Framerate, Hardware, Opinions?

[polygonpusher]

Given the hardware below, I still am not seeing smooth frame rates in the "Death Valley" mission. Is my experience typical of those with similar hardware?

 

BS clean install, patched to v1.1

 

Visibility: High

Scenes: High

Civ Traffic: Yes

Water: Med

Shadows: Active Planar only

Cockpit res: 1024

 

Screen Res: 1920x1280

AA: 2

AF:off

 

Hardware:

i7 930 @ 3.4 Ghz

6 GB Ram

Single Ati 5870 2 Gb ver

Windows 7 Home Basic

 

I was really expecting higher frame rates from BS given this rig. Were my expectations too high?

 

Your thoughts? Experiences?

 

Thank you so much,

 

D.

[EtherealN]

Patch up to 1.0.2 (I assume you mean you have 1.0.1 installed). New sound engine frees up additional resources for the main simulation thread.

 

You can boost up your AA and AF to 16 on both if you like with that graphics card. My computer with a lowly 9800GTX+ can do that, since there's enough GPU headroom left.

 

Define "smooth framerates" though. Some "stutters" can happen due to HDD lag, and there's really nothing your CPU can do to prevent being affected by a slow hard drive. This is usually related to loading sounds.

Also, are your mirrors on?

[polygonpusher]

Good point EtherealN! I'm sorry that it has taken me so long to reply, the dev work on 1.0.2 has made a wonderful improvement on frame rates.

 

It's nice to see a developer continuing to support it's product. I'm hoping that the next patch will focus on the AI wingmen behavioral issues that have been cropping up.

 

Thanks again DCS team!

[213]

my system's pretty new:

i7 930

3gb ram

9800gtx+

 

i'm using mostly medium range settings, with the fancy stuff(aa/anistrophic, etc) turned off. the game runs smoothly for the most part(70+ fps), however, it seems whenever there are battles involving cannons, the game slows to a crawl. also, cities lag. a lot.

 

i've learned that turning scene and drawing distance down will make the buildings pop up at a shorter distance, but cities still fare pretty badly in terms of fps.

 

i have set an experiment where i used the editor to create 2 scenarios. one involving lots of gunpods, cannons, etc, and the other missile oriented. the difference is about 40fps, which is astonishing...sometimes it'll even drop to single digits when ships starts spewing shells at eachother.

 

i think the ballistic calculation, with so many cannon rounds flying everywhere is costing performance. also, do they have to render the shadow on every single bullet? i think the ka 50's cannon uses no shadow, and it runs much better. maybe it's the shadow's fault.

[sofie_59]

I would also like some advice on how to get better fps!

My system:

Windows 7 64

i7 930

6gb ram

ssd 120 gb

460 gtx 1gb OC Sli

[Frazer]

You could do a walk through my "DCS Black Shark Tweak Guide Alpha Version" to understand the areas where you can win some FPS.

 

But I'm surprised that even on these high end systems people still suffer from low FPS. I think ED should stop implenting new eye-candies and take a serious look at how to improve the game engine.

 

Tweaking can be fun for some, but comeon, it should not be needed on high end systems.

[sofie_59]

Ok thanks Frazer i will try your tweak guide

 

And i get high 20+ on my rig

I will try lower water and see what happen

[Bucic]

C2D 3.7 GHz, GF9600GT 512 MB*, 4 GB RAM

 

in-game settings high except water (lower than GUI lowest, as per Frazer's guide) and shadows (active planar) + some Frazer's tweaks

driver settings: 8x AA, 8x AF, max quality texture filtering, vsync on

30-75 FPS

 

Additional info:

720MHz core; 1008MHz memory (2016MHz effective); 1800MHz shader

(factory overclocked)

1280x1024 fullscr

no additional monitors

Win7 x64

driver version - insignificant when using a driver from 2010

Edited October 22, 2010 by Bucic

[RightStuff]

I'm looking for an recommendation for an upgrade to get better FPS.

Actually I have only ~20-25 with some eye-candy settings, which I won't miss...

