BF4 and High Speed Ram

B

Bojamijams

Limp Gawd
Joined
Jun 5, 2008
Messages
178
Hi guys.

Has anyone tested Battlefield4 with high speed ram vs 1600 to see if there is a difference in frames like Corsair claimed?

I'd be very interested in non corporate testing/results.
 
If you give me a key I will be happy to test will some 2133 ram. I'll even oc it for you.
 
Yes, there have been tests and it's true. BF4 likes the faster ram.
 
Would be interesting to hear what others are noticing too... that's pretty awesome.
 
Has anyone found any reviews on this yet? Is it worth upgrading from 1600?
 
Dustins said:
Has anyone found any reviews on this yet? Is it worth upgrading from 1600?
Click to expand...

Been wondering the same thing. I went from 1600mhz to 2400mhz but can't say if it's improved anything since I also went from an i5 to a 6 core, and multiple patches/drivers have come out since.
 
Hooy said:
Been wondering the same thing. I went from 1600mhz to 2400mhz but can't say if it's improved anything since I also went from an i5 to a 6 core, and multiple patches/drivers have come out since.
Click to expand...

While keeping everything the same, you can downclock your DDR3-2400 RAM to DDR3-1600 to see if you note any differences.
 
I would, but I don't own the game. I am pretty curious about that claim, though. Faster RAM doesn't seem to matter much with anything else.
 
Marcdaddy said:
Would love to see this run on some DDR 3 3000mhz modules
Click to expand...

I'm more interested in 1600 and 1866. Would be interesting to see if it's linear straight to 2133 or is there a big jump from 1333 to 1600 and 1600 to 2133 shows much smaller changes.
 
Tsumi said:
I'm more interested in 1600 and 1866. Would be interesting to see if it's linear straight to 2133 or is there a big jump from 1333 to 1600 and 1600 to 2133 shows much smaller changes.
Click to expand...

I'm debating going with a 4930k and 16 gigs if DDR 3 3000 and benching in increments upto it would be awesome. Just trying to hold off until next year with Haswell-E , Maxwell and ddr4
 
Sprkd1 said:
http://www.overclock.net/t/1438222/battlefield-4-ram-memory-benchmark
Click to expand...

This testing cant be used to evaluate if memory above 1600MHz will help.
It is well established that 1600 to 1800MHz ram is the sweetspot for Sandybridge onwards, so its no surprise that 1333MHz isnt as good.
As the user is pretty new, this could be a shill.

trick0502 said:
Bit tech did some ram testing, but they used single player. Their results didn't show that much of a difference.

http://www.bit-tech.net/hardware/graphics/2013/11/27/battlefield-4-performance-analysis/9
Click to expand...

More in line with expectations.
 
I haven't OC'd memory in quite some time. It stopped making much of a difference when 1366 i7 showed up.
 
I actually dropped my CPU clock and raised my ram speed up to 1600mhz and got better results in BF4. Today I am upgrading from 6GB 1600MHZ to 12GB 2000MHZ. There should be an improvement, but I don't expect much.

I think more likely than not the faster ram does help. Remember the consoles are using GDDR5 as system memory right?
 
Blackstone said:
I actually dropped my CPU clock and raised my ram speed up to 1600mhz and got better results in BF4. Today I am upgrading from 6GB 1600MHZ to 12GB 2000MHZ. There should be an improvement, but I don't expect much.

I think more likely than not the faster ram does help. Remember the consoles are using GDDR5 as system memory right?
Click to expand...

GDDR5 is just a variant of DDR3. It is optimized for high speeds, but has high latency as well. The higher speed allows it to excel at highly parallel workloads (what GPUs work on), at the cost of lower performance in serial workloads (CPUs).

Also, I believe the Xbox One uses DDR3.
 
Well I can tell you this I just went from 6 GB of DDR 1600 to 12 GB of DDR2000. Ran both at stock speeds. I can now run textures and post processing on ultra at 1080p. The game absolutely chugged before. And the thing is, technically I am short on VRAM with only 1.5. GPU usage is now at 99-100 percent all the time, before I was getting dips.

I don't know if it was the speed or the amount of memory or both, but it was a HUGE upgrade and I am no longer considering a video card upgrade until Maxwell comes out.

I am using tri channel Corsair DDR3 1600 2000MHZ Dominator-GT.

Also I should mention EVERY aspect of system performance has improved including OS responsiveness, ect. And outside of BF4, I was NOT short on ram with 6GB. When it comes to memory I say go big or go home!
 
