnt matter. The data he then shows is the performance difference in memory bandwidth. Here’s the kicker, when the game runs out of memory the fps will only drop for a very short period of time like under 1 second then resume to play as normal until it needs to load something not in memory again. The test isnt even valid. Sorry op.
Honestly, your test method is poor. The only thing that it proves is that 16GB won’t benefit over 8GB when playing on empty servers, or in other words when load is minimal. But no one plays on empty servers, do they?
The game is much heavier on the system with 64 players on the server than it is with you alone on an empty server. Therefore, testing on one of the heaviest maps (like St. Quentin Scar for example) with 64players would be ideal. I think Argonne is one of the least demanding maps in the game for CPU.
I guess my thing is, RAM is the cheapest component these days, yet one of the most important… why not just get at least 16g of very good RAM and be on the safe side?
Sure your CPU usage is 100% , but your RAM usage is around 5000MB!
I’ve played the game with both 8GB and 16GB installed and with no other software than MSI Afterburner running in the background. In 64 player maps with 8GB installed, RAM usage is typically around 6000-7000MB. It doesn’t ever exceed 7000MB by much. With 16GB installed RAM usage is typically around 6500-8000MB and can sometimes even exceed 8000MB. I can definitely see this having an influence on performance and your test misses this completely.
On Argonne Forest RAM usage is lower than on the heavier maps and is pretty much the same whether I have 8GB or 16GB installed. Maybe because it has no vehicles and is smaller than other maps.