 

My system:

AMD X2 64 4800+ (2,5 GHz - no OC)

Asus M2N-E

2 GB RAM DDR800 (DualChannel)

GeForce 9600 GT, 512 MB @720 MHz

Win XP SP3

BS 1.0.2 playing at 1680x1050 (less is no-go)

 

Should I first upgrade the CPU, the GPU or both - perhaps incl. mainboard and RAM - I hope, I don't have to... :music_whistling:

 

Thx in advance!

[Bucic]

I'm looking for an recommendation for an upgrade to get better FPS.

Actually I have only ~20-25 with some eye-candy settings, which I won't miss...

 

My system:

AMD X2 64 4800+ (2,5 GHz - no OC)

Asus M2N-E

2 GB RAM DDR800 (DualChannel)

GeForce 9600 GT, 512 MB @720 MHz

Win XP SP3

BS 1.0.2 playing at 1680x1050 (less is no-go)

 

Should I first upgrade the CPU, the GPU or both - perhaps incl. mainboard and RAM - I hope, I don't have to... :music_whistling:

 

Thx in advance!

See my post above. No additional comments needed.

[RightStuff]

See my post above. No additional comments needed.

Sorry, but that answer is a little short, as the description of your rig is not really detailed:

How much RAM has your GT9600?

What clocks?

Which driver?

What's your game resolution?

What's your OS?

Playing in windowed mode or fullscreen?

Additional monitor(s)?

 

Meanwhile I got a AMD X4 9850 (Quadcore @2.5 GHz - big balls :smilewink:) but there wasn't the big advance in FPS as I had expected. The CPU is almost idling during game (20-30%) and graphics settings are kicked up to a reasonable value. Despite the new CPU that it's still around ~25-30.

 

So, the GT9600 seems now to be the bottleneck, hm?

Any other comments or helpful hints?

(Except getting a new graphics card as well... :cry:)

 

And yes: tweak guide was read and used...

[Bucic]

Sorry, but that answer is a little short, as the description of your rig is not really detailed:

How much RAM has your GT9600?

What clocks?

Which driver?

What's your game resolution?

What's your OS?

Playing in windowed mode or fullscreen?

Additional monitor(s)?

I hear you. Original post updated. However nothing except GC version and monitor resolution was needed evaluate my specs.

 

So, the GT9600 seems now to be the bottleneck, hm?

:shocking:

Are you kidding me? I thought that this 'puzzle':

GF9600GT + 2,5 GHz CPU -> crappy FPS

GF9600GT + 3,75 GHz CPU -> HIGH FPS

was trivial.

 

No, GF is not your bottleneck unless you use screwed up (original) flare trail effects or water on other setting than low.

Edited October 22, 2010 by Bucic

[RightStuff]

I hear you. Original post updated. However nothing except GC version and monitor resolution was needed evaluate my specs.

Well, first I thought, the different resolution (1280 vs. 1650) was the solution for getting more FPS, unfortunaltely it wasn't...

 

Are you kidding me? I thought that this 'puzzle':

GF9600GT + 2,5 GHz CPU -> crappy FPS

GF9600GT + 3,75 GHz CPU -> HIGH FPS

was trivial.

No time for kidding.

 

You noticed? Your CPU is Doublecore - Mine is Quadcore.

What's the exact brand of your CPU? Like to take a look at a performance-chart.

 

Btw.: With my old CPU DCS:BS was running at ~70% (avg of all cores). Never saw 100%. With the new Quadcore the CPU is "idling" at ~50% (avg of all cores). If my system isn't totally screwed up I would expect much more FPS - especially as my graphics card should be able to produce them.

 

Hmm... Thinking now about a re-install of my WinXP-SP3 - But not in the next three weeks. Or should I move up to Win7?

[Bucic]

You can trust me here. I did many FC/DCS benchmarks. I'm being immodest here but opposing what I say about DCS performance you're wasting your time ;)

 

Well, first I thought, the different resolution (1280 vs. 1650) was the solution for getting more FPS, unfortunaltely it wasn't...

Are you saying that you were hoping to get more FPS in

You will get more FPS by lowering resolution only when previously your graphics card was the bottle neck.

 

You noticed? Your CPU is Doublecore - Mine is Quadcore.

What's the exact brand of your CPU? Like to take a look at a performance-chart.

