[Review] Fermi vs. Cypress

[NCspecV81]

Okay, here is another one of my gpu comparisons just for you guys. It's pretty much a full comprehensive review of all the gpu's I could think one would upgrade to, and I hope it helps in answering a lot of questions.

Now, it's not going to be one of those lame-o stock reviews that still has you wondering about total possible performance. I will be overclocking the living piss out of these cards on the stock coolers in the most relative games I can think. I do want you to understand a couple of variables that one may encounter with my tests:

- Overclock ability from one card to the next is not identical so your frequencies will vary based on silicon quality, ambient temperatures, and cooling capacity. Again, my overclocks do not represent the maximum possible overclock for every card.

- For games that I will need to utilize fraps due to the lack of in-game benchmark the result with be slightly different. Keep in mind it's always useful to apply a certain margin of error. I usually leave around 2 percent or so before claiming any significance in power differences. Your will belief will vary from mine.

Now that we have that out of the way I'll let you get a little more intimate with the competition.

ATI

ASUS 5970



MSI 5870 Lightening



HIS 5850 Reference



MSI 5770 Hawks (Crossfire)




nVidia

ASUS GTX480



PNY GTX470



ASUS GTX465




As you can tell some of them are reference and some of them are not so also keep that in mind when overclocks are concerned. But enough about that and let me go into detail concerning the actual reason for this thread.

The test system will include:



Intel i7 980x (4.61GHz HT Off)
Corsair Dominator GTX2's
EVGA 4-Way Classified X58
Enermax EVO 1250W
Separated Win7 Installs to factor out driver conflicts
Samsung 2493HM - its a 1920x1200 Capable but will run tests at 1080p

Games/Benchmarks:

3DMark03
3DMark05
3DMark06
3DMark Vantage
Battlefield Bad Company 2 (fraps)
Metro 2033 (Maxed w/o Physx and fraps)
Street Fighter 4
Resident Evil 5 (Fixed Benchmark)
FarCry 2
Crysis (unmodded)
Dirt 2
Hawx (10.1 enabled)
Just Cause 2 - Desert Sunrise & Concrete Jungle
Left4Dead 2
Counter-Strike:Source
Stone Giant
Heaven 2.0 (Normal Tessellation Only)

Those will be the tests that I will be utilizing that I have at my disposal. Now this will be suicidal overclocks on the stock gpu and my only requirement to assure overclock stability is that I must complete each benchmark at the same overclock. Therefore, if a clock can run on one bench and not the other I have to run the benchmark that passes at a higher frequency lower. This is to ensure consistency in the tests.

[NCspecV81]

Universal resolution on all benchmarks are 1920x1080 since I figured a majority of people used that resolution. The only benchmark that was run at 1920x1200 was l4d2 b/c I forgot to change that so I keep it static across all cards.

Here is the list of benchmarks and settings

3dmark03 - default
3dmark05 - default
3dmark06 - default
3dmark Vantage - default no physx
AVP Benchmark - dx 11 4xAA all max
Heaven 2.0 - Default settings @ 1920x1080 with normal tessellation
Crysis - very high dx10 4xAA
Farcry2 - dx10 ultra high / max 4xAA
Dirt 2 - dx11 max and 4xAA
Stone Giant - default settings at 1920x1080 high tessellation
Resident Evil 5 - Max settings 8xAA dx10 fixed benchmark
Hawx - dx10.1 all max in game settings 8xAA
StreetFighter4 - All max settings 8xAA
Counter-strike: Source - All max settings including bloom 8xAA and max_fps 999
Just Cause 2 - All max settings minus the nvidia only functions 8xAA
Left4Dead2 - 1920x1200 all max in game and 8xAA
Metro2033 - dx11 all max in game settings minus DoF and Advanced PhysX 4xAA
Battlefield Bad Company 2 - dx11 all max 8xAA

[NCspecV81]

Okay before I post my graphs I would like to touch base on the gpu clocks aspect of the benching. The reference 5850 I had did not touch the same clocks as the 5870 lightning therefore I decided to experiment a little. I took a 5850 bios and flashed it to my 5870 lightning and everything worked to plan. I got the same clocks as the 5870 but with the 5850 shader count. All of these were done on stock air.

5770CF - 1055/1350
5850 - 1115/1200
5870 - 1115/1200
5970 - 1000/1200
465 - 900/2100
470 - 875/2100
480 - 900/2300

Also, on the Stone Giant benchmark I was unable to complete results with crossfire. Those results for the 5970 and 5770's were noted with 0's. The multi-gpu aspect of that particular benchmark is too bugged to consider crossfire/sli results accurate.

[NCspecV81]

3dmark03



3dmark05



3dmark06



3dmark Vantage



AVP Benchmark



Heaven 2.0



Crysis



FarCry 2



Dirt 2



Stone Giant



Resident Evil 5



Hawx



Street Fighter IV



Counter-Strike: Source



Just Cause 2



Left4Dead 2



Metro 2033



Battlefield Bad Company 2

[illectrik]

thanks for the info

[OverShocked]

Nice review, thanks!

[hirsch]

Thanks for a great review.

I would like to se 5970 CF vs. GTX 480 2 and 3-way SLI, do anyone know about a review with that setup?

[ScottALot]

Am I the only one who thinks the 5870 and 5850 scores on the 3DMark look strange?

[NCspecV81]

which 3dmark? I can rerun if necessary.

[Eugenester]

3dmark05 the 5870 and 5850 have pretty low scores. butt anyway, nice review. You sure have a heap of nice GPUs

[sniper_sung]

Excellent review, except that I dislike the figure of 3DMark05 and 3DMark06. Usually I regard such kind of plots as marketing reviews if the range of certain axis doesn't start from 0.

[gillll]

thanks for the review.
this is how a modern review should be looks like -> all benches are atleast in FHD res.
and i guess metro 2033 in the new beast .

[Johnny87au]

SLI 480s rape the 5970s hands down.. kinda old..

[Open1Your1Eyes0]

NCspecV81, you're a member here also? Awesome.

Excellent review, btw.

[sniper_sung]

SLI 480s rape the 5970s hands down.. kinda old..

With either watercooling weight or tractor's noise

[The Duke]

I didnt know stone giant gave out FPS? I ran it once and it didnt give me anything lol.

[NCspecV81]

[M.Beier]

nvidia; the way its meant to be paid

[V0id]

Nice review/benchmarks, thanks for your hard work.

[darckhart]

great review. i really like your short setup explanation and then just bunch of charts. thanks!

[clo007]

Great review out together. I appreciate your effort in this!