Crysis 3 CPU performance test

[bononos]

Anyone notice that that was tested with gk 110?

I did, I'd like to see benchmarks with GK104s to see what sort of bottlenecking the vast majority of gamers will face.

 

[Gikaseixas]

I call 2 FPS a large margin partially because the difference in cost, so I should rephrased that as "a large margin in performance/$" as some would consider a $300+ CPU losing to a competitor's $200 chip by any margin an embarrassing defeat.

That makes the FX 8350 stand out. If this trend continous the FX can become the gamers choice once again.

 

[cytg111]

Arrrggh .. wanted to play this game .. but aging C9450 & gtx580 (1.5G edition) .. even if reducing screen to say 720p.. might still be out of steam!

 

[Gikaseixas]

gtx580

Your card is still a beast. Grab one of those MC deals if you're close to one (say a 2500k or FX 6300) and you'll be able to play @720... i think

 

[dqniel]

AMD's finally competitive in a game at the high-end. It's been a while.

 

[cytg111]

Your card is still a beast. Grab one of those MC deals if you're close to one (say a 2500k or FX 6300) and you'll be able to play @720... i think

goddamnit, if hadnt been delayed haswell would almost be upon us .. coulda waited for that .. think im gonna run with my chances here..

 

[OS]

I noticed something else very interesting about the chart, the more cores/threaded processors have less percentage change between average and min.

 

[Deders]

Depends on the game, the i5-760 may be working at 90-100% when the Phenom X6 1100T only at 40-50%. So you could OC the Phenom and not the i5 with the default cooler.

My i5 750 uses about 75% max when clocked to 3800, I don't think I'd notice much difference if I left it a stock tbh but then my 560TI is the limiting factor. Can still play it reasonable well on with everything Very high and medium SMAA.

This is definitely the most CPU intensive multicore game I've seen so far.

 

[crashtech]

Is there any word on how many threads this game can actually bring to bear? More than 4, pretty obviously, but how many more?

 

[SPBHM]

it seems like there is a lot of variation,

under heavier action:

http://pclab.pl/art52489-9.html

 

[Mallibu]

it seems like there is a lot of variation,

under heavier action:

http://pclab.pl/art52489-9.html

And boom, there goes the AMD "We're 2 frames ahead in 1 of 100 games" party down.D:

 

[ShintaiDK]

And the Post Human one:

 

[KompuKare]

Sort of to be expected I guess. But what my take from Crysis 3 CPU scaling and other similar scaling was not FX8350 vs i5-3570K / i7-3770K, but rather that AMD's FX6300 and even FX4300 perform well against i3 especially since i3 and below do not overclock.

I know, AT forums is all full of enthusiast with big pockets and everyone buys top-end Intel CPUs plus top-end GPUs here, but for people on a budget the lower end FX's processors do have at least a niche. Problem is, the halo effect is an incredible power marketing tool and people tend to jump to conclusions: just because i7-3930K or some Xeon is the fastest doesn't make a Celeron good for gaming (although the IB Celerons are very good for power usage), ditto Titan's crown shouldn't say anything about HD7850 vs GTX660 etc.

 

[guskline]

In the for whatever it's worth department, here's a snippet of Fraps data I saved last night from playing first person Crysis 3 at the end of the first scene:
Frames, Time (ms), Min, Max, Avg
145frames
4836ms
29Min
31Max
29.983Average

That is on my 8350 @4.6Ghz with a EVGA670 FTW and settings on High ( 3rd machine in my sig below).
Game is demanding but fun and very playable at those rates..

 

[inf64]

Guskline can you bench with your 2nd machine in the sig?

 

[UaVaj]

it seems like there is a lot of variation,

under heavier action:

http://pclab.pl/art52489-9.html

good find.

so it was the gpu that was limited in "welcome to the jungle".

"root of all evil" pushed the cpu and amd falls behind again. so much for 8 core.

party over.

 

[guskline]

Guskline can you bench with your 2nd machine in the sig?

Will do

 

[Abwx]

good find.
"root of all evil" pushed the cpu and amd falls behind again. so much for 8 core.

party over.

Using only 4 cores is "CPU pushed"? Not for a FX...

 

[ShintaiDK]

More CPUs compared in the 2 other benches.

 

[UaVaj]

Using only 4 cores is "CPU pushed"? Not for a FX...

according to the graph.

4core - i5 or i7 at 4.5ghz = 45fps
8core - fx8350 at 4.7ghz = 33fps

seriously party OVER!

 

[inf64]

"8 core" is actually 8 thread in case of FX8xxx. It has 4 FP "units",same as i5 and i7 . Both designs can run two threads per fp unit(or core in intel's case).

 

[guskline]

OK for rig 2 below I ran a shorter segment. Here3 is the data.
Frames 54
Time 1687 ms
Min 31
Max 33
Avg 32.009

That's the EXACT same video card with settings on high and 2500k @4.5Ghz. BTW this was played on an Achieve Shimian 27" at 2560x1440

 

[inf64]

Much appreciated Guskline . So the difference is exactly 6% ,hardly relevant.

 

[guskline]

Why post findings with a 1024x768 resolution?

 

[UaVaj]

"8 core" is actually 8 thread in case of FX8xxx. It has 4 FP "units",same as i5 and i7 . Both designs can run two threads per fp unit(or core in intel's case).

amd themselves says. "eight native core"
http://www.amd.com/us/products/desktop/processors/amdfx/Pages/amdfx.aspx

cpu photo clearly shows 8 core
http://www.bit-tech.net/hardware/2012/11/06/amd-fx-8350-review/1

newegg say eight-core
http://www.newegg.com/Product/Product.aspx?Item=N82E16819113284

:'(