Layte's super-duper 480SLI Benchmark-O-rama!

[guadizba]

Here is mine. Like other posters with ATI setups, the gpu usage goes down significantly on scene 2. I'm running at 900/1200, basicaly 5970 toxic settings. CPU is at 4ghz as per screenshot. Not a bad result I suppose but scene 4 seems to suck.

have higher scores with similar gpu setups but my 980x was oc to 4.25ghz.

 

[Flanno]

Yeah, noticed your scores were higher allright. The other factor is I have a 759 mobo with the nforce 200 chipset which adds some latency to multi gpu setups.

 

[layte]

First post updated with some GTA:IV TfLC, DMC4 and FC2 results.

It would be nice to see something more comprehensive than RE5 numbers from the Tri-Fire crew, does that one scale nicely or something?

 

[Flanno]

I'll install some games later and run the benches. Currently in the middle of doing some cable sleeving.

Incidentally already had 2 5870's in crossfire, then went trifire and that was a noticeable boost in performance but was lacking with 3 screens due to the 1gig framebuffer limit. I moved straightt to 2 gtx 480's which solved that but the heat output and noise levels made me move to 2 5970 4gig cards and to be honest, the peformance to me feels about the same as the sli's 480's though I'm sure nvidia scales better overall.

 

[Flanno]

Just did the stalker bench in DX11 Ultra mode at 1920x1200. Not impressed to be honest.
I'm running at 4.15ghz now (bclk 160mhz). 4 GPU's are at Toxic speeds (900mhz). Like Resident Evil except in 1 test and at the very beginning did I see gpu utilisation over 80%. The rest of the time it was in the mid 40's to low 50's. I monitored it on my G19 with Afterburner. I noticed however that all gpu's including the 4th one were equally loaded. No microstutter visible.

Notice the suckage on the minimum framerates. Max are very good. But minimum brings the average right down. Comparing to the OP's score with GTX 480's in SLI, Quadfire is loosing out drastically in minimum framerates on the Day test, and has about half the minimum framerate compared to SLI on the Sunshafts test. Not a lot iin it on the othe 2tests. People have always said SLI gives better min. framerates.



Just re-ran it again. Seems there was a glitch allright on the 1st test with regards to minimum framerate. But the averages are still sucking. If it wasn't for the noise I would have kept my GTX 480 SLI setup which lasted all of 5 days.

 

[Razor Time]

Crysis Warhead Benchmark

 

[Razor Time]

Resident Evil DX10 16xAA

 

[rounddodger]

Finally Crysis is owned.

 

[layte]

Does anyone know of a Metro 2033 benchmark tool, or even a decent methodology for testing this game?

 

[Nemesis379]

Cool. Waits for pritty pictures.

+1

 

[Flanno]

I'm getting obsessed with this stalker benchmark. I finally got my cards stable at 950 (100mhz core overclock) with 1.25v which I know sounds high but stock voltage on the 4gig sapphires is almost 1.9. 980x at 4.16. Anyway, it seems to have made a difference to this bench. I've noticed the framerates are slowest just at the beginning of each scene, and scene 1 and 4 make heavy use of each gpu (over 85%), but scene 2 and 3 barely go over 50% on each gpu. But at least each gpu is being used so quadfire is working. Temps on gpu 2 get to almost 80 degrees though in scene 1 and 4. The rest stay around 70ish. Fans are at 100% on automode which is about the same as a 5870 around the 40-45% mark. So not too loud.

 

[layte]

So nothing more than the odd RE5 benchie from the previously chatty Tri-Fire crew.

I am disappoint.

 

[cainer]

So nothing more than the odd RE5 benchie from the previously chatty Tri-Fire crew.

I am disappoint.

if somone lends me a 5870 id run some benchies

 

[Phoenixdancer]

So nothing more than the odd RE5 benchie from the previously chatty Tri-Fire crew.

I am disappoint.

Sold my 5870 sorry lol

Quad fire help??

 

[layte]

Might as well, it will be interesting to see how well four GPU's scale. Which is why we are all reading this thread.

 

[Flanno]

So nothing more than the odd RE5 benchie from the previously chatty Tri-Fire crew.

I am disappoint.

You know people have better things to do Seriously, I only have bioshock 1 installed at present and am working my way through that. I will try to get time next week to do more benches.

I did Res Evil, I did Stalker. A couple of time I'll do the others when I get around to installnig the games.

Meanwhile here is Heaven at default setting at 2560x1600. This bench obviously favors Nvidia cards, especially with Tesselation and ATI cards seem to have pants min framerates but this is in 1 place only which causes the dip. It's the scene where you decend the strairs and you get shadows on the walls. It dips for just a second. I will also say there is very minor micro stuttering at the start but only at the start of the bench. I also find if you have afterburner running, sometimes it can cause major stuttering regardless of which driver you use. Only sometimes though.

Regarding scaling, here is my results. Bare in mind these are with old drivers, and stock speeds on gpu's and cpu. I didn't have my overclock running then. I think you will agree quadfire does indeed scale.

1 5970 4gig


Quadfire 4gig

 

[madindehead]

What program is that you are using to display that chip info on your G15 mate?

 

[Phoenixdancer]

Afterburner here

 

[Flanno]

Would you believe that's a G13 gamepad. And I'm using EVGA E-Leet.

 

[Duff-Man]

Okay, I got my GTX480 yesterday and I'm playing around with it. It seems stable at 825mhz with stock voltages, and it seems to roast if I up the voltage so I'm going to stick with that for now.

I've used the ASUS overclocking tool, but it's a bit clunky, and doesn't let you adjust memory timings, so I'm looking to use rivatuner. I can't see the usual driver button though... Hopefully I'm just doing something stupid? (see pic below)



What are you guys doing to use rivatuner for overclocking? Or are you using a different tool?