DCS A-10C final benchmarks

[RSharpe]

Now that the final has been released, has anyone been able to do some benchmarks comparing the performance of the final to say, Beta 4?

 

Thanks.

[hassata]

http://forums.eagle.ru/showpost.php?p=1112313&postcount=3

[RSharpe]

Thanks

[GregP]

I don't know -- I'm a bit hesitant to present this, because I feel like there must be some obvious thing I've overlooked and screwed up -- but I did a comparison last night between Beta 4 and the release version and I got a big DROP in framerate, DESPITE the fact that I have overclocked my processor even more now than I had when I did the Beta 4 test. My setup was/is:

 

WAS: Beta 4 64-bit version, i5-2500K @ 4.6 GHz, 8 GB DDR3, GTX 470 at 680/1360

IS NOW: Release 64-bit version, i5-2500K @ 5.0 GHz, 8 GB DDR3, GTX 470 at 680/1360

 

I ran two different tests (one a simple sitting-on-the-runway takeoff mission, the other during the LITENING Pod training mission) using a 3-screen SoftTH setup at 4630x1080 and got these results:

 

WAS: 52 and 57

IS NOW: 44 and 47

 

That's a disturbing 15-18% drop in framerate since Beta 4. As I was entirely expecting a framerate jump with the release version, I'm really confused here.

[monsterZER0]

Holy CRAP!!! You have your 2500K running at 5.0GHZ!? What kind of cooling and temps are you running? God I can't wait to get mine. Just waiting on the B3 Stepping mobos to hit the market...

 

That IS weird that you took a hit on frames, too... I definitely am getting much better performance than the beta.

Edited February 24, 2011 by monsterZER0

[alont]

Depending on the cpu, motherboard and memory, beyond a certain frequency one might see significant DROPS in games and other benchmark software.

Try lowering the clocks a little bit (5GHZ? seriously?)

[GregP]

Holy CRAP!!! You have your 2500K running at 5.0GHZ!? What kind of cooling and temps are you running?

 

I use a Sunbeamtech Core Contact Freezer w/120mm fan -- not top of the line, was using it for 2 years on my old QX6800 -- but DCS under load, in my testing so far, doesn't push the temps above 70, and idle in mid-30's.

 

Depending on the cpu, motherboard and memory, beyond a certain frequency one might see significant DROPS in games and other benchmark software.

Try lowering the clocks a little bit (5GHZ? seriously?)

 

Yeah, my next test was going to be dropping back down to 4.6 to see if I could replicate, or even just get closer to, the FPS numbers I was seeing in Beta 4. And why 5.0? Simply because the headroom is there, and it's fun to say your machine is running at 5 gigahertz :) With these Sandy Bridge chips o/c'ing is effortless, and I got at least a 10% FPS gain when going from the stock 3.3 up to 4.6, so I assumed going up to 5.0 would at least not hurt things. But perhaps I've pushed it too far. I guess I won't know til I try 4.6 again.

[Pyroflash]

I don't like the whole Sandy Bridge line because they gimped the RAM by shoving it back down to dual channel, and the new chipsets that Intel released for them have built in corruptions in their SATA controllers.

 

Though if they didn't do that stuff I would totally have bought one by now for the simple fact that you can get them running at 4.0GHz without any liquid cooling.'

 

Odd that you would be experiencing drops. Did you increase your QPI link speed with your BCLK? I know sometimes if it is too far out of sync, it can pose some issues as the CPU is processing more data than it can transmit, thereby bottlenecking it.

 

Also: RMA your motherboard quickly, the ASUS P8 line of motherboards contains one of the two corrupt Intel chipsets(H67 and P67) that Intel have discontinued for manufacturing and have started to implement a redesign of the processes in order to weed out the manufacturer defect. If you don't RMA it, there is a good chance that your chipset's SATA controllers will fail, leading to a possible malfunction of connected devices(HDD, Optical, SSD). http://www.anandtech.com/show/4142/intel-discovers-bug-in-6series-chipset-begins-recall

 

They say late February, but I don't have any exact dates. For all I know it could be fixed already, but you would have to check with Intel or a corroborating secondary news source on that.

Edited February 24, 2011 by Pyroflash
Link

[GregP]

I don't like the whole Sandy Bridge line because they gimped the RAM by shoving it back down to dual channel

 

I don't know -- unless you're specifically trying to run with 3 sticks of RAM (for example if that's what you already owned and didn't want to buy new sticks), does going from triple-channel to dual-channel really affect anything? With this platform, it doesn't seem like RAM is much of a bottleneck.

 

Odd that you would be experiencing drops. Did you increase your QPI link speed with your BCLK? I know sometimes if it is too far out of sync, it can pose some issues as the CPU is processing more data than it can transmit, thereby bottlenecking it.

 

With Sandy Bridge there is no de-linking of the FSB from anything else, so I don't think there's any way to play with things like QPI. All you have control over is FSB -- which pretty much has no headroom -- and multiplier.

 

Also: RMA your motherboard quickly, the ASUS P8 line of motherboards contains one of the two corrupt Intel chipsets(H67 and P67) that Intel have discontinued for manufacturing and have started to implement a redesign of the processes in order to weed out the manufacturer defect. If you don't RMA it, there is a good chance that your chipset's SATA controllers will fail, leading to a possible malfunction of connected devices(HDD, Optical, SSD). http://www.anandtech.com/show/4142/intel-discovers-bug-in-6series-chipset-begins-recall

 

Right, I'm waiting on Microcenter to begin stocking the fixed boards, at which point I'll do my exchange there. Either way it doesn't affect me because I just switched my SATA devices over to the 4 ports that are not affected by this defect.

[GregP]

So I dialed my overclock back down to 4.6 last night, to match what I was running with Beta4, and I get the exact same framerates I got at 5.0 with the release version: 44 and 47. Apparently, something has changed in the release version that has negatively impacted framerates on my system. I'm stumped!

[hassata]

did you clean install?

[zakobi]

Are you sure you have the exact same graphics options?

 

Not thinking on menu only, but only suggested optimization for B4 too.

[GregP]

Yup, clean install after uninstalling Beta4. And I have no graphics options set outside the game, so if I'm doing same in-game then I'm stumped as to what's different.

[EtherealN]

Do clean out the User\XXXX\SavedGames\DCS A-10C Warthog\ folders between runs? There are files in there that both Beta4 and release would use, and having them from the "wrong" version might impact your results. (Though I haven't tested that myself.)

 

Did you run the tests as track replays? If so, could you attach them so I can give it a go?