Jump to content
Placebo

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

Recommended Posts

d248vk7w.png

patch 1.92.145639 + Win 10 Pro 1903

R7/R9 3700X (8C/16T)/3900X (12C/24T) (SMT on) manually limited to 3.5 GHz all cores + 4x8 GB @ 3200 MHz 14-14-14-36 CR1 (2x F4-3600C15D-16GTZ) + Vega 64 Strix OC

 

+2.0 FPS @ 3600 MHz 15-15-15-36 CR1 (4x8 GB 1:1 mode)
+0.5 FPS @ 4000 MHz 16-16-16-36 CR1 (4x8 GB 2:1 mode)

ud71j9t.jpg


Some more FPS should be doable @ >3.5 GHz all cores + 3733/3866 MHz RAM (1:1 mode) with very aggressive timings.

 

*Going past 3600 MHz RAM, automatically switches to 2:1 mode, but user can manually set it to 1:1 mode and try going past 3600 MHz (with newest AGESA firmware). Not sure if a X570 mobo is required for that.

 

1730 GB/s read/copy L1 for R7 3700X @ 3.5 GHz all cores + 2933 MHz 14-14-14-35 CR1 AGESA 1.0.0.2 (Crosshair VIII Hero WiFi (BIOS 0066))

Which results in:

108 GB/s per thread (SMT on)

216 GB/s per thread (SMT off)

3700_aida_write_suakl5.png

 

Source: planet3dnow.de

 

P.S. - have 62.5 FPS in YAAB @ 1080p high with Win 10 1903 vs. 57 FPS R7 3700X 3.5 GHz all cores (SMT on) 3200 MHz CL14 RAM.
My 4790K: 4.8/4.4 GHz core/cache 2400 MHz CL10 RAM all timings optimized + latest perf build + CMA AVX2 dll + huge pages on + HT on

 

See no need to upgrade for ArmA (in my case).

 

if RAM runs in 2:1 mode, it means that infinity fabric runs 2x slower! if 2000 (4000) MHz RAM, then infinity runs only at 1000 MHz, which is crap!
So better to stick to highest possible RAM frequency possible in 1:1 mode and tighten the timings.
For sure it will be less than 4000 MHz 2x16/4x8 GB in 1:1 mode and this is what everybody will have to run to extract max performance possible.
Nobody will want to run infinity at half the speed to show 4000+ MHz RAM speed to ignorant users.
 

When running Intel CPUs with high frequency RAM, like 2x16/4x8 GB 4133/4400 MHz, CPU cache doesn't have to run at half the speed!

So current Intel CPUs (despite the price) have no competition in RAM latencies vs./and frequency.

  • Like 1
  • Thanks 1

Share this post


Link to post
Share on other sites

I am more focused, on the one hand on the operation of the processor "out of the box", and on the other hand on the balance between performance and cost.

Based on the first informations, Intel will keep the blue ribbon with Arma3 because of its highest-end processors higher clock speed and the capacity to reach 5 GHz, otherwise, AMD seems completely in the race.

So the i9-9900K and the i7-9700K are still on top at a price, but  Ryzen 7 3700X / Ryzen 9 3900X are not far behind, anyway in any case, the cost of the platform is very high,  because the cost of the motherboards based on the X570 rises to the same level as that of the motherboards needed for their Intel counterparts.

Now, I am waiting for R5 3600/ R5 3600X tests and reviews on previous B450/X470 motherboards for budget-conscious players 😏

 

  • Like 2

Share this post


Link to post
Share on other sites

Best combo (currently) is X470 mobo + 3700X

No flimsy chipset fan and better temps + when using new PCI-E 4.0 NVMe SSD on X470 (PCI-E 3.0), you get same speed as X570 (PCI-E 4.0) if you put a 3700X instead of 2700X.

You won't reach 4400-4600 MHz RAM speed, but it's just for numbers. Latencies and timings are more important for ArmA.