Quad cores give negligible performance increase in DCS. See the FAQ. My CPU is C2D 7300 2,5->3,75 GHz.

 

Btw.: With my old CPU DCS:BS was running at ~70% (avg of all cores). Never saw 100%. With the new Quadcore the CPU is "idling" at ~50% (avg of all cores). If my system isn't totally screwed up I would expect much more FPS - especially as my graphics card should be able to produce them.

You'd have to provide more info for me to be able to figure out something. First of all quit using the eye method. Use a diverse action track file, 1 minute long or so, and record performance data using FRAPS or something else. Make graphs. Graphs will tell you more than a crystal ball.

 

Hmm... Thinking now about a re-install of my WinXP-SP3 - But not in the next three weeks.

Quick system swap for testing:

1. Make a backup image of your current system partition using Acronis True Image.

2. Install new system and perform your tests.

3. If no increase => restore your old system from the image.

 

Or should I move up to Win7?

:thumbup:

 

Edit:

I can't seem to find the topics where I described my XP vs Vista benchmarks so here are just the graphs. There were even greater advantage in case of Win 7. You may as well take the opportunity to switch to Win 7 x64.

Edited October 25, 2010 by Bucic

[RightStuff]

Bucic, first of all I'd like to thank you for your detailed answers!

 

But - long story, short - testing (even with imaging partitions) is time-consuming and I definitely like more learning and flying the thingie than to test this and try that... :music_whistling:

 

So: I'll definitely move up to Win7 in a few weeks.

Case closed... Thx! :thumbup:

[Bucic]

You're welcome but I haven't posted the graphs yet because I'm on so crappy network that I can't load neither photobucket nor mediafire!

 

Edit:

 

 

 

Notes:

 

Done on vanilla 1.0

With 1.0.2 + the "affinity trick" kicks in automatically.

Win 7 may provide even greater performance increase compared to Vista.

Edited October 26, 2010 by Bucic

[EtherealN]

Just to re-emphasize something here, since it does apply to many fields even outside of this specific simulator:

 

An X GHz Quad-Core, does have a theoretical max computation power twice as an X GHz Dual-Core, assuming they are both of the same architecture. (An X GHz C2Q is however weaker than an X GHz i7, due to specifics in the compute pipeline and such.) But do remember that I placed a rediculous amount of emphasis on "theoretical". The best way to think of why this is important is to turn on the verbose boot option on your windows install and note where it loads a multi-processor kernel. Those who have been around for a while will remember how awesomely cool it was back in the nineties with those high-performance workstations with multiple processors in them, but might also remember how pretty much nothing on the consumer market was made to take advantage of a multi-processor setup.

 

A dual/quad/hexa/octa-core processor is really the same thing: it's several processors (as far as the OS is concerned) but they just happen to be on the same physical silicon die. To make use of the extra cores in the same process, the process needs to be specifically programmed for this.

 

So for example, for many games a 3GHz DualCore (2x3=6GHz total, so to speak, even though that's not quite right to say either) will outperform a 2GHz QuadCore (2x4=8) simply because the game cannot spread it's thread(s) efficiently across enough cores to even try using all that the Quad has to offer. If the application is made to run in two threads only, it will get a total of 6GHz on the Dual (minus OS etc), but only 4GHz on the Quad. (Though there the OS and background will end up on the "free" cores.)

 

This is relevant to DCS:BS in that DCS effectively runs on two threads: the sound engine has it's own thread, and "the rest" is in another thread. There's some extra threading gains from Vista and 7 through their better multicore handling and Dx10/11 features native to those operating systems, but it should be said that we are effectively expecting to "saturate" only one and a half core with DCS:BS. So for DCS:BS you are better off with a really high-speed dual-core than a moderate-speed QuadCore, even if the "brute force" of the quad is greater.

 

Sadly, creating a truly "multithreaded" application is extremely complicated, so this problem in fully utilizing quad/many-core processors will take a while to fix. My understanding is that it'll be a gradual process such that as each underlying component in the technology is updated or replaced, it will be investigated to see if it can be made to run in it's own thread. The new sound engine is an example of this.

Edited November 5, 2010 by EtherealN
Correcting terminology - I said chipset where I should have said architecture, way different things.