If Anyone is Curious how the Radeon 7970 Performs in RO2

[mattlach]

I was discussing my plans to get a Radeon 7970 and overclock it in another thread, and I was asked to post my performance results when done.

I figured I might as well give it its own thread so its not buried in an unrelated one, just in case anyone else is interested.

http://g3ar.co.za/wp-content/uploads...HD-7970-OC.jpg

When I received my Gigabyte custom cooler card, it was defective. (blue snow and flickering lines from DVI output, displayport output is fine, overheats in stress testing at stock clocks, very high idle temperatures), so I have decided to return it.

That being said, I was able to do some minor overclocking and benchmarks despite its problems.

1150Mhz core was the highest I was able to reach stably (except in Metro 2033 where this setting overheated, probably due to the board problems). I was able to run the RAM stably at 1650Mhz. This was the highest setting in both CCC Powertune and in MSI Afterburner. I have seen others get higher, so there must be a way to override this.

All of my testing was on my core i7-3930K@4.6Ghz with 16GB of RAM.

I benchmarked 3 games I have, at a few different settings.

1.) One single 6970, Asus DCII overclock version (890mhz)

2.) 7970 OEM Clocks (925core/1375ram)

3.) 7970 Gigabyte OC (1000core/1375ram, the clock speed the board ships with)

4.) 7970 OC (1150core/1650ram, the highest overclock I was able to achieve with the bad board)

5.) two 6970's in Crossfire, Asus DCII overclock version (890Mhz)


Results are as follows:

Civilization 5:

I ran the built in benchmark function, and collected the late game full render score.

Settings were as follows:
2560x1600 Everything High, Vsync Off, 8xMSAA

http://farm8.staticflickr.com/7160/6...2ee4c747_o.jpg


Metro 2033:

I played the opening scene of the game (up until where the flying think knocks over the armored car and the scene ends), tracking my average framerate using FRAPS. Again, not the lack of results for the highest overclock, as it overheated due to the bad board.

Settings were as follows:
DX11, 2560x1600, Very High, AAA, 16x AF, Tesselation ON, DOF off.

http://farm8.staticflickr.com/7161/6...e5c8f021_o.jpg


Red Orchestra 2: (The one you have been waiting for)

I played the single player German campaign of the Railway Station map in at easy difficulty level (Single player, as it is more reproducible, Railway station as it looks like a more graphically intensive map, easy difficulty level so it doesn't drag on) I recorded average FPS using FRAPS.

Settings were as follows:
2560x1600, War Movie, Everything max, Bloom on, Motion Blur on, Light Shafts On, Ambient Occlusion On, Framerate smoothing off, Oe Frame Thread Lag on.

http://farm8.staticflickr.com/7032/6...2fe67ae3_o.jpg


So as you can see, at 1150/1650 the 7970 gets pretty close to two 6970's in crossfire. I'm hoping once I get a good board, and mod it with a better cooler, the results will be even better.

[Colt .45 killer]

nice card. Thats also a rather impressive leap in performance in one generation. Good to see the video card wars are ongoing.

[Westernesse]

Interesting. Thanks for posting the info.

[Sah.]

no, not curious sry

[mattlach]

Originally Posted by Sah.

no, not curious sry

Then why click the link an reply?

I figured I'd post my results as there aren't exactly a plethora of articles out there testing RO2 performance on different hardware.

[mattlach]

Originally Posted by Goten

Better test a CPU for this game.

Agreed, it's pretty CPU intensive. I just don't have a bunch of different CPU's to test on

[Andrew Blake]

Matt have you ever tried to overclock you CPU to 5.2? If I remember correctly it can get as powerful as a Xenon core. Or do you just use it for gaming?

[[Mad_Murdock]]

Will not be interested until I win the lottery, 'nuff said. Cool info though man.

[mattlach]

Originally Posted by Andrew Blake

Matt have you ever tried to overclock you CPU to 5.2? If I remember correctly it can get as powerful as a Xenon core. Or do you just use it for gaming?

4.6 is as high as it goes stably with my 180mm sealed water cooler (MainGear) which performs slightly better than a corsair h100.

With more extreme cooling I might be able to get there, but SB-E produces a lot more heat than SB, so I'd need some pretty extreme cooling.

I also dot know were you get the 5.2 number from. I have never seen anydata supporting that a XEON is notably faster than a SB or SB-E at the same clock (they are really just the same core with more cache, and cache can only do so much).

My understanding is that currently the highest clocked XEON is the E3-1290 at 3.6Ghz (4.0 turbo), and XEON motherboards dint typically have very good (any) overclocking features. My chip at 4.6 ought to already - as it is - blow any XEON on the market away

[Warlord]

Very interesting stats...thanks for posting.
Three fans? How loud is this card?

[Andrew Blake]

Originally Posted by mattlach

4.6 is as high as it goes stably with my 180mm sealed water cooler (MainGear) which performs slightly better than a corsair h100.

With more extreme cooling I might be able to get there, but SB-E produces a lot more heat than SB, so I'd need some pretty extreme cooling.

I also dot know were you get the 5.2 number from. I have never seen anydata supporting that a XEON is notably faster than a SB or SB-E at the same clock (they are really just the same core with more cache, and cache can only do so much).

My understanding is that currently the highest clocked XEON is the E3-1290 at 3.6Ghz (4.0 turbo), and XEON motherboards dint typically have very good (any) overclocking features. My chip at 4.6 ought to already - as it is - blow any XEON on the market away

Sorry I mixed up some a CPU with the i7-3930K. What I meant was two not overclocked Xeon cores against yours. They are pretty close with the Xeon X5650. You did good overclocking it
Which is amazing considering the price.


Here is a comperison by bit tech: http://www.bit-tech.net/hardware/cpu...dge-e-review/4