 

Crosshair VII Hero (X470) vs. VIII Hero (X570) RAM capabilities

05093707886li8kzv.jpg

 

3900X X470 (PCI-E 3.0) vs. X570 (PCI-E 4.0) + AORUS NVMe SSD 2TB ( PCI-E 4.0)

05094119405sefjg9.jpg 05094119218slkks2.jpg

 

3900X X470 (PCI-E 3.0) vs. X570 (PCI-E 4.0) + 5700XT (PCI-E 4.0)

3DMark Fire Strike (1080p), Extreme (1440p) and Ultra (2160p)

05093820651lodkl8.jpg

 

3900X X470 (PCI-E 3.0) vs. X570 (PCI-E 4.0) + 5700XT (PCI-E 4.0)

3DMark Time Spy Extreme (2160p)

05093820606lv9ksv.jpg

  • Thanks 1

Share this post


Link to post
Share on other sites

btw, ASRock Extreme4 and Gigabyte Aorus Pro cost 80-90 € more than their Z390 counterparts!

Share this post


Link to post
Share on other sites

You guys are all way over my head but any laymans reasons to now go AMD (for Arma) -the last I heard was no matter the CPU speed, Intels just played Arma smoother. Im still on a 6600k at 4.3 and have been Intel for 20+ years but this Ryzen looks promising

Share this post


Link to post
Share on other sites

As soon as there is a prize target of <600€ for a X570+3700X combo met, I will give my X470+2700X combo to a friend and get a new.

But it helps just for SP and pure client experience.

MP still suffers from the bad utilization of server HW by BIS.

BIS promised massive battles which I translate into 100 human players among 1000+ AI.

BIS server SW + 2x HC on a monster server can't do that.

Since the server must pump all he gets from the HC also throught the entire stack and sync it like all from a normal client, I do not see how BIS want to fullfill their promise with this approach. So I am happy with all eye-candy served by BIS, the MP experience is ...well...decades after initial promise still not simulating battles properly....maybe shooting of some gangs in the backyard is performing well.

Share this post


Link to post
Share on other sites

Hi, could you tell us where is the source of this ..."BIS promised massive battles which I translate into 100 human players among 1000+ AI." ... ?

 

RV Engine was not built for that task, let's remember what Joris Van't Land has said about this when he leaves A3 project Lead

Quote

 

We're also developing using an ageing engine and tool-set. Real Virtuality still does certain things better than the newest versions of popular alternatives out there, but we're not oblivious to its short-comings (which is why a new Enfusion engine is in development). Still, for an engine never designed with 100+ player persistent and deeply modded servers in mind, amazing things are accomplished with it every day. The freedom offered to modders is another extremely powerful double-edged sword. It gives us plenty of head aches trying to keep backwards compatibility with mods in mind, or to adapt to changes in Windows, hardware drivers, third-party libraries and Steam. In many ways Arma 3 is an uncontrollable beast that we all love very much.

 

Source : 'Report In' dev interview of 2016

 

Anyway, I can't get the relationship between this personal wish and the hardware requirement need to play Arma3, which remains the subject of this topic.

Share this post


Link to post
Share on other sites

What do you people think in regards to the Ryzen 3000 processors strong performance over previous gen and even Intel?

Is the improved IPC mostly based in the refined architecture (and new lithography) or because of the immensely increased cache size?

I wonder to what degree the performance improvement stems from which aspect of Matisse.

 

If the major IPC improvement stemmed primarily from their "game cache", then that would let me to believe that perhaps, if Intel had the guts or was insane enough to go for some sort of i9-9900KSC with a substantial amount of victim cache, it may just be enough to beat even the Ryzen 3950X in certain workloads (as for instance, ArmA).

There were such rumours a couple of months ago, I came to believe however that is was just a hoax/ prank - AMDs release however has peaked my interest in such a processor again.

 

Anyone else has thoughts on this?

 

And truely nobody's got any numbers in regards to R5 3600Xs performance in A3?

@Groove_C, you got any data on R9 3900X with SMT off?

I am curious as to performance of R7+ chips with SMT off. Since ArmA cannot use this many threads anyways and turning SMT off would, at least in theory, substantially improve overclocking headroom/ turbo boost duration due to (much?) lower temps, it might just be worth a try, for A3 specifically that is.

 

Share this post


Link to post
Share on other sites
2 hours ago, alessiomoreno said:

What do you people think in regards to the Ryzen 3000 processors strong performance over previous gen and even Intel?