Last edited:
Blackstone said:
Well I can tell you this I just went from 6 GB of DDR 1600 to 12 GB of DDR2000. Ran both at stock speeds. I can now run textures and post processing on ultra at 1080p. The game absolutely chugged before. And the thing is, technically I am short on VRAM with only 1.5. GPU usage is now at 99-100 percent all the time, before I was getting dips.

I don't know if it was the speed or the amount of memory or both, but it was a HUGE upgrade and I am no longer considering a video card upgrade until Maxwell comes out.

I am using tri channel Corsair DDR3 1600 2000MHZ Dominator-GT.

Also I should mention EVERY aspect of system performance has improved including OS responsiveness, ect. And outside of BF4, I was NOT short on ram with 6GB. When it comes to memory I say go big or go home!
Click to expand...

I experienced the same when I went from 8GB 1600 MHz CL8 >to> 16GB 2133 CL9.

The ~$40 I spent after selling my old RAM was well worth it!
 
I've tested the difference between 1600-1866 2x8GB.

I went from 29 FPS to 34 FPS using a single card in eyefinity.

I will be doing more testing to see the difference once some faster ram comes in.
 
Priller said:
I've tested the difference between 1600-1866 2x8GB.

I went from 29 FPS to 34 FPS using a single card in eyefinity.

I will be doing more testing to see the difference once some faster ram comes in.
Click to expand...

Not possible.
You are saying that the ram is a 100% bottleneck for everything and the framerate jump is even higher % than the % increase in ram speed.

Something peculiar is going on with these claimed memory speed (not size) performance figures.
 
I should mention that I did not get the same improvement in Crysis 3. It was Battlefield 4 that gave the improvement. Now, I still have a video card vram bottleneck, and it rears its ugly head everything things get really crazy. It just isn't as much as a bottleneck as I thought it was before I made the ram upgrade. Ultra textures and post processing at 1080p was impossible before the upgrade. Keep in mind I went from 6 to 12 gigs.

What is more interesting to me is that to achieve the 2000GHz ram speed, I lowered my CPU clock speed from close to 4 ghz to more like 3.4 ghz. Absolutely no performance loss in BF4 downclocking the CPU.

I have no clue what it means. I need to run more tests. For one thing I need to drop the memory speed back down to see if wasn't the quantity of memory that made the difference.

Also, food for thought. I run a lot of high definition audio and video files on a program called JRiver Media Center. Even when I had only 6GB of ram, a 1080p movie never used more than 15-20% of my ram. Memory usage for those types of media files was negligible it was so low. But now that I have the new sticks in, memory usage is even lower of course because i have more ram, but now movies and music load almost instantaneously, and on a 1080p movie i can skip to any part of the film instantly whereas the 1600 sticks would lag for a few seconds. And all this with a 500 mhz downclock on the cpu to get to the higher memory speed.

I see a lot of comments on the forum indicating that x number of ram and x speed is more than enough. The community seems rather dismissive on the importance of ram but I wonder if subjectively the higher speed memory or having a larger pool of memory makes a bigger difference in everyday usage than that crazy CPU overclock we all strive for.
 
Last edited:
Nenu said:
Not possible.
You are saying that the ram is a 100% bottleneck for everything and the framerate jump is even higher % than the % increase in ram speed.

Something peculiar is going on with these claimed memory speed (not size) performance figures.
Click to expand...

Didnt say ram was the bottleneck.

I tested it multiply times between taking the averages and comparing the 2.

This was in BF4.

Also the difference is 16.6% is ram speed versus 17.2% in FPS so its within the margin of error went you start talking rounding several different runs together.
 
Priller said:
Didnt say ram was the bottleneck.

I tested it multiply times between taking the averages and comparing the 2.

This was in BF4.
Click to expand...

You explicitly did state ram was the bottleneck.

The only change in your system was memory speed which increased by 16.6%.
Your framerate increased from 29 to 34fps which is a performance increase of 17.2%.
Even if they matched, it wouldnt be possible, the graphics card is not ALWAYS reliant upon system ram speed for its performance.
 
Nenu said:
You explicitly did state ram was the bottleneck.

The only change in your system was memory speed which increased by 16.6%.
Your framerate increased from 29 to 34fps which is a performance increase of 17.2%.
Even if they matched, it wouldnt be possible, the graphics card is not ALWAYS reliant upon system ram speed for its performance.
Click to expand...

So you're a telling me that tests that I did over and over again when I didnt even believe the results are somehow wrong...

