Bioshock Infinite Tested, Benchmarked

[Julio Franco]

Read the full article at:
[newwindow=https://www.techspot.com/review/655-bioshock-infinite-performance/]https://www.techspot.com/review/655-bioshock-infinite-performance/[/newwindow]

Please leave your feedback here.

 

[TomSEA]

Bioshock Infinite also comes with its own benchmarking tool. I run two MSI GTX 660ti's in SLI and at 2560 x 1440 resolution with the ultra setting was getting 60-90 FPS depending on scene.

As the benchmarking shows here in the article, you can get some really nice FPS with a mid-range machine and card.

 

[Rage_3K_Moiz]

Have you guys seen any difference in performance with the alternate post-processing option enabled? The game certainly looks quite different with it to me.

 

[LNCPapa]

I occasionally hit the memory limits of my 2GB 680 SLI setup running 2560x1440 at ultra during the benchmark. It makes for some seriously stuttery moments during the benchmark. Watching the memory usage on my G15 plugin shows it maxed out as well. If I drop it down just one level on the quality settings it runs much smoother and VRAM usage maxes out around 1.6-1.7GB for me.

 

[Jad Chaar]

It is interesting that this game isnt very CPU intensive. Usually you will see a 5-10FPS increase from an increased clock speed from 2.5GHz to 4.5GHz. Also it is odd how OCing the CPU made the minimum FPS drop.

 

[hahahanoobs]

Also it is odd how OCing the CPU made the minimum FPS drop.

Those aren't min framerates, they're frame times, and lower is better.

 

[Jad Chaar]

Those aren't min framerates, they're frame times, and lower is better.

My bad, didnt see that.

 

[hahahanoobs]

My bad, didnt see that.

It's cool.

 

[Guest]

Hmmm, those 6970 results seem strange to me. It performs worse than a 5870?

 

[hammer2085]

Have you guys seen any difference in performance with the alternate post-processing option enabled? The game certainly looks quite different with it to me.

http://www.youtube.com/watch?v=4oIusxRznG4

 

[Guest]

As always, excellent graphical comparison from TechSpot. I'd love seeing this game on a PC, currently own it on the 360 and looks marvelous.

Regarding the game itself, I highly recommend it to everyone. Especially to those who like utopia-themed environment.

 

[Twixtea]

Sorry if this is a dumb question but what causes performance drop when measured frame time in milliseconds compared to the regular FPS benchmark? I'm not really understanding the difference.

 

[Guest]

No GTX 690 ?

 

[Deejay214]

Why no GTX 690 ?

 

[Skidmarksdeluxe]

I'm really not interested in frame rate & time frame numbers. As long as I can have smooth gameplay and an enjoyable gaming experience with most of the eye candy turned up @ 1980x1080 I'm more than satisfied. I'll definitely get this game as soon as it hits bargain basement price. I'm sure my GTX 670 will be up to the task. If not, I want my money back. (for both game & card)

 

[Guest]

Come on, TechSpot. You measure average framerate and then 99th percentile frametimes and you are shocked when there's a big difference.

Steven Walton obviously doesn't understand frametimes testing: it's not an average, it's a threshold under which 99% of the frames were produced. Much like minimum framerate, it's there to give an idea how each card performs in the worst-case scenario in the test. If you were doing average frametimes, or 50th percentile frametimes, it would match the framerate when converted.

 

[slh28]

Come on, TechSpot. You measure average framerate and then 99th percentile frametimes and you are shocked when there's a big difference.

Steven Walton obviously doesn't understand frametimes testing: it's not an average, it's a threshold under which 99% of the frames were produced. Much like minimum framerate, it's there to give an idea how each card performs in the worst-case scenario in the test. If you were doing average frametimes, or 50th percentile frametimes, it would match the framerate when converted.

I believe the point Steve was trying to make is that there's a much bigger fluctuation in fps in this game compared to other games, hence the statement:

It isn’t uncommon to see a gap between frame time performance and average frames per second, but we generally observe a smaller 5 – 10 fps reduction with the former. With BioShock Infinite the margins were massive, anywhere from 10 to 30fps.

 

[amstech]

I see a triple core or two on there but one chip that is still widely used for games that I wouldn't mind seeing, is the 720 X3/740 X3. You don't have to even unlock/overclock it. I would send you mine but its still doing a nice job pushing my 570 in my HTPC for 1080p gaming.

 

[Steve]

Come on, TechSpot. You measure average framerate and then 99th percentile frametimes and you are shocked when there's a big difference.

Steven Walton obviously doesn't understand frametimes testing: it's not an average, it's a threshold under which 99% of the frames were produced. Much like minimum framerate, it's there to give an idea how each card performs in the worst-case scenario in the test. If you were doing average frametimes, or 50th percentile frametimes, it would match the framerate when converted.

I believe the point Steve was trying to make is that there's a much bigger fluctuation in fps in this game compared to other games, hence the statement:

It isn?t uncommon to see a gap between frame time performance and average frames per second, but we generally observe a smaller 5 ? 10 fps reduction with the former. With BioShock Infinite the margins were massive, anywhere from 10 to 30fps.

slh28 you are correct, thank you.

It shouldn't be a surprise though as I said virtually that in the article.

From the conclusion...
"It isn’t uncommon to see a gap between frame time performance and average frames per second, but we generally observe a smaller 5 – 10 fps reduction with the former. With BioShock Infinite the margins were massive, anywhere from 10 to 30fps."

 

[Rage_3K_Moiz]

I've already seen it in action. I'm just wondering if the performance hit is higher on some cards compared to others.

 

[brkbeatjunkie]

In your screenshot comparison high settings show way more AA around edges ie trees and leaves compared to ultra which is quite aliased IMO

Why does ultra have inferior/less anti aliasing??

 

[Guest]

This is extremely biased towards Nvidia, they intentionally left post processing to "normal" which is DX10 instead of maxing it out to "Ultra" which is DX11 because Post-Processing Ultra is significantly faster on AMD cards, shame on you techspot !
This is not the first time that you intentionally manipulated the settings in favor of Nvidia .

 

[Steve]

I guess we have been caught red handed once again, or should I say green handed.

 

[Jad Chaar]

I've already seen it in action. I'm just wondering if the performance hit is higher on some cards compared to others.

Well, PostProcessing has been used for a while by companies like nVidia and AMD. So as the tech progresses and drivers mature, the hit will be more and more minor.