Is the improved IPC mostly based in the refined architecture (and new lithography) or because of the immensely increased cache size?

I wonder to what degree the performance improvement stems from which aspect of Matisse.

 

If the major IPC improvement stemmed primarily from their "game cache", then that would let me to believe that perhaps, if Intel had the guts or was insane enough to go for some sort of i9-9900KSC with a substantial amount of victim cache, it may just be enough to beat even the Ryzen 3950X in certain workloads (as for instance, ArmA).

There were such rumours a couple of months ago, I came to believe however that is was just a hoax/ prank - AMDs release however has peaked my interest in such a processor again.

 

Anyone else has thoughts on this?

 

And truely nobody's got any numbers in regards to R5 3600Xs performance in A3?

@Groove_C, you got any data on R9 3900X with SMT off?

I am curious as to performance of R7+ chips with SMT off. Since ArmA cannot use this many threads anyways and turning SMT off would, at least in theory, substantially improve overclocking headroom/ turbo boost duration due to (much?) lower temps, it might just be worth a try, for A3 specifically that is.

 

i posted some  vid of mine with ryzen 3000

Share this post


Link to post
Share on other sites

@KrazyBee

What RAM frequency, timings and quantity, which GPU, SSD or not? In example: 2x8 GB 3200 MHz 14-14-14-34 CR1

Which quality preset was chosen in video settings? Have you changed anything within the chosen quality preset? Was v-sync disabled? Which AI difficulty was seleted under "OPTIONS -> GAME -> DIFFICULTY -> AI LEVEL"?

How have you recorded and at which settings? Browser, Discord, Skype, any monitoring software (Afterburner/PrecisonX/Core Temp/HWiNFO/Aida etc) were running in the background?
Paging file in Windows forced to specific size or managed by system?

(System (right klick on Windows button) -> System Info (right upper corner) -> Advanced system settings (left upper corner) -> Advanced -> Performance -> Settings... -> Advanced -> Change..)

Share this post


Link to post
Share on other sites

I wonder will if Thread Ripper 3 be will even faster with 1:1 RAM at 3600-4000 since it has Quad Channel RAM?  I would buy an 8 Core TR just for the RAM channels if it can be sped up to that?

 

Or just wait for old second hand 7700K and OC to 5+ GHZ and keep running until ARMA 4 comes out.

Share this post


Link to post
Share on other sites

Quad channel doesn't bring anything in ArmA. Already tested by numerous users.

 

Buying Z270 mobo, DDR4 RAM and 7 gen CPU is not worth perf difference and even less -> cost, comapred to your Z97 4790K DDR3 2400.

Share this post


Link to post
Share on other sites

Like one can see here, 1800 MHz Infinity and 2167 (4333) MHz RAM in manual/asynchronous mode is worse than 1800 MHz Infinity and 1800 (3600) MHz RAM in synchronous (1:1) mode, bandwidth + latency.

readbandwidthfclk_mcloxkbb.png memorylatencyfclk_mclcsj3p.png

Share this post


Link to post
Share on other sites

untit2kwk.jpgung7jyv.jpg

 

R7 3700X vs. R7 2700X @ 3600 MHz 16-16-16-36 CR1 both, has 2110 GB vs. 1000 GB L1 cache read/copy bandwidth + twice the size of L3 cache, which is very important for ArmA, Squad and Tarkov

Share this post


Link to post
Share on other sites
2 hours ago, Groove_C said:

@KrazyBee

What RAM frequency, timings and quantity, which GPU, SSD or not? In example: 2x8 GB 3200 MHz 14-14-14-34 CR1

Which quality preset was chosen in video settings? Have you changed anything within the chosen quality preset? Was v-sync disabled? Which AI difficulty was seleted under "OPTIONS -> GAME -> DIFFICULTY -> AI LEVEL"?

How have you recorded and at which settings? Browser, Discord, Skype, any monitoring software (Afterburner/PrecisonX/Core Temp/HWiNFO/Aida etc) were running in the background?
Paging file in Windows forced to specific size or managed by system?

(System (right klick on Windows button) -> System Info (right upper corner) -> Advanced system settings (left upper corner) -> Advanced -> Performance -> Settings... -> Advanced -> Change..)

so here is my specs BTW ARMA is running on SSD and only arma is stored on it

the game setting is only for the KOTH vid

Infantry showcase is all ultra

the coop mission is mostly ultra with 2500 view

YABM=standard setting

------------------------------------------------------

ryzen 3600 stock(but it never boosted to 4.2 as advertised cuz of bios bug or something)

motherboard-gigabyte b450 aorus elite bios f40

amd chipset version 1.07.07

gtx 1070 super clocked

Ram=Crucial Ballistix Sport LT 16GB 2 x 8GB DDR4-3200 PC4-25600 CL16 Dual Channel Desktop Memory Kit BLS2K8G4D32AESB 

------------------------------------------------

Recording

Shadowplay max setting=1080p 60fps

MSI afterburner for the overlay of stats and had firefox open with steam,origin,uplay

-----------------------------------------------------

game settings

Everything ULTRA setting besides

Texture/Object/Terrain=All High

Visibility

Overall=2500

Object=2000

AA=4x

Post processing is all OFF

Resolution=1920X1080(16:10) ratio

Vsync=75hz since my monitor refresh rate is 75

 

 

 

 

 

 

Share this post


Link to post
Share on other sites

@KrazyBee

Once again, paging file in Windows is forced to specific size or managed by system?

To check this, go to ->System (right klick on Windows button) -> System Info (right upper corner) -> Advanced system settings (left upper corner) -> Advanced -> Performance -> Settings... -> Advanced -> Change..

 

Don't record with Shadowplay - make screenshots instead.

 

GTX 1070 is able to run ArmA at  max FPS at ultra 1080p.

I myself always disable post effects, but always played with everything else at ultra 1080p with GTX 1060 6 GB, v-sync off.

But for consistency/accuracy/comparability of YAAB runs/results, you should avoid changing anything within the chosen standard/ultra video quality preset!

You always can set everything to what it's going to be for you 24/7 after YAAB runs/results.

Otherwise it won't be possible to compare if everybody will change some setting here and there.

 

I strongly recommend you to go for more than 3 YAAB runs, if FPS differs to much between runs, for your liking. This way you can drop run(s) with FPS that differed to much compared to other runs.

 

- close all possible background programs (like Firefox, Origin, Uplay and Shadowplay) prior to launching/running ArmA/YAAB (except sound and mouse software), to minimize "false" results due to background programs interference

- leave only Steam and MSI Afterburner, if this is what you're going to use to take a screenshot or assign a button in "Steam -> Settings -> In-Game" to take a screenshot

- disable all mods in the ArmA launcher
- launch ArmA

- under "OPTIONS -> GAME -> DIFFICULTY -> AI LEVEL" select Normal (regular)

- under "OPTIONS -> VIDEO -> DISPLAY -> VSYNC" select Disabled

- under "OPTIONS -> VIDEO -> OVERALL QUALITY" select Standard

- restart ArmA

- run YAAB 3 times and post here average FPS from each of 3 runs, so people with low end/midrange CPUs and GPUs can compare

 

- under "OPTIONS -> VIDEO -> OVERALL QUALITY" select Ultra and restart the game

- don't press "S" at the start of YAAB, otherwise it will modify selected video quality preset by lowering landscape and objects view distance from ultra (3800/2300 m) to standard (1600/1300 m)

- run YAAB 3 times and post here average FPS from each of 3 runs, so people with powerful CPUs and GPUs can compare

 

Whole ArmA community is waiting for your Ryzen YAAB results, as you're among first who has the new Ryzen system!

Share this post


Link to post
Share on other sites

@KrazyBee

Have reproduced exactly all your "kinda ultra" settings, except no browser and no other unnecessary background programs running, except Steam, ArmA and Bandicam to take a screenshot instead of recording.

56.7 FPS avg., 38 min. and 73 max. with my rig (see signature) from summer 2014.

arma3_x642019-07-0922nlkmw.jpg

46.5 FPS avg., 23 min. and 68 max. with your R5 3600 rig with 2x8 GB RAM 3200 MHz 16-18-18 + SSD + GTX 1070

 

"kinda ultra" 1080p preset:
texture/objects/terrain = high

visibility overall/object = 2500/2000 m + pressed S in YAAB, which reduced it once again to just 1600/1300 m

FSAA = 4x

post processes off (bloom, rad. blur, rot. blur, field)

v-sync on

 

So you definitely need to retest like I suggested above.

