**Mafia 2 Benchmark thread**

[Rroff]

I should say this isn't a great benchmark for real comparisions - the demo seems to use 1000/whole ms timeslice to measure framerates so you end up with values being rounded to:

1000/2 = 500
1000/3 = 333.3
1000/4 = 250
1000/5 = 200
...
1000/11 = 90.9
1000/12 = 83.3

and so on - you only get any kinda accuracy at low framerates. Which means 2 people could be getting results 5fps apart but the benchmark showing them over 20fps apart.

Not a great testament to the programming team as that was the kinda school boy error I made when learning to program video game engines.

 

[shuzzbutt]

If anyones interested - same PC, same settings, etc. top is GTX260 SLI bottom is GTX470 + GTX260 for PhysX. Not a huge difference but framerate is more consistant and the game a bit smoother - also I can turn up the AA ingame (using exe rename/control panel force) and still get 60+ fps on the GTX470 whereas on the GTX260 SLI it would drop below 40fps.

All settings are maxed including PhysX at 1680x1050.

sorry to be a pain but have you run the benchmark with a single gtx 260 rendering whilst the other is used for physx rather than in sli mode?

 

[Rroff]

Yeah but that was before the PhysX fix - it didn't make much difference to the framerate but was marginally smoother.

 

[Biglunn]

All settings maxed. PhysX off.

With physX on, the FPS takes an absolute hammering.

 

[Leeroy123]

Out of interest, whats making you sell x2 gtx 295's for a single gtx 480? was it dx11 support or is quad sli just not that good? Just intrigued to know cos I'd have thought thats a powerful enough gpu setup for anything thats out now or will be anytime soon?

Don't get me wrong quad sli with the 295's is superb, they rip through any games with 16x aa and everything maxed at my resolution. I play on 1920 x 1080. The only single thing making me even consider changing my cards is the dx11 for the games. Im always about pushing that extra bit for gaming. Id love to be able to go 32x instead of 16. Am I wrong to change now? do you think i should wait.

 

[arc@css]

^^Stay with what you've got. I'll bet there's nothing that really taxes that, save for something like GTA4 / Metro.

 

[Bizarre]

I should say this isn't a great benchmark for real comparisions - the demo seems to use 1000/whole ms timeslice to measure framerates so you end up with values being rounded to:

1000/2 = 500
1000/3 = 333.3
1000/4 = 250
1000/5 = 200
...
1000/11 = 90.9
1000/12 = 83.3

and so on - you only get any kinda accuracy at low framerates. Which means 2 people could be getting results 5fps apart but the benchmark showing them over 20fps apart.

Not a great testament to the programming team as that was the kinda school boy error I made when learning to program video game engines.

I noticed that too. If you randomly pause the video below you can see that fraps at times is recording the framerate 10+ frames higher than the actual benchmark.

Edit: removed vid - check post below.

 

[Bizarre]

Linked wrong vid:/ Here's a better example.

http://www.youtube.com/watch?v=uSbj1zrW45A&feature=player_embedded

 

[jmc007]

You can set it up to show both GPUs. The advantage of a GPU dedicated to PhysX hardly seems worth the cost, heat and power tho - it is a bit smoother playing and framerate is a bit higher but not massively so - maybe 20% higher average fps at best.

Even when using an ATI card? Bloomin' just bought a B grade gts 250 from OcUK for physx.

 

[Rroff]

If your using ATI for rendering then its another story assuming you can get PhysX to work with that setup.

 

[jmc007]

T.G. for that. Kind of looking forward to the challenge of getting it working, (Hopefully.).
b.t.w. Sig changed from 9800gtx+ to gts250 for physx as, well to not cut a long stort, I gambled on buying the gtx of ebay though it was faulty; believing I would be lucky with the oven trick.
Low and behold the thing just started smoking and sparked to a flame on one of the transistors.
Hence why I opted for the gts from OcUK yersterday.
Sig now changed.

 

[jmc007]

Pre, dedicated physx benchmarks.
5850 @ 1006/1300
Physx off.

Physx medium.

 

[jmc007]

Wow, wondered why some people were getting 250 Maximum FPS and decided to try Vsync off; really didn't think it would make such a difference. Pretty darn stupido thinking about it now.

Same settings as the first try above, except Vysnc off too.

 

[Razor Time]

Interesting review on Mafia 2 and different hardware setups.

http://www.hardwarecanucks.com/foru...tion-nvidia-s-graphics-plus-technologies.html

 

[RavenXXX2]

lol suck up much, physx in mafia 2 is a bad joke, obviously they have no clue.

 

[shuzzbutt]

lol suck up much, physx in mafia 2 is a bad joke, obviously they have no clue.

I've only played the demo but have to say that I'd rather play it with physx on rather than off- to such an extent that I will be upgrading my gpu setup to enjoy it without affecting performance too much

 

[kmufc77]

My copy of Mafia 2 was despatched today from a well known jersey etailer

 

[Vince]

All high AA on Physx off.

5850 @ stock

More of the same but at 950 core, although I am not sure AA is working in the benchmark. Mind you I haven't bothered updating since 10.2 so it might be time.

Either way overclocking the core seems to give a decent return in minimum and average framerates

 

[Sone]

all high with PhysX and AA off @ 1650 x 1050

3850 X2 47fps
2900XT 33fps

Makes it much more playable, just waiting till tomorrow now

(rest of PC E8500 @ 4ghz, 6gb, IP35)

 

[blackninja]

AA on .. all else on or high:

AA off .. all else on or high:

AA on .. all else on or high: