Test 1: Battle of the DDRs (2GB DDR3 vs 4GB DDR2)

[Kevin W]

**I'm not an expert. What works on my system doesn't mean it works on others**

Intro

We all know that fsx is CPU dependent, but how much do the other items in a computer affect the fps. I attempt to find out. I know that frame rates don't tell the whole story, but it's hard to measure qualitative items, but I will attempt to try as well.

How the tests are run

For these benchmarks I'm more interested in the low-end performance than top end, such as big cities and clouds. By running fsxmark07 and looking at the results I found that the low end numbers are during the first half of the test. So I only run fsxmark07 for 2.5 min instead of 5 min. For clouds I tried different cloud settings and found that fog had the lowest fps. Each series of tests starts off with a trial run then the next 3 runs are recorded and the average of the three runs are shown here. I figured this would give a better result as I've found some variation between same tests.

All OS tray items were shut down such as skype, anti-virus, etc.

System Setup

Motherboard: ASRock P43Twins1600 (P43 chipset, supports DDR2 & DDR3)
CPU: E8400 @ 3.6GHz (can run DDR2 @ 1066 & DDR3 @ 1333)
PSU: Antec 550W
GPU: XFX 8800 Ultra
Hard Drive: 500GB WD SATA II (OS) / 160GB SATA II Samsung (FSX)
OS: Vista32 SP1 (not a fresh install)
FSX SP2
Addons: GEX, UTX, SceneryTech NA Landclass, and a bunch of planes
nHancer setup: http://www.simforums.com/forums/forum_p ... 769#156769

Test 1: Battle of the DDRs (2GB DDR3 vs 4GB DDR2)

I am going to upgrade my ram from 4GB DDR2 800. The price difference between 4GB DDR2 and 2GB DDR3 was very close together so I bought and tested them out here. The contenders:

* 2GB DDR3 1333 OCZ Reaper 6-6-6-18
* 4GB DDR2 1066 Corsair 5-5-5-15


I used the 16xS in nHancer as it had less blurries than the 8xS setting. Results:





Conclusion

In the default run the DDRs are close together, but in the clouds the DDR2 is the clear winner. So in this case more RAM is better than faster. Also, when I fly I often have a browser open which would also use some ram. Qualitatively I didn't notice any difference between the two.

Up Next

Test2: Battle of 8xS vs 16xS (nHancer)

-Kevin

[anikun07]

Very cool test!  It's always nice to see recent tests for FSX and hardware as opposed to the tests I've been seeing done that are from 2006/2007.  You've gotta nice machine put together.  I'm still running XP Pro, but I've been considering Vista to see DX10 graphics.  I've running it with 2x1GB G.Skill DDR2 800 4-4-4-12  - although my BIOS are running it at 5-5-5-15.  I tried changing it in the BIOS to what G.Skil said it can run at but the system didn't boot.

Cool test!

[NickN]

Sorry Kevin there is no battle if you know how to use DDR3 and select the right memory/motherboard/processor to do the job.

Your tests are flawed.. they rely on the frame counter which can not in any way shape of form relay correct data about perf in FSX
I have demo'd a very long time ago the difference between DDR2 and DDR3 speed on the correct timing and settings

The only thing that was changed in the tests I ran were memory speed and memory timing... everything else was exactly the same

http://forums.avsim.net/dcboard.php?az= ... 7210&page=



[quote]I would probably pay more attention to those +3% to +5% TRUE average increases in benchmark frame rates in conjunction with the lower timing, lower latency memory products

[macca22au]

I am unclear.  I thought DDR3 v DDR2 is a mattrer of motherboards.  Surely the choice is automatic once you move into the latest CPUs?

[NickN]

With i7 DDR3 perf is obtained through Quickpath and requires no setup or clocking


With any setup before i7 you must be running at minimum 1333 5-5-5- or 1440 6-6-6 (or faster) @ 400-450FSB or there is no advantage to DDR3 over DDR2

[Kevin W]

Thanks for your responses!

[NickN]

I quit flying my FSX becuse honestly, my system, as you have put it, is still in the FS9 era, and since I have niether the money or knowledge it seems to build one capable of running FSX like it will FS9, I am starting to feel kinda cheated by MS that it wasn't made very clear that I would need to know how to overclock and fully understand computer science in order to enjoy my FSX experience.

Dont blame MS

MS designed FSX just like FS9.. no one was flying full-on FS9 till 2006 without a massive budget and experience and many still dont. MS designed the title to have a shelf life and that is based on hardware and the next release in 2010

Many wont migrate to FS11 right away and MS knows that too... so in 2010 you will be saying the same thing... you cant run FS11 but run FSX flat out

[NickN]

[quote]Thanks for your responses!

[Kevin W]

Nick,

I run a cheapo ASRock P43.

[Kevin W]

These are the trial version so I'm not sure if you can do anything with them but here it is.  This one's at 3.6 GHz FSB 400 1333 MHz 6-5-5-15 PL 7



This one's 3.75 GHz FSB 417 1388 MHz 6-5-5-15 PL 8

[NickN]

Looks like you are doing a bit better Kevin however that memory is simply not going to allow you to run at CAS performance for the speed

The next step up is 1600 7-7-7-20 trd 6 or 7 for 2x2GB and 6-7-7-24 for 1x2GB

you can try 7-7-7-24 and see if the memory will run it

You are using 1x2GB and becasue of that the proper timing is 1CLK lower than what you would typically use at 2x2GB

1x2GB

1333 5-5-5

1440 6-6-6

1600 6-7-7

If it wont do that and the memory is running the highest voltage allowed by OCZ, and, the northbidge voltage can not be used to assist, then the memory is simply not good performance product and that is as far as you are going to get

[Kevin W]

Thanks Nick.

Looks like this RAM just won't cut it, but I'll see if she'll run at 1600.

What would you recommend for a motherboard / ram combo fairly cheap, say under $400.

I would like to run DDR3 and a 450+ FSB (try to get in the 4 GHz range) if possible.

[Kevin W]

No go at 1600...

Thanks for your help Nick!

If I want to run this test again, what timings would you recommend for 4gb DDR2 1066?

[UU]

2.1V

[NickN]

No go at 1600...

Thanks for your help Nick!

If I want to run this test again, what timings would you recommend for 4gb DDR2 1066?

minimum 5-5-5-15 or 18 tRD 6