Jump to content
Placebo

Will-my-pc-run-Arma3? What cpu/gpu to get? What settings? What system specifications?

Recommended Posts

Hope that the additional L3 cache will also be available for 5800X and 5600X and not just for 5900X (demonstrated by AMD) and 5950X.

Share this post


Link to post
Share on other sites

5600x here, with 16GB RAM 3600 18-22-22-42 and the CPU on default (no OC or anything else) i get around 43 FPS on the YAAB map on Ultra settings. 

 

With my previous 2600, i would get around 30 FPS, 33 if OC. 

 

How much would i stand to gain if i got better RAM? If it's only a few FPS more, i may decide to stay like this. If it's like 10 FPS, i could be compelled to upgrade to get close to the 60 FPS sweet spot. I mean i could do it now by tinkering with the graphics but hey, it's the principle of it.  

 

Ho BTW resolution is 1440p and video card is Red Devil 5700 XT running on the "quite" bios. 

Share this post


Link to post
Share on other sites

So i turned the game boost on and set the memory to 3800 using the memory try it option and i got 41 FPS. That's inspiring. 

 

[EDIT] Turned off game boost and reset memory to XMP2 and tried Precise Boost Overdrive, set on "enabled". No gains, i'm back to 43. 

 

This isn't fun. At least pretend i'm making a difference damn you. Guess i have to set down and try a proper OC of both CPU and memory and see how much i can get out of what i have. 

Share this post


Link to post
Share on other sites

@opus132 simply tune the one you have already.

Changing only primary timings won't give you much.

It's secondary and teritary timings that count.

Share this post


Link to post
Share on other sites

Seems i'm unable to tune anything, possibly because i'm running this on a B450 board and the bios may be limited in what i can do. Everything appears to be set on auto and i can't change any of the secondary or tertiary timings. 

 

I think i'll settle with this. A friend of mine has some 3600 CL16 memory i'll see if i can convince him to bring it over to do some test, just for curiosity's sake. 

Share this post


Link to post
Share on other sites

According to my tests, the differences will be marginal, as here:

 

ocsf.jpg

 

With a given processor, the only effective way to increase the FPS level is to decrease the Global Visibility setting.

Share this post


Link to post
Share on other sites

Leaving secondary and tertiary timings on auto at which they are very high, yes, gains will be marginal.

Share this post


Link to post
Share on other sites
On 6/16/2021 at 9:38 AM, oldbear said:

According to my tests, the differences will be marginal, as here:

 

ocsf.jpg

 

With a given processor, the only effective way to increase the FPS level is to decrease the Global Visibility setting.

 

Ok, i think there's something wrong with my configuration. 

 

I tried running the test in standard by pressing S while the benchmark was running and i got 51 FPS. 

 

So i tried manually setting the game to standard from the graphic options, and i ended up with 61 FPS. 

 

What kind of memory do you have? Perhaps primary timings do make a difference, or else something is wrong with my BIOS configuration. 

Share this post


Link to post
Share on other sites

Well ...

First, I would suggest, first, to create some sort of John Doe profile specifically for testing.

This achieved, for comparison purposes, before the test, set the presets to Standard, then at the start of the test press the "S" key.

This allows the parameters to be set identically.

 

For information, for these tests, I was using a 1080p screen.

 

I was using ...

G.Skill TridentZ Neo 16 Go (8Gox2) 3600 MHz C16 [16-19-19-39]

G.Skill TridentZ Neo 16 Go (8Gox2) 3800 MHz C16 [16-19-19-39]

 

Share this post


Link to post
Share on other sites

Ha yes, i forgot i was at 1440p. 

 

I followed what you said, and i got 64 FPS. This is with precision boost overdrive and game boost off. 

 

I'm going to borrow my friend's memory and see what difference that actually makes. 

Share this post


Link to post
Share on other sites

And again, increasing only RAM frequency and/or maybe lower only few primary timings won't give you much FPS.

 

Dozens and no just few secondary and tertiary timings are all on auto and are very high and this is where additional FPS is lost.

Leaving 3600 MHz XMP timings and simply going with same timings but 3800 MHz is useless.

 

XMP profile only determines primary timings, RAM voltage and frequency and doesn't touch any of dozens of secondary and tertiary timings.

 

So when you go from 3600 MHz to 3800 MHz and you think your timings are still same - they're not. It's only 4 first timings that don't change, but they also don't influence FPS much.

