Jump to content
Sign in to follow this  
216th_Jordan

Overclocking DDR3-1600mhz RAM greatly reduced microstutters

Recommended Posts

So, after reading for several years that a RAM overclock is total nonsense and has no effect I finally decided to try it myself:

 

I Overclocked my RAM from 1600Mhz 9-9-9-27 to 2133 Mhz 10-11-10-30 (hell I'm still amazed at that number..) with following effects:

- microstutters almost completely gone, TrackIR is mostly buttery smooth (I put it on high priority in the Task Manager still)

- average fps have risen by about 15

 

 

It appears that Il-2 quite heavily relies on RAM so that it can indeed be a bottleneck on some systems. So If you have microstutters and a motherboard and ram that support overclocking this is well worth a try.

Share this post


Link to post
Share on other sites

Pretty sure I wanted to post this in the Hardware, Software and Controllers subforums. :huh: I'd appreciate if this could get moved when seen by some mod :)

Share this post


Link to post
Share on other sites

My ram is clocked at 3200Mhz and for sure I don't have micro stutters, but i can't say it is because of it. Any way good find Jordan can help others. 

Edited by 307_Tomcat

Share this post


Link to post
Share on other sites

mine came preclocked at 1.65v so I did not change anything with voltage. (Going beyond 1.65 seems to be rather dangerous)

Share this post


Link to post
Share on other sites

I Overclocked my RAM from 1600Mhz 9-9-9-27 to 2133 Mhz 10-11-10-30 (hell I'm still amazed at that number..) with following effects:

- microstutters almost completely gone, TrackIR is mostly buttery smooth (I put it on high priority in the Task Manager still)

- average fps have risen by about 15

 

Can't say much about microstutters (cause I never noticed them yet), but the massive increase in average FPS raised my attention.

Unfortunately you didn't say what FPS you were coming from.

To get some kind of figures, I assumed it would be below 100 so the increase would be 15% at least.

This would mean for me, with an average of 70FPS on a test QMB sortie I usually fly for such kind of measurement, I would have to reach an average of 80FPS now.

This definitely isn't the case. Even more, I can't notice any change in FPS at all.

So I thought "let's take this to a test" and benchmarked the memory with both standard settings and overclocked.

Standard in my case is DDR3-1333 with 9-9-9-24 timing.

Overclocked in my case is DDR3-1600 with 10-10-10-26 timing.

 

This is the result (click for full size display):

39677105962_2cd03442c1_c.jpg

 

Without going into detail, I think it's plain to see that the difference falls within measuring tolerances.

IOW: Overclocking memory is a placebo tuning measure.

 

Cheers!

Mike

  • Upvote 1

Share this post


Link to post
Share on other sites

It appears that Il-2 quite heavily relies on RAM so that it can indeed be a bottleneck on some systems.

As I said I am pretty sure this depends on the system. I'm coming from around 60 fps and went to around 75. (average Berloga 30 players) Most noticable the increase comes in MP (I fly a lot on Berloga and watch average figures every flight). In my testmission in SP (6 fighters on Kuban summermap) it is noticable equally and this doesn't vary (here it went from 100 to 120 fps). But what was really sticking out were the missing microstutters.

 

In my case the ram seems to be the bottleneck with my i4820k CPU at 4.6 Ghz and my Gtx 1070. What are your specs?

 

Please don't just accuse me for falling for the placebo trap, I know about confirmation bias and compared hard numbers. Systems are different and I have had quite some troubles with mine recently which I could now partly solve.

Edited by 216th_Jordan

Share this post


Link to post
Share on other sites

Sorry Jordan didn't want to accuse you of anything, however that 25% performance increase by raising one tiny little single PC's component speed by 21% (33% clock rate increase - 12% latency increase) sounds rather optimistic.

Especially when we take into consideration that your Ivy Bridge CPU isn't built for DDR3-2133 support...

 

My test system was an i5-2500K @4.2 GHz, Nvidia GTX 970 GPU (Palit Jetstream 4GB GDDR5) and 16GB DDR3-1333 RAM.

I'll run a comparative test on another system with i7-7700K CPU @4.5 GHz and 16 GB DDR4-3000 (aka PC4 24000) RAM as soon as I find time to.

 

But you can see what others achieved with different RAM speeds e.g. here: https://techbuyersguru.com/does-ram-speed-matter-ddr3-1600-vs-1866-2133-and-2400-games

For those who are not bothered to read the tech article, they've tested DDR3-1600, 1866, 2133 and 2400 RAM (the latter in CL10 and CL11 fashion) with a wide variety of synthetic tests and actual games, and the result was an average difference of 1.4% between "slowest" and "fastest" RAM, and a maximum peak difference of 3.9% (under very special circumstances) between the two.

This differs in dimensions from 25% (and the OC factor is twice as high in that test) and again, it's highly unlikely anyone would ever notice the difference.

 

Cheers!

Mike

Edited by SAS_Storebror

Share this post


Link to post
Share on other sites

I ran some tests with a memory testing program and came up with 9.2% increase in read and 8.5% increase in write on average.

Ingame the results were not as high as I had them noted from last tests, outside view on my mission averaged ~114 fps for 1600Mhz and ~125 fps for 2133Mhz, which is about 9.6% increase. Cockpit view averaged ~110 fps and ~118 fps respectively, which is a 7.2% increase.

 

Multiplayer I did not test again as the method is not repeatable, only realtime measurement is possible.

 

When panning around however the microstutters became noticable again with the 1600Mhz setting. So in conclusion the errors in measurement before were likely due to not properly setup tests (I did them on the fly, so to say) which leads to a 8-11 fps increase on average now in SP. Microstutters still severely lessen for my system with a higher RAM frequency. So the result is not so much measurable in quantity, but in playing quality. As long as you don't have any problems with the game there's no need to fiddle with this anyways, but in some cases like mine it can definitely help.

 

Don't worry by the way, I was a bit offended this morning, but also got up on the wrong leg. ;)

Share this post


Link to post
Share on other sites

Don't worry by the way, I was a bit offended this morning, but also got up on the wrong leg. ;)

 

Everything's fine Jordan, I know I've got a straight tone in my messages sometimes  ;)

Your figures make sense.

Here is the promised comparative check with the i7-7700K, running DDR4-3000 memory @2133 MHz (BIOS default with XMP disabled), 3000 MHz (XMP enabled) and 3333 MHz (overclocked).

The maximum difference in synthetic figures is slightly less than 10% (click for full size):

 

39707609482_418461be8c_c.jpg

 

Cheers!

Mike

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×
×
  • Create New...