What's new
  • Please do not post any links until you have 3 posts as they will automatically be rejected to prevent SPAM. Many words are also blocked due to being used in SPAM Messages. Thanks!

AMD Radeon HD 6970 and HD 6950 Review

Status
Not open for further replies.

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
Metro 2033 (DX11)

Metro 2033 (DX11)


There has been a lot of buzz about Metro 2033 which has mostly centered on its amazing graphics coupled with absolutely brutal framerates on even the best GPUs on the market. For this test we use a walkthrough and combat scene from The Bridge level which starts at the beginning of the level and lasts for about 3 minutes of walking, running and combat. Famerates are measured with FRAPS and Advanced PhysX is turned off.


1680 x 1050

HD6900-73.jpg


1920 x 1200

HD6900-74.jpg


2560 x 1600

HD6900-75.jpg
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
Unigine: Heaven v2.0 (DX11)

Unigine: Heaven v2.0 (DX11)


Unigine’s Heaven benchmark is currently the de-facto standard when it comes to simple, straightforward DX11 performance estimates. While it is considered a synthetic benchmark by many, it is important to remember that no less than four games based on this engine will be released within the next year or so. In this test we will be using a standard benchmark run with and without tessellation enabled at three resolutions. Note that tessellation IS NOT set to extreme.


1680 x 1050

HD6900-76.jpg


HD6900-77.jpg


1920 x 1200

HD6900-78.jpg


HD6900-79.jpg


2560 x 1600

HD6900-80.jpg


HD6900-81.jpg
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
Unigine: Heaven - Normalized Tessellation Performance

Unigine: Heaven - Normalized Tessellation Performance


Supposedly, AMD’s new architecture makes use of increased tessellation efficiency when compared to the outgoing HD 5800 series cards. In this short section we will be taking the results from the Heaven benchmark tests on the previous page and plotting them on a comparative graph.

HD6900-86.jpg

For whatever reason, the HD 6900 series did indeed have better all-round performance than the HD 5870 in Unigine but the performance drop with tessellation enabled is virtually identical between the two architectures. As we saw in past tests the HD 6970 and HD 6950 are able to eat high geometry scenes for breakfast so this could simply boil down to a lack of optimization in the 10.12 beta drivers.
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
8x MSAA Testing

8x MSAA Testing


In this section we take a number of games we have tested previously in this review and bring things to the next level by pushing the in-game MSAA up to 8x. All other methodologies remain the same.

BattleField: Bad Company 2 (DX11)
Note that 8x MSAA is enabled via the game’s config file for the NVIDIA cards since it is not a selectable option within the game menu

HD6900-52.jpg


Dirt 2 (DX11)

HD6900-59.jpg


F1 2010 (DX11)

HD6900-97.jpg


Just Cause 2 (DX10)

HD6900-66.jpg
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
Core Temperature & Acoustics / Power Consumption

Core Temperature & Acoustics


For all temperature testing, the cards were placed on an open test bench with a single 120mm 1200RPM fan placed ~8” away from the heatsink. The ambient temperature was kept at a constant 22°C (+/- 0.5°C). If the ambient temperatures rose above 23°C at any time throughout the test, all benchmarking was stopped. For this test we use the 3DMark Batch Size test at it highest triangle count with 4xAA and 16xAF enabled and looped it for one hour to determine the peak load temperature as measured by GPU-Z.

For Idle tests, we let the system idle at the Windows 7 desktop for 15 minutes and recorded the peak temperature.


HD6900-83.jpg

Considering AMD added a good portion of size to their GPU die in order to create the HD 6900-series, low temperatures on both cards were definitely a welcome surprise. However, it should be mentioned that while both cards would never be heard above the normal din of gaming, the HD 6970 was (subjectively) slightly louder than other GPUs we have tested recently.


System Power Consumption


For this test we hooked up our power supply to a UPM power meter that will log the power consumption of the whole system twice every second. In order to stress the GPU as much as possible we once again use the Batch Render test in 3DMark06 and let it run for 30 minutes to determine the peak power consumption while letting the card sit at a stable Windows desktop for 30 minutes to determine the peak idle power consumption. We have also included several other tests as well.

Please note that after extensive testing, we have found that simply plugging in a power meter to a wall outlet or UPS will NOT give you accurate power consumption numbers due to slight changes in the input voltage. Thus we use a Tripp-Lite 1800W line conditioner between the 120V outlet and the power meter.

HD6900-82.jpg

Power consumption for the HD 6970 is quite good when compared to the GTX 400-series products but it does loose out to NVIDIA’s newer GF110-based products. The HD 6950 meanwhile also pushes the upper limits of what we have come to expect from a second generation DX11 card. It should be mentioned though that AMD does give you the option to reduce these consumption numbers through their PowerTune technology though this will cause framerate drops as we will see on the next page.
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
The Effect of PowerTune on Performance

The Effect of PowerTune on Performance


AMD’s PowerTune is one of the first tools on the market which allows users to customize the TDP and power consumption based on their needs. However, many will be worried that a feature such as this one could artificially cap performance as well.

In order to put this debate to rest (or open up a can of worms depending on how you look at things), we installed our system with the HD 6970 into a closed Cooler Master 690 Advanced case and let it build up its internal temperature to approximately 35 degrees celcius. Our standard suite of benchmarks was then run and results compared to the open-system 23*C benchmarks. In addition, we also took the reduced power states out for a run around the block in order to see what kind of effect lowered PowerTune settings would have upon overall performance.

