BANANA Benchmark Results Comparison

I cant imagine a graphics card really making that much difference? There might be something wrong if a newer more powerful is lowering your score, or changing it at all really.

 

Exactly Some of the vehicle Processing is part of the Graphics card (obviously to render the cars) But to Process more Cars that uses up CPU power

 

Um, no. GPU should have no effect on the BANANAbench. More cars would stress the GPU more too in the game. Seriously, please stop spreading false information.

 

BananaBench only tests the physics core, it isn't rendering anything so wont touch the GPU

 

I didn't mean it that way i meant rendering in-game not improving the BANANAbench

 

No effect? Then why are my Banana scores different? The two images I uploaded above (left is 1GB ATI HD6780, right is the new 2GB GTX770) clearly depicts a change in the performance of my benchmark score. I never said it lowered my score. It increased the score but only by a little bit. Beforehand six cars was my max and now it's seven cars before the performance drops bellow 100%.

I've looked at many of the Banana scores. Plenty of people out there with the same identical 2600K processor at stock speed, exhibit varying score results, obviously affected due to different components within their computers.

 

Unsure about the GPU thing, but recently I did a test with HT on and HT off. Thought you guys might find it interesting. Oh, and Task Manager included for proofs.


(imported from here)

 

I have no idea why the scores are relatively different, but yes the second score on the right with the new 770 is lower. 48Mbeams/s vs 57Mbeams/s. Thats quite the difference, one which i doubt has anything to do with the GPU. Thermal throttling, power saving mode?

- - - Updated - - -

Yeah thats interesting, maybe Hyperthreading is off in your new benchmark. Definitely something to check.

 

New results!

4770k @ 4.4GHz

 

Um, are you using auto-overclocking? 4.4Ghz and 1.35v is pretty high for my tastes and it probably kicks higher (maybe 1.38-1.4v) during AVX loads.

I prefer nothing over 1.3v honestly. My chip (which is definitely not golden) even hits 4.3Ghz at 1.27v. I'd look into lowering that.

 

No, I've been doing it manually from the bios and I need 1.35v to get it stable, and anything for 4.5 is ridiculous pretty high jump from the meer 1.2v for 4.2Ghz, I haven't had the time to try 1.3v though I may try that tomorrow but untill then I've down clocked it back to 4.2 due to running Intel burn test it was hitting 100c which is waaaaaaaay to high for me.

 

Can someone explain to me why different GPUs like to run at higher voltages? I mean, my FX-8320 and my old A8-3850 are both 1.4v stock I think. (I run the 8320 at 1.5v) I've also had my old P4 rig up to 1.8v on air cooling

Also, my new laptop GPU (FX 1700M) runs hotter than my old one even though it runs at a lower voltage. (Old one is 160M)

 

Well, there is always the architecture and general silicon shenanigans, those can mess with voltage/clock things along with how high damaging voltage is. Main issue is the process node I think. As we get smaller chips it doesn't take as much voltage to cause electromigration or whatever. Imagine pushing 1l/sec of water through a hose. Now reduce the size of the hose by 2. Now it's higher pressure and it's easier to punch a hole in a weak spot. That's why vishera at 45nm (right?) can take 1.5-1.6v while haswell at 22nm gets iffy around 1.35-1.4.

(I think all this is right. I'm not really a guy that knows most of this.)

 

Installed Drive and ran Banana on my laptop. (3GHz Core 2 Duo.) I got about 17.5 max Mbeams. Definitely playable on Grid.

 

Yeah, I read about the pipe analogy somewhere. Care to link it?

 

more susceptible to interference the smaller you go too

 

Yeah, apparently radiation too. Smaller chips don't work in space. (I just moved closer to NASA in Alabama.)

I redid my test at 4.1Ghz because my PC was feeling slow. CPU is fine I guess. (It's really weird.)

 

Radiation was the implication actually and why curiosity has electronics manufactured on a 250nm scale along with a host of other measures. But yes, I have heard background radiation of a level considered average across the united states theoretically kicks in at about 12nm.

 

Someone should take the 5960x to 4.5Ghz and bench. I think we might break 100mbeams. Is there a way to (legally) send someone the BANANA? I know a guy over at anand who's got a 5960x.

 

You could get whoever it is to play beamng, as it is a fun game/physics engine

Sent from the 3rd galaxy from the talks of tapping