Crysis 3 Tested, Benchmarked

Julio Franco

Posts: 9,097   +2,048
Staff member
Read the full article at:
[newwindow=https://www.techspot.com/review/642-crysis-3-performance/]https://www.techspot.com/review/642-crysis-3-performance/[/newwindow]

Please leave your feedback here.
 
I think that this is the first game that justifies me getting a second 670 for myself. Curse you, o mighty CryEngine 3.
 
Bit disappointed with AMD's performance, let's hope for a quick driver updatae!
 
Yeah pretty pedestrian from AMD, they seriously need to get cracking on their drivers. This review doesn't even take into account their poor frame latency which may degrade performance even further.
 
As for me... I'll wait till the game hits basement bargain. I have no intention whatsoever of spending serious dough on upgrades for a single (not so hot) game.
 
Holy waaaaaaaaahh. This game looks straight up sexy. Now I'm debating whether to buy three identical 1080p monitors (my current one is too expensive to buy two more) or stick with my 1920x1200 and get a second 680. I def can't afford both.
 
CPU benchmarks on Medium only? Come on, you guys said yourselves that the LOD and draw distance was too low on Medium, and those are the two most CPU-intensive parts. At least test the CPUs on high settings, if not Very High. Test at a lower resolution to avoid GPU-bound results.
 
Bf3 plays best on amd cards which has better graphics than crysis 3....screenies below
 

Attachments

  • bf3 1.PNG
    bf3 1.PNG
    44 KB · Views: 3
  • bf3 2.PNG
    bf3 2.PNG
    45.8 KB · Views: 3
The numbers throughout the review look low to me, Ive seen lots of other benches that suggest different results. Also it is all 16:10 — we could use a nice 1080p column to those figures. :)

I've seen on youtube a video of a guy running 1080p at very high quality but no AA, and he gets in the mid 40s fps with a GTX 460 — that's an old gpu, it only pushes ~900 gflop/s (compared to 2.5 tflop/s for an overclocked 660TI).

FXAA only uses ~2% overhead because it smooths edges on the final 2d output rather than true AA.. to me, I actually prefer it to MSAAx2 and think it is about the same as MSAAx4. MSAAx8 is going to ruin your graphics performance, even MSAAx4 halves the fps.
 
What? No i5-3570k or i5-2500k listed? I would have guessed those are the two most popular cpus out there!
 
This setup isn't ideal for everybody, but the game runs great on CF 7950 OC (Gigabyte model.) It cost $600.00, but the current promotion gets you free copies of Crysis 3 and Tomb Raider. As I was buying these anyway, I tell myself these cards "only" cost $500.00. I sold off my old card for a $100.00 as well, further reducing the total cost.

I have not been experiencing the notorious micro-stutters and performance is terrific (2600K @ 5GHZ.) I have only crashed once, post-patch, and yesterday's beta drivers supposedly fix a crash due to the patch...
 
Bf3 plays best on amd cards which has better graphics than crysis 3....screenies below

I dunno, I see the benches and everything but I have a friend who has duel 6950 I believe and he actually sold them for Nvidia 670's as Battlefield refused to stop stuttering or crashing, he has an x58 setup, 920, 8GB RAM and 900watt PSU.

Once he got the Nvidia's it stopped, incidentally, I know a few people who had trouble with crossfire setups and Battlefield, its strange though because as single cards, your right, ATI's play battlefield just as well if not better?

Guess it must be driver issues, he did update the drivers, in fact he completely re-installed windows xD
 
Your CPU section is insufficient, and I haven't experienced any problems, whatsoever, running this game at 1080p, Very High, FXAA using two 670s in SLI (which are enough to maintain 60fps at these settings at all points in the game that aren't CPU-limited, which unfortunately is a good chunk of the game), on 314.07. Rerun your CPU tests at the lowest possible resolution, and display the differences between meduium and very high at 1080p...dropping to medium drops fps from 45 to 30 in some sections of the game, where CPU-limiting is clearly the case (GPU utilization at ~50% and fps below 60 equals CPU limiting). Additionally, I've not seen CPU utlization over 70%; not sure as though this game makes the best use of quads, either.
 
Did I miss something because in beta I was playing on high comfortably with my 7870 crossfireX, was sitting on about 75 the hole time
 
I'd wager that number is within 1-2 fps of the 3770K @ 4.5 as is the 2600K @ 4.5
 
Really lol , I guess you have not seen crysis1 and 2, this one blows compared to the 2 and that blows compared to 1... whats the point even of good grafix if 90% plays in the dark... worst crysis yet
 
I've got a 920 at 4ghz and a gtx 480 o/c'd and it bends this game over and spanks it silly. It seems that if you have a high enough bandwidth graphics card a really good cpu clock speed is more important.
 
I've got a 920 at 4ghz and a gtx 480 o/c'd and it bends this game over and spanks it silly. It seems that if you have a high enough bandwidth graphics card a really good cpu clock speed is more important.
what settings?
 
"Assuming you clear that hurdle, you'll likely want dual GTX 680s or HD 7970 GHz Editions to play on very high at 1920x1200 or more."

But...but...but people told me multi-GPU setups @ 19x10 is stupid! You HAVE to have a resolution of 25x14 or higher!!! :|

+1
Their have been quite a few games @ 19x10 that have brought single GPU flagships to their knees over the years that could benefit from another card. If one card could always do the job at all single monitor resolutions, we wouldn't have had multi-gpu technology years before multi-monitor gaming setups came out.
 
Back