HD6900-89.jpg

The results are eye-opening to say the least since it doesn’t look like increasing PowerTune’s overhead had any positive effects upon performance. Most of the tests stayed well within an understandable margin of error. It seems like AMD’s technology is keeping things well in hand without affecting top-end performance in the games we test.

Further constraining the TDP did have a negative impact upon performance in some cases but not once did it reduce framerates to a non-playable level. Interestingly, there were only a few games that exhibited a noticeable framerate dip. This leads us to believe that our sample at least wasn’t operating at anywhere near its PowerTune-induced limits.

HD6900-90.jpg

In terms of power consumption, increasing the PowerTune limit didn’t noticeably decrease efficiency simply because our card didn’t actually need any of the overhead provided. By tightening the setting, we were able to knock a few watts off and as we saw above this really didn’t hamstring performance in any way. Using this method, users should be able to strike a perfect balance between performance and efficiency.

We do however have one request for AMD: a profile system that allows users to set certain PowerTune settings for particular games.
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
Overclocking Results

Overclocking Results


Unfortunately for us, we weren’t able to get our hands on a piece of software that would push the HD 6900 series’ cores over their CCC limits in time for this review. This means we were limited to the caps within AMD’s own software which was detrimental for the HD 6950 since AMD was very careful to ensure its overclocked frequencies couldn’t match those of the HD 6970. We will hopefully have the time to revisit this in the near future.

HD 6970 2GB

Core Speed: 950Mhz (CCC limited)
Memory Speed: 5744Mhz (QDR)


HD 6950 2GB

Core Speed: 840Mhz (CCC limited)
Memory Speed: 5300Mhz (QDR) (CCC limited)


HD6900-99.jpg
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,841
Location
Montreal
Conclusion

Conclusion


Before the release of a new generation of graphics cards, there is always a fair amount of hype and assumption flying around. This sometimes results in a pleasant surprise come release day and in other situations a crushing reality check for those who expected far too much to begin with. There have been all too many products which fall victim to over inflated expectations and this situation will likely repeat itself with the HD 6970 and to a lesser extent the HD 6950. That really is a shame because these are highly competitive products at excellent price points.

HD6900-87.jpg

In most ways, AMD hit the nail on the head with the HD 6970. Its focus upon improved geometry horsepower and a massive framebuffer allows for some impressive DX11 performance gains over the HD 5870, particularly at ultra high image quality settings.

Things become a bit less cut and dry when the higher-end Cayman board is compared directly against the GTX 570. At lower resolutions NVIDIA’s card is able to pull away and at 1920 with AA enabled it really becomes a toss-up between the two cards. For the vast majority of gamers who don’t have 27 or 30 inch monitors, numbers like this mean the $369 AMD card is a bit too expensive for the performance numbers it displays. However, if you game on an ultra high resolution setup this will likely be the card you’ll be looking for since at times it even one-ups a GTX 580.

HD6900-88.jpg

While the HD 6970 does compete very well in the upper end of the price spectrum, the real star of this show is the HD 6950. Currently, there isn’t anything that even comes close to matching its capabilities in the $299 market. Performance is well above that of the $270 GTX 470 and even AMD’s previous single GPU flagship –the HD 5870- struggles to keep pace; particularly when AA is enabled. To our way of thinking, this is the closest thing to perfection we have seen for a while.

So performance for the Cayman XT and Pro is generally very good and the feature set for these products is as well rounded as they come. PowerTune is a great concept that is highly user friendly when used in conjunction with the Overdrive settings and in our testing, it never did step in to limit in-game performance. The dual BIOS toggle is interesting for the overclockers among us but it does have dubious value for anyone who will stick to software overclocking at default voltages. Nonetheless, it adds a serious safety net for users who fancy a walk down the BIOS hacking block.

The efficiency of higher end GPUs is always a massive concern for manufacturers. AMD seems to have now run into NVIDIA’s Fermi situation: adding the items necessary for higher DX11 performance comes with a cost of decreased efficiency and a higher TDP. PowerTune does help mitigate this hurdle but in no way does it eliminate Cayman’s seemingly high power consumption numbers. Against the GTX 480 and GTX 470, the XT and Pro fare quite well but they do tend to fall flat in the overall efficiency category against the newer GF110-based cards. That being said, at ultra high resolutions AMD’s new cards are able to edge ahead in performance per watt.

Both the HD 6970 and HD 6950 are incredibly enticing but the HD 6970 in particular isn’t without its faults. Unfortunately, AMD aimed these cards at NVIDIA products that have either been discontinued (the GTX 480) or will be replaced in very short order (the GTX 470). This leads to the HD 6970 fighting an uphill battle at the most popular resolutions against a lower priced yet similarly performing GTX 570. The HD 6950 on the other hand avoids this quagmire altogether by hitting a market where there is very little to no competition which makes it infinitely more appealing than its bigger brother.

With the HD 6900 series, AMD ensured that improvements were made in key areas where Cypress was left wanting. The result is a pair of GPUs that perform well but ultimately fail to deliver a killing blow to NVIDIA’s refreshed Fermi lineup. Whether or not these products are a sales success is a real question mark here but regardless of expectations, AMD has still released cards that are able to give the competition a run for their money.


 
Status
Not open for further replies.

Latest posts

Twitter

Top