Also your RAM is dirty cheap and this is where a lot of performance is lost as well.

But for the price of a B450 mobo + R5 3600 + 70-80€ 2x8 GB 3200 MHz 16-18-18 one can't complain!

Share this post


Link to post
Share on other sites
59 minutes ago, Groove_C said:

@KrazyBee

Have reproduced exactly all your "kinda ultra" settings, except no browser and no other unnecessary background programs running, except Steam, ArmA and Bandicam to take a screenshot instead of recording.

56.7 FPS avg., 38 min. and 73 max. with my rig (see signature) from summer 2014.

arma3_x642019-07-0922nlkmw.jpg

46.5 FPS avg., 23 min. and 68 max. with your R5 3600 rig with 2x8 GB RAM 3200 MHz 16-18-18 + SSD + GTX 1070

 

"kinda ultra" 1080p preset
texture/objects/terrain = high

visibility overall/object = 2500/2000 m + pressed S in YAAB, which reduced it once again to just 1600/1300 m

FSAA = 4x

post processes off (bloom, rad. blur, rot. blur, field)

v-sync on

 

So you definitely need to retest like I suggested above.

Also your RAM is dirty cheap and this is where a lot of performance is lost as well.

But for the price of a B450 mobo + R5 3600 + 70-80€ 2x8 GB 3200 MHz 16-18-18 one can't complain!

maybe you are getting better than mine cuz of ur clock speed.what ram is better in ur opinion cuz people in reddit are recommending the one i have for good ram for ryzen.

Share this post


Link to post
Share on other sites
3 minutes ago, Groove_C said:

@KrazyBee

Please run YAAB like suggested and report here.

so tell me which ram is better so i can try go grab the best one

Share this post


Link to post
Share on other sites
Posted (edited)

Why does this always happen to me 😔

I'm giving up

🙄!!!

Edited by alessiomoreno
KrazyBS

Share this post


Link to post
Share on other sites
28 minutes ago, alessiomoreno said:

Why does this always happen to me 😔

I'm giving up

🙄!!!

why are u giving up and on what?

Share this post


Link to post
Share on other sites

G.Skill Trident Z Neo Ryzen 2 & X570 compatible

csm_bildschirmfoto_20vkj12.jpg

 

Tomb Raider 1080p & 1440p RAM scaling @ i7-8700K

*M2 is mode 2 for RAM on Maximus X Hero

Up to 39% more FPS 4133 MHz CL16 vs. 2666 MHz CL15 @1080p
Up to 22% more FPS 4133 MHz CL16 vs. 2666 MHz CL15 @1440p

At same CPU frequency after only RAM fine tuning, without and/or with CPU OC.
So lets continue to affirm/pretend, that RAM frequency and timings do not matter (that much) and continue to recommend/buy dirty cheap RAM and then OC the CPU and complain about poor performance increase 🤔

rottr_ram_stock_vs_oc4mj1g.pngrottr_ram_stock_vs_oc2qkko.png

Share this post


Link to post
Share on other sites

An interesting AMD Zen 2 Memory Performance Scaling with Ryzen 9 3900X review on TECHPOWERUP.

It looks like that 2x 8 GB G.SKILL DDR4 3600 runs well with ZEN2, the reviewer was unable to reach ... "the DDR4-3733 "sweet spot" configuration recommended by AMD because our CPU simply was not able to run at the required 1867 MHz Infinity Fabric—1800 MHz was the highest, which we're covering with the DDR4-3600 test." ... and if I understand well, there are some issue with Infinity Fabric over 4000 MHz.


 

 

Share this post


Link to post
Share on other sites

Nothing interesting in this review. It's even useless, because only 3200 MHz RAM has adequate timings. Other sticks have exaggerated timings. 4000 could have had CL18 or even 17, 3600 could have had CL16 or even 15. Also games video quality settings used are in GPU limit.

All this explains only measurable difference, but no real one.

Really poor. 

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

×