 

Dozens of secondary and tertiary timings that are already very high on auto at 3600 MHz, when switching to 3800 MHz are even much much higher after frequency change.

Because the motherboard tries to keep the system stable and makes secondary and tertiary timings too high rather than not let the system boot or crash.

Share this post


Link to post
Share on other sites

And the fact that you will try the RAM of your friend with slightly better but again only primary timings, will only confirm your suspicions that gains are only marginal.

 

But it's completely wrong, since it's only primary timings, which are less important than secondary and tertiary timings and there are only 4 timings out of dozens of them in total.

Share this post


Link to post
Share on other sites

Well, at this time i'm simply trying to figure out why i am behind oldbear. We have the same CPU and a similar GPU we should be roughly the same. I'm concerned i may be losing performance because of something else. 

Share this post


Link to post
Share on other sites

From my point of view, what I think about first ...

1° 1080p vs 1440p. The extra pixels weigh heavily on performance.

2° Golden sample. It is possible to verify the qualification with CTR. The differences between a Gold CPU and a Bronze one can be significant.

3° Over-ventilated test platform. It' a bench table encased in a 19" rack featuring 3 x Noctua NF-P12 redux-1700 PWM just in front.

Noisy but very efficient. Adding to the 140 mm Noctua cooler, CPU temp never over 58° C during the tests.

Share this post


Link to post
Share on other sites

Well i did run the test at 1080p so that's out. 

 

I'll look into CTR. Is the diagnostic safe? I'm not sure i want to risk damaging the CPU just to see how far i can push it and even if i only get a bronze rating i may just keep it. I'm not really planning to OC (i'm broke can't afford to mess around with expesive parts, heh) i just want to make sure everything is in order. 

 

Share this post


Link to post
Share on other sites

Yes, there is a big variance in quality of Ryzen 5000 CPUs, which can result in noticeable performance difference in different applications.

Share this post


Link to post
Share on other sites

Well, a straight run on the diagnostic in CTR tells me i have a golden sample but i'm not sure that counts, everything is in default i think you need to at least tune something, right? 

 

Tried my friend's RAM. Three tests, no OC, and i got 68, 69 and 70 FPS. So i gained a couple of FPS, about 4 or 5. I also tried Cinebench and i basically gained nothing there. Definitely not enough to justify replacing what i have. I suppose if i really tried i could match the results of oldbear, with new ram and tuning through CTR but to be honest i think what i have is good enough. Definitely better than what i used to have on the 2600. I think the game is perfectly playable now. 

 

Share this post


Link to post
Share on other sites

Cinebench doesn't care about RAM at all.

And again, no tuned secondary and tertiary timings but yet such conclusions.

 

Just tune the RAM you already have as much as it's possible and that's it.

 

Geekbench is really sensitive to RAM. 

Share this post


Link to post
Share on other sites

Well, i thought i was locked out as all the seconary and teriary timings are stuck in auto but i just figured out you can change them somewhere else in the BIOS menu. It's kinda of confusing because they had to remove every graphic in the bios to make room for the support of newer CPUs and now it's a mess to find anything. 

 

To be honest i'm also getting crosseyed just by looking at them: 

 

Zen-Timings-Screenshot.png

 

 

Share this post


Link to post
Share on other sites

All your timings are super mega ultra high.

 

An exemple is tRFC 631!

It could be 300 or even lower.

 

TWR is at least 2 higher than it could have been, or even 3 times higher, if not 4 times higher.

 

tREFI is 14029, but should be 65535.

 

Etc.

 

A lot of work and a lot can be optimised.

Share this post


Link to post
Share on other sites

Do i really have to change so many or are there "key" ones that matter most? Since i have to go by trial and error it would cut time if i could start with the more important ones. 

Share this post


Link to post
Share on other sites

Simply changing random timings to random values won't help you much.

Performance might be even worse or system unstable/crash.

 

Need to google this. There are dozens of guides. You need to understand what works how and the dependencies and only then you start to OC RAM.

It's science, not easy like GPU/CPU OC.

Share this post


Link to post
Share on other sites

Hey guys, I'm about to buy a new PC System and before buying I want to make Sure, that ARMA 3 runs properly. The System I am interested in has an AMD Ryzen 7 5800x CPU and a GeForce RTX 3070 (8GB) GPU. In the past I heard about issues of AMD CPUs running ARMA 3. Is this still problenmatic or can I expect that everything runs well?

Thank you guys!

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now

×