Why dont you go try it then come back and report your findings instead of saying something you havent tested yourself is wrong.
 
Found my test sheet.

Highest average was 29.435 FPS with memory at 1600

Highest average was 34.133 FPS with memory at 1866

difference was 15.960% in performance with memory increase was 16.6% in speed.


Like I said before, within the margin of error when you start rounding numbers.
 
Priller said:
Found my test sheet.

Highest average was 29.435 FPS with memory at 1600

Highest average was 34.133 FPS with memory at 1866

difference was 15.960% in performance with memory increase was 16.6% in speed.


Like I said before, within the margin of error when you start rounding numbers.
Click to expand...

I'll be very surprised.
What are you measuring average fps with?
I'll have a go too.
 
Nenu said:
I'll be very surprised.
What are you measuring average fps with?
I'll have a go too.
Click to expand...

FRAPS.

I did all the tests identically. Down to what programs where running in the background.

BF4's in game FPS counter seem to report higher FPS so I stuck with FRAPs since that is what I started with.
 
First run with memory at 2220MHz before rebooting, loads of things open/running, just out of curiosity.
100.305fps average

Clean boot at 1600MHz
102.745fps average

Clean boot at 2220MHz
101.473fps average

A bit quicker at 1600MHz, but there was only one test so it could be one of those things.
In my case there is no bottleneck at all from system memory.

Can I ask you to provide an Afterburner graph showing CPU use on all cores as you bench it with fraps with the slower memory?
It wont display framerate but this might help me diagnose why it occurs for you.
 
Nenu said:
First run with memory at 2220MHz before rebooting, loads of things open/running, just out of curiosity.
100.305fps average

Clean boot at 1600MHz
102.745fps average

Clean boot at 2220MHz
101.473fps average

A bit quicker at 1600MHz, but there was only one test so it could be one of those things.
In my case there is no bottleneck at all from system memory.

Can I ask you to provide an Afterburner graph showing CPU use on all cores as you bench it with fraps with the slower memory?
It wont display framerate but this might help me diagnose why it occurs for you.
Click to expand...

What CPU do you have? Your testing BF4? What Settings did you use? Cores unparked?


I didnt log Afterburner.

It could be textures or 3 monitors putting more of a strain on the ram.

I also logged several runs in the same server on different maps(same game mode), averaged those together. Then did the same at 1866 in the same server.

My average as I said was a tick over 34 FPS but atleast 1 map came in at 31.986.
 
I'm on a 2500K at 4.4GHz, cores unparked, 290X gfx.
All settings on max, no AA, vsync off, FOV 70.

Can you run it again at and do afterburner screen caps with all cpu cores?
I have an idea why this happens for you, I would like to confirm it.
 
I cant seem to find it now but I came across a review awhile back where they were testing different memory from 1600 to 2400 on an AM3+ board with a FX8350 and there was very little difference in any of them. I remember the 1866 at 9-9-9 timings had the best performance but it was only by a very small margin. I don't recall them testing BF4 specifically though. it seems to me that with newer systems either intel or amd ram is no longer a bottle neck so long as its 1600 or higher and you have enough of it and that's pretty much the conclusion they came to also.
 
I went from 1866 corsair 9/9/9/24/1t to 2133 9/10/9/24/1t and I'm getting about 8fps extra on the low side and the game feels smoother. 2600k@4800mhz, windows 7, 2560x1080 29in lg Ultrawide, 3 EVGA SC 780 Gtxs 1180/7000
 
Marcdaddy said:
I went from 1866 corsair 9/9/9/24/1t to 2133 9/10/9/24/1t and I'm getting about 8fps extra on the low side and the game feels smoother. 2600k@4800mhz, windows 7, 2560x1080 29in lg Ultrawide, 3 EVGA SC 780 Gtxs 1180/7000
Click to expand...

Are you speaking of BF4 specifically or are you getting the same gains across the board in all games and\or programs? It almost seems that BF4 was coded to take advantage of system memory. did any one monitor ram and cpu usage during any of these tests I mean were either one highly loaded at the time these gains were noticed?
 
tcar said:
Are you speaking of BF4 specifically or are you getting the same gains across the board in all games and\or programs? It almost seems that BF4 was coded to take advantage of system memory. did any one monitor ram and cpu usage during any of these tests I mean were either one highly loaded at the time these gains were noticed?
Click to expand...

So far just in BF4 is all I tested. Ill moniter ram and also test out Crysis 3 and Tomb Raider.
 
You must log in or register to reply here.
Back
Top