Monday, 9 January 2012

ATI 7970 - Gigabyte, Asus and XFX Comparison Review

Introducing the latest 7 Series cards from ATI. 

Up today we have something very special straight out of our offices. We have in our hands 7970 cards from XFX, Gigabyte and Asus. Straight off the manufacturing conveyer belt we get the opportunity to test out these cards for our Centre Com and worldwide fans!

These cards have managed to be kept quite secretive up to latest point in time as possible, our suppliers weren’t even able to provide any sort of tech specs for the cards but we have them right here for you:

7970 GPU
PCI-Express 3.0
3GB GDDR5 Memory
925MHz Core Clock
384-bit Memory Interface
2x Mini-DsplayPort
1x DVI

For testing we have gathered and built our system with the following parts and components:


The packaging of the items is in direct reflection of their prices. The XFX, cheapest of the three offers basic packaging for the card with just a single display port adapter within the box. Asus on the other hand have beefed it up featuring moulded foam and boxed paper which includes all the accessories and includes 2 display port adapters and a power extension cable, just in case yours doesn’t reach or you are intricate about your wiring.

The XFX Core Edition supports up to 3 simultaneous displays with independent resolutions and controls with support for stereoscopic 3D gaming. Support for 4k resolutions via HDMI and DisplayPorts. Up to 2560x1600 via DVI output. XFX have also added in their own Hydrocell Thermal Solution for improved cooling for those who like to overclock their cards.

The Gigabyte model supports up to 4 simultaneous displays via AMD Eyefinity and doesn’t feature any extra features of the card on the box.

The Asus card features its own GPU Tweak utility allowing you to monitor and optimize your graphic details.

All the cards in our hands today are your basic reference models straight out of the factory, but don’t worry; you can get yourself overclocked editions of these cards (not to say you can’t overclock them yourself)
Looking at the cards in detail, we really could not find any different between the brands. Apart from the obvious branding stickers and logos, the cards are almost identical from tip to end. The only real difference we could spot was different colour power sockets. The cards are quite lengthy and are about the same length as the 6990 so make sure you have good space to allow room for the cards. In our Cooler Master case, the cards fit pretty well and left enough room for decent airflow.
The cards come with a single fan located towards the end of the case so ensure good airflow is happening beneath the card and you don’t have other cards blocking its airway.

Let’s face it; we all want to know how these bad boys perform. First problem or issue we encountered was ATI’s way of presenting its driver software. The catalyst centre provides a very mixed view of high school graphics and tech specs for the I.T experts. The views between simple and advanced could not be any larger of a difference and the simple view almost provides no decent ability to play around with settings.

One of the frustrating things was VSync and ATI’s own way to describe it as ‘Wait for Vertical Refresh’. Basically it’s called VSync. Turning it off doesn’t seem to also work all the time. Even with our game settings VSync turned off, we were seeing limits still enforced by the driver. After Googling around for individual games we needed to change a few config files we were able to get it running. With that note, we used the same driver versions for all the cards (not that you can get any other at the time of testing as the card did not exist yet)

With the 28nm architecture within the card we expected it to draw a bit less power. Voltage consumption came in at 1.170v during full load and about .935v during idle. bit higher than expected but it definitely puts out that power consumption in terms of performance.

For testing purposes we ran benchmarks including a spec-heavy selection of games, Cinebench, OCCT, 3dMark11 and Heaven.

For our selection of games we had:
Crysis 2
Modern Warfare 3
Battlefield 3
Metro 2033

All our games were played at 1920x1080 with max settings on all available options. For consistency purposes we played the same levels on all cards for equal length of time. Our results consisted of minimum, maximum and the average FPS of the games. At the same time we were recording the temperature of the card and system in general.

Check out our results for the games.

The cards as expected performed well and equally, with minor differences not worthy of a clear cut winner, yet the GTX 580 seems to continue to hold its thrown on single-monitor gaming setups.

Our Cinebench tests proved some interesting results with the OpenGL results as follows:
FPS Result

XFX: 76.18fps
Gigabyte: 77.38fps
Asus: 79.16fps

The XFX and Gigabyte performed evenly with the Asus slightly edging out the two with a mere 2fps. Sounds like an extremely tiny difference which many will agree will have no difference in the real world.

Our Heaven benchmarking software was set at Full HD resolution with maxed setting and extreme tessellation. After all, these cards are marketed as having very good capabilities with high tessellation models. The cards performed as below (Scored in Points):

XFX: 807
Gigabyte: 806
Asus: 831

Somehow the Asus managed to cream the other two with its result. As cards that are supposed to be identical (which is shown with the XFX and Gigabyte) the Asus pulled out a rabbit from the hat. 

Lastly, we ran the popular 3dmark11 software to see how each of these would perform. As expected there were no surprises in results from card to card as each gave a good and consistent result. Our tests were done with extreme settings and at Full HD resolution.

XFX: X2346


Gigabyte: X2340


Asus: X2339

As you can see each card performed quite evenly in the Extreme category.

Temperatures on the cards were pretty much on par with each other with a few differences at times which could be blamed on fluctuating room temperatures. However with the 28nm chips within these cards, we were expecting a slightly better performance in temps. Logic dictates smaller chip equates to less heat production. Temperatures we noted ranged between 60 degrees on idle up to 80 degrees on full load, so it is quite the warm card so be sure to have good cooling within your case or you could potentially damage these cards. we hope to see non-reference cards greatly improve on these temperature results in the future.

To finish up, these cards provide very good graphical power, but the true power of these cards will show with EyeFinity displays and higher resolutions than Full HD screens. I can’t imagine anybody complaining when owning one of these cards, if not two of them as they offer the end user great current-gen gaming performance with good proofing for tomorrow’s games.
Each brand offers good cards in relation to their price points. In the end it will fall on the end users which model they would prefer to go with.

Centre Com now have these cards in stock!

The benchmark results may differ from user to user depending on what background software you are running and versions of benchmark software. These results aren’t portrayed to be seen as exact performance figures but merely as a rough estimate on the performance of the machine. These results are in no way bias to any company or person and are here to provide the end user in depth details and to provide extra assistance of potential purchases. All information on this page is subject to copyright. Please do not copy any parts of this article.

No comments:

Post a Comment

Note: only a member of this blog may post a comment.